Grammaticalization
Uploaded by
Silvio DominguesGrammaticalization
Uploaded by
Silvio DominguesOUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
Grammaticalization
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
O X F OR D T E X T B O O K S IN L I N G UI S T I C S
PUBLISHED
The Grammar of Words Compositional Semantics
An Introduction to Linguistic Morphology An Introduction to the
Syntax/Semantics Interface
by Geert Booij by Pauline Jacobson
A Practical Introduction to Phonetics The History of Languages
An Introduction
by J. C. Catford by Tore Janson
An Introduction to Multilingualism The Lexicon
Language in a Changing World An Introduction
by Florian Coulmas by Elisabetta Ježek
Meaning in Use A Functional Discourse Grammar
An Introduction to Semantics for English
and Pragmatics by Evelien Keizer
Grammaticalization
by Alan Cruse
by Heiko Narrog and Bernd Heine
Natural Language Syntax Diachronic Syntax
by Peter W. Culicover by Ian Roberts
Principles and Parameters Speech Acts and Clause Types
An Introduction to Syntactic Theory English in a Cross-Linguistic Context
by Peter W. Culicover by Peter Siemund
A Semantic Approach to English Grammar Linguistic Typology
by R. M. W. Dixon by Jae Jung Song
Semantic Analysis Cognitive Grammar
A Practical Introduction An Introduction
by Cliff Goddard by John R. Taylor
Pragmatics Linguistic Categorization
by Yan Huang by John R. Taylor
IN PREPARATION
Codeswitching Cognitive Grammar
by Jeff MacSwan An Introduction
Diachronic Syntax
by John R. Taylor
by Ian Roberts
An Introduction to Phonetics
and Phonology
by Nathan Sanders
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
Grammaticalization
HEIKO NARROG AND BERND HEINE
1
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
3
Great Clarendon Street, Oxford, ,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Heiko Narrog and Bernd Heine
The moral rights of the authors have been asserted
First Edition published in
Impression:
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
Madison Avenue, New York, NY , United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number:
ISBN –––– (pbk.)
ISBN –––– (hbk.)
Printed and bound by
CPI Group (UK) Ltd, Croydon,
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
Contents
Preface xi
List of figures and tables xiii
Abbreviations xv
1. Introduction
1.1 What is grammaticalization?
1.2 Primary vs. secondary grammaticalization
1.3 Unidirectionality
1.4 Direction of semantic change in grammaticalization
and expansion of scope
1.5 Grammaticalization as explanation
1.6 How to study grammaticalization
1.6.1 Diachronic grammaticalization studies
1.6.2 Grammaticalization as a synchronic concept
1.6.3 Reconstruction
1.7 Overview of the remainder of the book
Discussion points
2. Criteria, parameters, and other variables
2.1 An overview
2.2 Discussion
2.3 Conclusions
Discussion points
3. Four parameters of grammaticalization
3.1 Context extension
3.1.1 The context extension model
3.1.2 Discussion
3.1.3 On the differential behavior of stages
3.2 Desemanticization
v
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CONTENTS
3.3 Decategorialization
3.4 Erosion
3.5 Conclusions
Discussion points
4. Directionality of semantic change in grammaticalization
4.1 Bleaching, generalization, abstraction
4.2 Discourse orientation
4.2.1 Increase in speaker orientation
4.2.2 Increase in hearer orientation
4.2.3 Increase in textual orientation
4.2.4 Discourse orientation in grammaticalization
4.2.5 Counterexamples to (inter)subjectification and
their status in discourse orientation
4.2.6 Conclusion
Discussion points
5. Steps and phases of grammaticalization
5.1 Gradualness
5.2 Stages
5.3 Grammaticalization chains
5.4 Grammaticalization in semantic maps
5.5 Polysemy, transcategoriality, and heterosemy
5.6 Cycles
5.6.1 Renewal and innovation
5.6.2 The morphological cycle
Discussion points
6. What drives grammaticalization? Mechanisms
and motivations
6.1 Mechanisms: reanalysis and analogy
6.1.1 Reanalysis
6.1.2 Analogy/extension
6.1.3 Summary
6.2 The language system
vi
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CONTENTS
6.3 Communicative needs or goals of the speaker
6.4 Inferences and contexts
6.5 Cognition and conceptualization
6.6 Frequency of use and language processing
6.7 Discourse
6.7.1 Grammaticalization from discourse structures
6.7.2 Discourse as the arbiter
6.8 Conclusion
Discussion points
7. Grammaticalization in language contact
7.1 Grammaticalization vs. polysemy copying
7.2 A case study: Basque
7.3 Identifying contact-induced grammaticalization
7.3.1 Evidence for contact-induced grammaticalization
7.3.2 Diagnostics for determining the direction of transfer
7.4 Pidgins and creoles
7.5 Conclusions
Discussion points
8. Grammaticalization and language typology
8.1 Typological features guiding grammaticalization
8.2 Grammaticalization as a possible explanation for
typological features of languages
8.2.1 Grammaticalization and word order
8.2.2 Grammaticalization, the suffixing preference,
and morpheme order
8.2.3 Grammaticalization and expression types of
grammatical categories
8.2.4 Grammaticalization and the cycle of synthetic
vs. analytic expression of categories
8.3 Conclusion
Discussion points
vii
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CONTENTS
9. Theory-specific approaches
9.1 Generative grammar
9.2 Functional Discourse Grammar
9.3 Variationist Theory (sociolinguistics)
9.4 Cognitive Grammar
9.5 Construction Grammar
Discussion points
10. Beyond grammaticalization
10.1 Lexicalization
10.2 Degrammaticalization (anti-grammaticalization)
10.3 Exaptation/regrammaticalization
10.4 Other ‐izations
10.5 Conclusion
Discussion points
11. Discourse markers
11.1 What are discourse markers?
11.2 Problematic features of discourse markers
11.2.1 Do discourse markers belong to grammar?
11.2.2 From syntactic constituent of the sentence to
syntactically unattached status
11.2.3 From prosodically integrated to non-integrated
or less integrated status
11.2.4 From meaning as part of a sentence to meaning
outside the sentence
11.2.5 From sentence function to metatextual function
11.2.6 From positionally constrained to largely
unconstrained placement
11.2.7 Conclusions
viii
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CONTENTS
11.3 Hypotheses on the development of discourse markers
11.3.1 The grammaticalization hypothesis
11.3.2 The pragmaticalization hypothesis
11.3.3 The lexicalization hypothesis
11.3.4 The cooptation hypothesis
11.4 Conclusions
Discussion points
12. Summary and conclusion
Glossary
References
Author index
Language index
Subject index
ix
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
Preface
The study of grammaticalization has been a particularly vibrant field of
linguistics in the past couple of decades. The words “grammaticaliza-
tion” and “grammaticalized” can be found in literature ranging from
historical linguistics to language description, as well as in theoretical
studies of all color. Nevertheless, to our knowledge only one introduc-
tion to the topic written in English has been available so far, and we felt
that now is a good time to offer a new introduction and textbook that is
not only updated but also offers an alternative view of the topic.
Many people have contributed to this book in one manner or the
other. In a sense, this book is supported by the whole field of study and
the scholars working in it, and the copious publications which have
informed this book and its authors. We wish to particularly thank
Alexandra Aikhenvald, Walter Bisang, Laurel Brinton, Hilary Chappell,
Kristin Davidse, Olga Fischer, T. Givón, Martin Haspelmath, Kees
Hengeveld, Yūho Kitazaki, Tania Kuteva, Haiping Long, Andrej Mal-
chukov, Toshio Ohori, Seongha Rhee, and Elizabeth Traugott for their
input over the years.
Proceeding with the book project, we received great feedback and
support from Oxford University Press. We thank our editors Julia Steer
and Vicki Sunter for all their understanding and support, and Abirami
Sinduja for overseeing the production. It goes without saying that all
remaining errors are our own.
Heiko Narrog also wishes to thank for the support received from the
Japanese Society for the Promotion of Science through grant number
H.
Finally, we hope that our book can serve as a useful point of
orientation both for students who are just learning about the field
and scholars who are already working in it. The concept of grammat-
icalization offers practically infinite possibilities for application, and we
hope to see many more of them in the future.
Heiko Narrog and Bernd Heine
Sendai and Köln, August
xi
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
List of figures and tables
Figures
. Navajo verb template (according to Young and Morgan
: –)
. A sequence of semantic changes in grammaticalization
. Grammaticalization chain of ‘give’ in Kabba
(Moser : )
. Semantic map of ‘breathe’ (François : )
. Semantic map of the instrument/companion domain
(Narrog and Ito )
. Semantic map of the instrument/companion domain
with directionality of meaning extension
(Narrog a: )
. Two mechanisms leading to renewal
. Inflection outside of derivation (Mithun : )
. Paths for the development of future morphemes
(Bybee : )
. Grammaticalization of be going to in Cognitive
Grammar (Langacker : )
. Lexicalization, degrammaticalization, and exaptation
vis-à-vis grammaticalization
Tables
. Rise in normalized frequency of the English progressive
by register, – (excerpt from Kranich a: )
. Rise in normalized frequency of the English
progressive by use, – (excerpt from
Kranich a: )
. ‘Threaten’-constructions in some European languages
(Heine and Miyashita : )
xiii
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
LIST OF FIGURES AND TABLES
. The parameters of grammaticalization (Lehmann
[]: )
. The context extension model (Heine )
. Features of speaker orientation
. Verbal correspondences between two Romance
model languages and varieties of Basque (based
on Jendraschek : )
. Portuguese expressions grammaticalized by Tariana
speakers on the model of evidential categories in
Tariana (Aikhenvald : –, –; Heine
and Kuteva , Table .)
. Prefixing versus suffixing in inflectional morphology
(Dryer : )
. Features of lexicalization vs. grammaticalization
(Brinton and Traugott : )
xiv
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
Abbreviations
Grammatical categories
future
ablative
genitive
absolutive
gerund
accusative
honorific
adnominal
indefinite
adverb, adverbial
identity marker
applicative
illocutionary modifier
article
imperative
aspect
inessive
causal
infinitive
, noun class , . . .
. . . instrument
classifier indirect object
complementizer imperfective
concessive locative
comitative narrative
copula negation
dative neuter
demonstrative non-future
determiner nominalizer
directional nominative
enunciative non-past tense
epistemic object
modality partitive
ergative participle
essive perfective
evidential perfect
focus progressive
xv
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
ABBREVIATIONS
present first person (combined
proximative with number)
possessive second person (com-
past bined with number)
purpose third person (combined
with number)
interrogative
-digit portmanteau combinations
quotative
(examples)
relative (clause marker)
first person singular
reflexive
second person plural
relator
first person feminine
subject singular
subjunctive third person masculine
specific singular
subordinator feminine singular
switch-reference
marker of tag questions Parts of speech and
term of address morphological categories
topic
adjective
transitive
adverb
transnumeral
article
Person-gender-number discourse marker
portmanteau labels modal particle
noun
Building elements noun phrase
()() dual particle
()() feminine verb
()() masculine verbal adjunct
()() neuter verb phrase
()() non-feminine
() non-masculine Word order
() non-neuter
()() plural subject-object-verb order
()() singular subject-verb-object order
xvi
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
ABBREVIATIONS
verb-object order
verb-subject-object order
Symbols
= clitic boundary
? morpheme of unknown
meaning
* reconstructed item/
ungrammatical
() tentative interlinear glosses
by the present authors
xvii
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
1
Introduction
1.1 What is grammaticalization?
1.2 Primary vs. secondary grammaticalization
1.3 Unidirectionality
1.4 Direction of semantic change in grammaticalization
and expansion of scope
1.5 Grammaticalization as explanation
1.6 How to study grammaticalization
1.7 Overview of the remainder of the book
Discussion points
1.1 What is grammaticalization?
We define ‘grammaticalization’ as the topic of this book as follows:
() Grammaticalization is the development from lexical to grammat-
ical forms and once the grammatical form has evolved, the devel-
opment of further grammatical forms. Since the development of
grammatical forms takes place in specific contexts and construc-
tions, the study of grammaticalization is also concerned with
constructions and context, including even larger discourse
segments.
This definition refers to change in both contents and form, as linguistic
expressions have a semantic and a formal component, including the
constructions of which they are a part. Thus, it differs from some other
definitions that are restricted to a change in meaning (or function),
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
such as that by Hopper and Traugott () in (), or to discourse
prominence, such as that by Boye and Harder () in ().
() [Grammaticalization is] a term referring to the change whereby
lexical items and constructions come in certain linguistic contexts
to serve grammatical functions and, once grammaticalized, con-
tinue to develop new grammatical functions. (Hopper and Trau-
gott : )¹
() Grammaticalization is the diachronic change that gives rise to
linguistic expressions that are by convention ancillary and as such
discursively secondary. (Boye and Harder : )
The definition presupposes that we know what is ‘grammatical’ and
what is ‘lexical.’ But this is not always the case. First of all, the term
‘grammatical’ is associated with a number of uses. On the one hand, it
may refer to the fact that a linguistic expression conforms to the rules of
grammar. On the other hand, it refers to linguistic forms and structures
that complement lexical items (contents words) as the building blocks
of constructions and sentences. In this use, a term like ‘grammatical
category,’ also called ‘functional category,’ stands in direct contrast with
‘lexical category.’ In this book, the latter sense is meant. Whenever
there is a possible misunderstanding between the two conventions we
will use the term ‘functional (category)’ instead of ‘grammatical (cate-
gory)’ to contrast with ‘lexical (category).’
Secondly, the answer to the question what is grammatical depends
on a variety of factors, especially on the specific framework or theory of
grammar within which it is asked. There are certain prototypical
grammatical categories like ‘past tense’ or ‘indefinite article’ that will
be treated as ‘grammatical’ in any grammar, and words like procrasti-
nate or committee that will be treated as a lexical entry in any lexicon.
And there are categories such as complex prepositions, for example on
behalf of, in the light of, so-called semi-modals, such as have to and
seem, or so-called light verbs like do, that occupy a transitory zone
between the lexicon and grammar. Here, classifications may differ from
grammar to grammar, and from description to description. In general,
grammatical forms constitute closed classes of items with shared
¹ Note that () covers essentially only one of the four main parameters that will be used
in this book to describe grammaticalization, namely context extension (Chapter .).
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. WHAT IS GRAMMATICALIZATION?
morphological and syntactic properties, in contrast to nouns and
verbs, which are open-ended classes. That is, the latter have a poten-
tially open-ended number of members. Grammatical meanings are
typically abstract, procedural, and non-autonomous, that is, they gain
their significance only in conjunction with words that carry specific
content, and often serve to organize speech or relate words to each
other. For example, the concept of past tense is an abstraction relying
on our processing of time instead of on concrete visual or auditory
or tactile perception. A past tense as such is meaningless without
a predicate that can designate an event or state that took place in
the past.
One of the most attractive proposals with respect to the distinction
between the grammatical and the lexical comes from Boye and Harder
() in their framework of discourse prominence. Using a number of
analytic tests, such as on addressability and focalization, they classify
linguistic items into discursively primary and discursively secondary
ones. In the view of these authors, demonstratives such as English this
and that qualify as lexical, that is, non-grammatical expressions: Like
lexical items, they can be both addressed and focalized (e.g. Look at
that.—What?); hence, they are classified as discursively primary rather
than secondary (Boye and Harder : ).
In whichever way the distiniction is made, lexical forms become
grammatical and from there on may develop even further grammatical
functions. The development from lexical to grammatical is often called
‘primary grammaticalization,’ and the development of further func-
tions from there on ‘secondary grammaticalization’ (Givón ;
Breban and Kranich ; see Section .). These further grammatical
functions have sometimes been labeled as ‘more grammatical,’ for
example, in the frequently quoted definition of grammaticalization by
the Polish scholar Kuryɬowicz:
() Grammaticalization consists in the increase of the range of a
morpheme advancing from a lexical to a grammatical or from a
less grammatical to a more grammatical status, e.g. from a deriv-
ative formant to an inflectional one. (Kuryɬowicz : )
The problem is then how ‘more grammatical’ can be defined. There
are basically only two options. The first is with a circular definition:
“what comes later in grammaticalization is more grammatical”. This is
clearly not desirable. But one could go one step further and create a
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
grammatical model based on grammaticalization, in which what
is more grammatical is defined by stage of grammaticalization. The
second option is with reference to a specific model of grammar that has
an independent criterion for degree of grammaticality. Indeed, there
are such models within which it is possible to clearly define what is
‘more’ grammatical (cf. Chapter ). The common denominator in these
models is that further grammaticalized categories have wider scope,
that is, the linguistic units whose interpretation they influence (e.g. verb
vs. verb phrase vs. clause) belong to an increasingly higher level of
grammar. In this book we also assume that continuous grammaticali-
zation is characterized by widening (or at least not narrowing) of scope.
We will return to the issue of ‘scope’ in Section ..
The evolution of the verb go (to) to future marking in specific
constructions is a paradigm example of continuing grammaticalization.
It has been cited many times, but we will still repeat it here, because it is
cross-linguistically so common. Instead of English go/going to, we take
a look at the corresponding expression in Mexican Spanish. This also
serves to illustrate that in grammaticalization the same kind of devel-
opment is often found in many languages, even across language fa-
milies and linguistic areas.
According to Dávalos (), the construction ir a + infinitive ‘go to’
originally had a motion meaning, as in example ().
() Mexican Spanish (CORDE, fiction, ) (Dávalos : )
E es asy como el que va a pescar peces
and is so like the who go. to fish. fishes
al monte o el que va a buscar
to-the mountain or the who go. to seek.
liebres al mar.
hares to-the sea
‘And it is like the one who goes to the mountain to fish or the one
who goes to the sea to catch hares.’
In the fifteenth century it then developed an aspectual reading as
prospective aspect, that is, prediction of some immediate occurrence
based on a present state. Examples open to an aspectual rather than
motion reading were first available with human subjects, based on the
overall interpretation of the situation in its non-linguistic context. With
non-human subjects that do not move based on their own volition, as
in example (), the aspectual reading became even more preferred.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. WHAT IS GRAMMATICALIZATION?
Note that in example (), there is a remnant of the meaning ‘go’ because
motion is involved, and the main verb is an activity that would be, in
principle, plausible as the end point of a motion event. However,
because of the nature of the subject, it is clear that no intentional
motion can be involved.
() Mexican Spanish (CORDE, history, ) (Dávalos : )
[Description of an indigenous show with a stick which is being
kicked like a football; the writer is amazed by the elegance of the
player]
Y muchas veces parece que le va a
and many times seem. that him go. to
dar en la cabeza, y cuando menos catamos
give. on the head and when least look.
acude con el un pie y lo recoge . . .
come. with him a foot and it get.
‘And many times it seems that the stick is going to hit him on his
head, and when we least expect it he comes with his foot and gets
it . . . ’
This aspectual reading became dominant in the nineteenth century.
Furthermore, at a third stage, in the late nineteenth and the twentieth
century, an ‘immediate future’ use emerged that was characterized by
the coocurrence with temporal adverbs such as en lo sucesivo ‘from now
on’ in (). Also note the ‘be’-verb in the complement, which is incom-
patible with a ‘go’ reading.
() Mexican Spanish (CREA, fiction, ‒) (Dávalos : )
[ . . . ] porque uste, en lo sucesivo, va
because you in the following go..formal
a ser mi secretario.
to be. my secretary
‘[Yes, sir—he answered—go have a break. And you know: from
this evening onward you stay with me. Right now I have the office
prepared that belonged to Luisito,] because from now on you are
going to be my secretary.’
Lastly, in contemporary language, a ‘remote future’ reading, as in (),
with adverbial expressions that support the temporal remoteness, such
as para el próximo siglo ‘for the next century,’ has emerged.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
() Mexican Spanish (CREA, oral, n.d.; Dávalos : )
[ . . . ] van a ser para el próximo siglo aquellos
go. to be. for the next century those
que va a condenar esta historia.
whom go. to condemn. this history
‘[However, history, the new history is writing itself, today you
are the winners,] but in the next century you are going to be the
ones that are going to be condemned by this history.’
Thus, the verb ir ‘go’ in a construction with the preposition a and a
following infinitive came to be grammaticalized as a marker of aspect
and then of tense. The step from motion verb to aspect marker can be
labeled as ‘primary grammaticalization,’ and the further step to tense as
‘secondary grammaticalization.’ As mentioned above, this is a devel-
opment that, apart from details that vary from language to language,
has taken place in many languages, and in some languages it has even
gone further. English be going to, for example, has additionally acquired
an epistemic modal use, as illustrated in ().
() I think that there’s going to be incompetence in every profession
(ICE-GB) (Collins : )
Be going to in () arguably does not express a prediction about a future
event but the speaker’s inference about a present general state of affairs.
That is, it is not temporal, but ‘epistemic.’ The next step, from future to
epistemic modality is in fact quite common. The English modal will has
taken it as well, as have the French synthetic future form, for example,
and the Spanish synthetic future.
Beyond the Indo-European languages, Bybee et al. () found
future markers derived from motion verbs in seventeen unrelated
languages in a sample of ninety-four languages. They range from Africa
to South America, China, and Papua New Guinea. This strongly sug-
gests that grammaticalizations are based on cognitive mechanisms that
do not depend on specific language structures. Recurrent developments
such as the one of a movement verb to future have prompted scholars
to suggest that it is diachronic mechanisms leading to grammatical
structures that are universal, rather than the synchronic structures
themselves (Bybee : ).
On the other hand, many languages cited among those seventeen
unrelated languages from Africa, South America, and so on, do not
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. WHAT IS GRAMMATICALIZATION?
have a long written history. In fact, in the vast majority of the languages
of the world there is no tradition of writing. And for more than
percent of all languages there are no appropriate written records for
reconstructing their history. This raises the question whether grammat-
icalization is a valid approach for only a very small number of languages.
To address this important question, reconstruction is a central issue
of this book. The approach sketched here rests on the extrapolation
from known to unknown data. It is contingent upon the knowledge
of findings that were made on the basis of evidence from written
languages.
The underlying idea is that when ‘hard’ proof from written docu-
ments is not available, we can still take the path of accumulating
evidence until a plausible hypothesis which is superior to alternative
hypotheses that exist can be formulated. Givón () succinctly for-
mulated the usefulness of reconstruction of earlier language states for
linguistics as follows:
Quite legitimately, then, one may wish to study diachrony not for its own sake, but
rather to gain better understanding of seemingly-synchronic universals that govern the
range of possible grammatical typological diversity. (Givón : )
The main goal of the study of grammaticalization, including reconstruc-
tion, in our view, is to contribute to explanations of why grammar is
structured the way it is. We will further explore this idea in Section ..
Further reading
For a substantial list of definitions that have been proposed for grammati-
calization, see Campbell and Janda (). The history of grammaticalization
has been dealt with extensively in previous monographs, especially in Leh-
mann ( []: –), and Heine et al. (: –). General overviews
and introductions into the study of grammaticalization are provided by Heine
et al. (), Hopper and Traugott (), and Lehmann ( []).
Wiemer () discusses the concept of discursively primary and secondary
in detail. Budts and Petré () is an excellent corpus study on (part of) the
development of be going to as a future marker, which was given as an example
here. Relatedly, besides the more established qualitative notions of what is
more grammatical, more recently there have been attempts to develop quan-
titative techniques for measuring relative degrees of grammaticalization both
in diachrony and synchrony (Saavedra ; Bisang et al. ).
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
1.2 Primary vs. secondary grammaticalization
Definitions of grammaticalization usually refer not only to the
development of elements of grammar out of lexical words, but also
to the development of further grammatical functions in already
grammaticalized items. While the latter is covered by our definition
in () of Section ., it is not ‘grammaticalization’ in a strict sense
of a development of something grammatical out of something
lexical. Therefore, it has often been distinguished from so-called
‘primary’ grammaticalization, or grammaticalization proper, being
called ‘secondary’ grammaticalization. The term was coined by
Givón (: ), who defined secondary grammaticalization as a
process where “earlier-grammaticalized morpho-syntax can give rise . . .
to other morpho-syntactic patterns.” When introducing the term,
Givón () was referring to the development of relative clauses,
adverbial clauses, and complement clauses out of nominal clauses in
Hebrew.
Givón’s () definition has largely remained in place. However,
some scholars argue that secondary grammaticalization is likely to have
other properties than primary grammaticalization (cf. Breban ;
Breban and Kranich ; Kranich ; Narrog ). Most promi-
nently, the type of meaning change found in secondary grammatical-
ization may be different from the one involved in primary
grammaticalization: while primary grammaticalization usually involves
subjectification, secondary grammaticalization usually does not
(cf. Section .). There have also been proposals to further distinguish
different types of secondary grammaticalization, for example, second-
ary grammaticalization involving structural change vs. mere semantic/
functional extension (Breban ). Lastly, it has also been suggested
that secondary grammaticalization should not be included in the study
of grammaticalization (e.g. von Mengden ). In this book, we take
the more common stance of a simple distinction between ‘primary’ and
‘secondary’ grammaticalization. In doing so, we try to elucidate in
which respect ‘secondary’ grammaticalization may indeed be different
from ‘primary’ grammaticalization when it comes to issues relevant to
the distinction.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. UNIDIRECTIONALITY
1.3 Unidirectionality
If grammaticalization is “the development from lexical to grammatical
forms and once the grammatical form has evolved, the development of
further grammatical functions,” (see () above) ‘unidirectionality’ of
grammaticalization means that the reverse process is not found, or else
is not covered by the term grammaticalization. That is, expressions of
grammatical functions do not develop back into more basic grammatical
functions, and eventually become lexical items again. This tenet of
grammaticalization theory is also known under the label of ‘irreversibil-
ity.’ It is in the first place an empirical observation, and has been a more
or less explicit part of the modern concept of grammaticalization from its
beginnings in the s, so that it is difficult to identify a first proponent.²
Recall some of the examples of grammaticalization from the previous
sections. A verb with the meaning ‘go’ in a construction with a prepo-
sition of direction and infinitival complement becomes a marker of
aspect and then future tense (Section .). In a rigorously chosen sample
of ninety-four languages of the world (Bybee et al. ), the type of
change from motion verb to future marker was found in nineteen
languages. How likely is it that in even one of those languages, a future
marker became an aspect marker and then further became a lexical verb
with the meaning ‘go’? We do not know of any such case and clearly it is
not very likely. Or, think of an indefinite article that has become a marker
of non-specificity which reverts and becomes a presentative marker and
eventually the numeral ‘one’ (Section .). How likely would that be?
Beyond being a generalization over empirical observances, the prin-
ciple of unidirectionality also has some theoretical significance. If
grammaticalization were easily reversible, it would be a rather random
change among many, and it would not be very useful for the purposes
of generalization and reconstruction.
On the other hand, the existence of this principle and its importance
has also challenged scholars to find counterexamples. Counterexamples
² Possibly, Vincent (: ) was the first: “Chains of grarnmaticalization are unidi-
rectional or unilateral—i.e., put at its most general, lexical items may be grammaticalized,
but grammatical items do not become lexicalized.”
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
have been presented under labels such as ‘degrammaticalization,’ ‘de-
grammation,’ ‘antigrammaticalization,’ and the like, but the question is,
in which sense, and to what extent they exist. A more detailed discus-
sion of this matter will be found in Section .. To give just one
example here, think of the formation of the noun ism as an abstraction
over a lot of nouns that end in the suffix -ism (e.g. bilateralism,
socialism, fascism, etc.). This is indeed the development of a lexical
entry from a suffix, but not as a reversal of the process of grammati-
calization, but as an intentional and spontaneous formation rather than
a gradual process (see Börjars and Vincent : ; Haspelmath
). Furthermore, typically for many proposed cases of ‘degramma-
ticalization,’ it is primarily the morphological direction of change that
is reversed, and not the semantic/functional one. It seems that the
semantic/functional aspect of grammaticalization is its core and not
the formal one, especially not the degree of morphological dependence.
What we do find, though, is a small number of category pairings in
which neither of the categories seems to be further grammaticalized
than the other and for which change in both directions, that is bidir-
ectionality, is attested. These pairings include necessity and possibility
in the domain of modality (cf. Narrog a; Section .), causal and
concessive subordination (cf. Narrog ), and marker of attributive
possession and recipient (cf. Kuteva et al. : , ).
Overall, even after three decades of discussion of potential counter-
examples, for most scholars in the field the unidirectionality hypothesis
still holds up, either as an absolute, or at least as a probabilistic
generalization on grammatical change. The interesting thing is that it
holds up for scholars from a large range of theoretical backgrounds.
The remainder of this section is devoted to presenting some of the
reasoning why grammaticalization is unidirectional.
From a functional perspective, two hypotheses are of particular inter-
est. On the one hand, it has been argued that the goal of speakers and
hearers to communicate successfully underlies grammaticalization. To
this end, Heine et al. () draw on what Werner and Kaplan (:
) call the ‘principle of the exploitation of old means for novel func-
tions’ to explain major lines of directionality in grammatical change:
By means of this principle, concrete concepts are employed in order to understand,
explain, or describe less concrete phenomena. In this way, clearly delineated and/or
clearly structured entities are recruited to conceptualize less clearly delineated or
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. UNIDIRECTIONALITY
structured entities, and nonphysical experience is understood in terms of physical
experience, time in terms of space, cause in terms of time, or abstract relations in terms
of physical processes or spatial relations. (Heine et al. : )
On the other hand, Haspelmath (, ) argues that grammatical-
ization emanates from a number of principles of communication and
behavior to which degrammaticalization would not conform. Without
going into every detail here, Haspelmath suggests that the basic prin-
ciple of verbal behavior is to “talk in a manner that you are successful in
society.” Some derived more concrete principles are to (a) talk with
clarity, (b) talk economically, (c) conform with others, and (d) be
‘extravagant,’ that is, talk in a way that you stand out. (c) and (d)
may seem contradictory at first sight, but the idea is that successful
communication is only possible by basically using the same lexicon and
grammar as the other speakers of the language. Sometimes intersper-
sing an ‘extravagant’ expression gives the speaker more attention. Now,
incipient grammaticalization fulfills the need of ‘extravagance,’ because
by using a novel expression for some grammatical category the speaker
stands out. By doing so, the principle of ‘clarity’ is also fulfilled, since
lexical expressions are as a rule semantically richer and more specific,
and therefore clearer than grammatical ones. In the course of gram-
maticalization, ‘economy’ takes over: expressions that are habitually
used in a specific environment tend to get phonetically de-emphasized
and shortened. Degrammaticalization would run against these prinici-
ples, with the exception of ‘extravagance.’ In principle, one can also
speak in a manner that stands out by using a grammatical form where a
lexical item would have been expected. But the creative use of gram-
matical expressions is much less likely than that of lexical expressions
since grammatical expressions are usually processed automatically and
unconsciously and not chosen in the way words are. It therefore takes
some extra mental effort for this to happen. With the item ism,
apparently an individual made this conscious creative effort, but it is
a different type of process than the gradual extension of the verb ‘go’
across linguistic context to acquire new grammatical meanings.
From a language processing perspective, Bybee () suggests that
grammaticalization is unidirectional because if the frequency of a
certain word or construction increases in grammaticalization, meaning
and structure and phonetic shape are reduced, or even deleted, but if
frequency remains unchanged or decreases, the opposite mechanism
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
does not take place. That is, a grammatical morpheme that for some
reason becomes obsolete, as has happened with the person endings on
the English verb (with exception of the third person -s), does not
become longer or gain in morphological complexity. Quite the oppo-
site, in many cases it is reduced to zero.
In Relevance Theory, one of the two major theoretical paradigms of
linguistic pragmatics, grammaticalization means the acquisition of
procedural meaning in addition to, or in place of conceptual meaning.
Now, as Nicolle (: ) argues, when meaning fades away it can
only be the conceptual meaning but not procedural meaning, because
procedural meaning optimizes the relevance of an utterance. Therefore
grammaticalization is unidirectional.
The idea that grammaticalization constitutes a type of optimization
can also be found in formal grammar. For Kiparsky (), grammat-
icalization is a case of grammar optimization, and this optimization is
constrained by universal principles. There is only one thing that can go
counter to this universal change, and that is exemplar-driven analogy.
In generative grammar (see Chapter for details), grammaticalization
means a change from movement to a higher category to late merge, that
is a development towards more economy, or ‘structural simplification’
(cf. Roberts and Rousseau ; van Gelderen ). The opposite
direction of change would run counter to both economy and the
fundamentals of hierarchical clause structure. More recently, grammat-
icalization has also been framed in terms of ‘feature economy’: Since it
is more economical for an item to have uninterpretable features, items
develop towards having uninterpretable features (that is being base-
generated higher up in the tree) (see van Gelderen a). In these
cases, the reasons for unidirectionality of grammaticalization are prac-
tically in-built into the grammatical systems of any language.
Lastly, Maslova () has brought forward the interesting idea that
the expansive change that is a central feature of grammaticalization
(cf. Section . on ‘context extension’) has a strong perceptive advan-
tage over reductive change in contexts when it comes to speakers
selecting a change: speakers can perceive an extension of context, that
is, when a verb like ‘go,’ developing into a future marker, is used with
new types of subjects and verbs, because these contexts are novel, but it
is very difficult to notice a reduction in contexts of use.
Note that the ‘unidirectionality’ of grammaticalization is not defined
in a sense that all changes accompanying grammaticalization are
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DIRECTION OF SEMANTIC CHANGE IN GRAMMATICALIZATION
one-sidedly reductive or one-sidedly expansive. As we will show in
Chapter , there are both reductive and expansive changes in gram-
maticalization, and while the reductive ones may be more salient, the
expansive ones are more essential. Relatedly, the idea has been floated
that approaches to grammaticalization can be divided into reductive
and expansive ones (Traugott d; see also Section ..), but except
for very early studies, especially Lehmann ( []), few approaches
to grammaticalization are exclusively reductive, and probably no
approach is exclusively expansive. Most approaches, even earlier ones,
recognize both reductive and expansive aspects of grammaticalization
(cf. Heine a).
In conclusion, we accept the hypothesis of unidirectionality as one of
the premises for many issues concerning grammaticalization in the
remainder of this book. This does of course not mean that unidirection-
ality cannot be challenged. We will return to this topic in Section ..
Further reading
For a general discussion of unidirecionality, see Hopper and Traugott (:
–). Besides the literature already mentioned in this section, Börjars and
Vincent () provide an overview of thinking on unidirectionality in gram-
maticalization; see also Norde (). There is a wide range of studies challeng-
ing the unidirectionality hypothesis in some form or other, see in particular
Ramat (), Newmeyer (), Campbell (), Campbell and Janda (),
Janda (), and Joseph (). An approach to test existing hypotheses on
grammaticalization by means of a worldwide sample of languages is presented
by Bisang et al. (). The authors conclude that while some processes of
grammaticalization are very widespread in the languages of the world, they
found no process that is common to all languages. However, counterexamples
are rare and may in fact often be artifacts of classification (Bisang et al. : ).
The findings in Saavedra’s quantative text study of synchronic and diachronic
grammaticalization also support unidirectionality (: –).
1.4 Direction of semantic change in grammaticalization
and expansion of scope
An important notion associated with unidirectionality in many models
of grammar (cf. Chapter ) is that of scope expansion. The further some
linguistic unit grammaticalizes, the more its scope expands. ‘Scope’ is
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
originally a concept from logic (cf. Lyons : ) and denotes “[the]
portion of a particular sentence which is interpreted as being affected
by an operator present in that sentence” (Trask : ).³
Let us take the example of a fictional verb denoting ‘have,’ and
undergoing grammaticalization until it becomes a past tense ending,
a path of grammaticalization that is common cross-linguistically
(cf. Bybee et al. : ). Of the lexical verb ‘have,’ we may say that
it has scope over the object NP that it accompanies; for example a
-year-old son in I have a -year-old son. The verb may grammaticalize
to become a marker of resultative aspect (I have closed the door). It then
has scope over (that is, affects the interpretation of) the verb phrase close
the door, whose temporality it indicates. Finally, going through a phase
of ‘perfect’ interpretation, it may become a past tense ending like English
-(e)d in I had closed the door, as has in fact happened in a number of
languages (see Kuteva et al. for examples). Past tense is generally
taken to have scope not only over the verb phrase but over the whole
clause. The past -(e)d in I had closed the door, for example, not only has
scope over the phrase close the door but also over the aspect have -ed. In
this manner, semantic scope keeps expanding during grammaticalization.
It is a distinguishing feature especially for ‘secondary grammaticalization,’
that is, grammaticalization from one grammatical category to the next.
While scope is essentially a logical-semantic notion, it has been
integrated into the Logical Form of generative models of syntax
around the time of the Revised Extended Standard Theory in the
s (cf. Hornstein : ; Helbig : ). Logical Form is
conceptualized as an output component of syntax, mediating between
syntax and semantics (cf. Haegeman : –). It is then
commonly assumed that in Logical Form scope is determined by
c-command. Therefore, in generative grammar, ‘scope’ may be
referred to as part of the syntax model. Nevertheless, it is essentially
the same logical notion as in other non-formal models of grammar,
and it expands during grammaticalization (cf. Roberts : ).⁴
³ As we will see in Chapter , the term has been used for a different concept in some of
the preceding literature on grammaticalization.
⁴ In his model of grammaticalization, Lehmann ( []: ) has claimed that
‘structural scope’ (also known as ‘syntagmatic weight’) shrinks during grammaticalization.
However, Lehmann’s ‘structural scope’ is something fundamentally different from ‘scope’
in the rest of linguistics. It is therefore problematic as a label and should not be confused
with scope in the logical sense (cf. Sections ., . for more detail; see also Narrog ()).
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DIRECTION OF SEMANTIC CHANGE IN GRAMMATICALIZATION
In conclusion, scope expansion is an important aspect of the
unidirectionality of grammaticalization.⁵
Now, how does this expansion of scope relate to the functional/
semantic change that we consider as the core of grammaticalization?
The grammatical categories that have wide scope and develop late in
grammaticalization—that is, not directly from lexical categories but
from less grammaticalized, earlier, grammatical categories—typically
have one of the following properties:
(i) They encode a deictic relation to the speaker; e.g. tense, episte-
mic modality.
(ii) They encode a deictic relation to the addressee; e.g. imperatives
and other mandative modalities, politeness.
(iii) They modulate the speech act; e.g. illocutionary modification.
(iv) They serve the organization of text and discourse; e.g. sentence
adverbs, conjunctions.
Examples of grammatical categories that do not share these properties,
that is, have narrower scope, have more concrete meanings that do not
deictically refer to speaker, hearer, speech act, or discourse, and come
early in grammaticalization, that is, they are often grammaticalized
directly from lexical items and constructions. They include aspect,
benefactives, possession, adpositions, and certain event-oriented mod-
alities such as ability or volition.
We label this overall movement towards speaker, hearer, speech-act,
and discourse-related functions as ‘discourse orientation’ and discuss it
in more detail in Chapter . The term ‘subjectification’, as it is com-
monly understood and used (e.g. Traugott b), refers only to a minor
part of this development, namely, to an initial stage in grammaticaliza-
tion in which meanings become expressive of the speaker’s self.
With respect to the relationship between scope expansion and dis-
course orientation we submit that it is the latter which is the driver of
change, and scope expansion is the result of it. The more concrete and
objective lexical categories are the building blocks of sentences at the core.
Grammatical categories function like the cement on the outside that
embed sentences in consecutive layers of less to more advanced
⁵ We mentioned in Section . a small number of category pairs where change is
bidirectional, e.g. possibility and necessity, and causal and concessive subordination. In
the case of these pairs, we assume that there is no significant difference in scope.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
grammatical categories, creating connections between the building blocks
and to the builders (the speech act participants, especially the speaker).
Further reading
Jäger and Rosenbach () suggested in a programmatic paper that a
psychological mechanism called asymmetric priming might be the cause
for unidirectional language change in general and grammaticalization in
particular. In simplified terms, asymmetric priming means that one meaning
or form (in this case, the grammatical one) is likely to be associated from
another one (in this case, the lexical one) but not vice versa. Against this
hypothesis, Hilpert and Saavedra () provide experimental evidence that
implies negative priming effects between grammatical and lexical meanings,
thus weakening the case for asymmetric priming. However, the authors
admit that their evidence is limited in several ways, so that asymmetric
priming cannot yet be ruled out as a cause of unidirectionality.
1.5 Grammaticalization as explanation
Grammaticalization is of great importance for the study of language for
a number of reasons. It is, of course, highly relevant in the description
of language change in general, and in the description of the genesis of
grammars in their synchronic states in particular. In the maximal case,
we may assume that practically all elements of grammar as we know
them are the result of grammaticalization, so that every functional word
or morphological segment ultimately goes back to some lexical expres-
sion. In practice, however, in any language there are grammatical items
that are so old that it is impossible to track down their origins with
certainty. We know for example, that the modals and the tense-aspect
auxiliaries in English have developed from lexical verbs, but we do not
know, and will presumably never know for certain, a lexical origin for
the genitive -s or the plural -s.
In this section, we want to show that grammaticalization is not only
relevant for language history and language description but may also
function as an explanation for synchronic states of language.
The first major area in which grammaticalization has been successfully
applied as an explanation is morpheme order, that is, the question why in
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. GRAMMATICALIZATION AS EXPLANATION
morphologically complex words and phrases each element occupies just
the place that it does. This includes the question why, cross-linguistically,
languages have far more suffixes than prefixes, not only when they are
head-final but also when they are head-initial (see Section . for details).
The hypothesis that grammaticalization is responsible for morpheme order
is known as the “fossilized syntax hypothesis”, that is, “the position of an
affix is the same as the position of the non-bound lexical or grammatical
material from which the affix developed” (Bybee et al. : ).
A good example is the position of the definite article in Bulgarian.
The language is predominantly head initial and demonstratives, being
the historical source of the articles, are preposed, that is, precede the
noun (example ()). However, the definite article is postposed, that is,
it follows the noun (example ()).
() Bulgarian (Kuteva and Heine : )
Tazi masa
this. table.
‘this table’
() Bulgarian (Kuteva and Heine : )
Masa-ta
table-the.
‘the table’
So why do demonstratives precede the noun and definite articles follow
it in Bulgarian? The answer is found in the grammaticalization of the
definite article. At the time of its grammaticalization as a definite
article, between the ninth and the thirteenth centuries, a demonstrative
could both precede and follow the noun. Example () shows it fol-
lowing the noun in a twelfth-century text.
() Bulgarian (Mirčev : ; Kuteva and Heine : )
I pridošǎ na mjasto to, ideže stoaše
and came.they on place this where stood
na kameni s̃tyotc’ Joann’ molja boga
on stone father John praying.to god
‘And they came to THIS PLACE, where the saint Father John
stood on a stone, praying to God.’
Among the two options available at that time, the demonstrative became
cliticized and affixed in the postposed position, probably for prosodic
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
reasons (it had less prominence there than in preposed position). How-
ever, in the later historical development, the position of the demonstra-
tive became fixed to the position before the noun, thus leading to the
apparent contradiction in word order in Modern Bulgarian.
A much more extensive case for grammaticalization explaining
morpheme order has been presented by Mithun (), who analyzes
the highly complex structure of the Navajo verb, which had puzzled
researchers for decades. The Navajo verb has been described in terms of
a template, consisting of a large number of fixed positions. Figure .
renders a traditional description with thirteen positions for prefixes,
followed by the verb stem.
0 Ia Ib Ic Id Ie II III IV V VI VII VIII IX X
PP Ø PP PP REFL REV SMI ITR DISTR OBJ 3 SBJ THM MODE 1, 2 SBJ CLF stem
OBJ ADV PLU ADV
NOM
Disjunct prefixes Conjunct prefixes
Figure 1.1 Navajo verb template (according to Young and Morgan : –)
() Position classes in Figure . (according to Young )
Object of a postposition [applied objects]
Ia Null postposition
Ib Postpositions [applicatives], Adverbial-Thematic, Nominal
prefixes
Ic Reflexive
Id Reversionary: ‘returning back’
Ie Semeliterative ‘once more’
II Iterative
III Distributive plural
IV Object pronominals
V Subject pronominals: third person
VI Thematic and adverbial prefixes [ slots]
VII Mode [modality, aspect]
VIII Subject pronominals: first and second persons
IX [Classifiers] (valency markers)
X Stem
According to Mithun (: –), the following four points are
especially challenging about this morpheme order.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. GRAMMATICALIZATION AS EXPLANATION
(i) Languages with verb-final syntactic structure are expected to be
suffixing. Verb-final Navajo, by contrast, is exclusively prefixing.
(ii) Mutually dependent morphemes should be contiguous, but in
Navajo some are scattered throughout the verb.
(iii) Inflectional affixes are expected to occur farther away from the
root than derivational affixes, but in Navajo derivational and
inflectional prefixes are interwoven.
(iv) Paradigmatically related affixes are expect to occur in the same
position in a template. But they do not do so in Navajo.
Mithun () argues that it is impossible to explain this order in terms
of principles of syntax or semantic scope, as other scholars had at-
tempted (Baker ; Rice ; Hale ). Instead, prefixes in Navajo
are the result of eventual grammaticalization, and their present shape
reflects their current stage of grammaticalization. The decisive principle
is: those words that were grammaticalized very early are closer to the
stem as affixes, while the more recent grammaticalizations are further
from the stem. For example, third person subjects are found in position
V, while first and second person subjects are found in position
VIII. This is due to the fact that Navajo grammaticalized the first and
second person morphologically independent pronouns first, and the
third person pronoun later. Or, subject prefixes occur closer to the verb
stem than the objects, since highly topical, frequent, unstressed mor-
phemes in a relatively fixed position were more likely candidates for
grammaticalization than objects. This can be reconstructed by the
increase in () phonological reduction, () generality and abstraction,
and () diffuse meaning from left to right in the template. In other
words, there is an increasing degree of grammaticalization from left to
right. Therefore, “the positions of prefixes in the verb correlate with
their age: those closest to the stem are the oldest, and those furthest the
youngest” (Mithun : ).
A second area where grammaticalization has been applied success-
fully as an explanation for synchronic language structures is in the
explanation of differences and commonalities of the expression of the
same grammatical category across languages. First, indefinite articles
(e.g. English a(n)) are an example of grammaticalization leading to
similar structural features across languages. In many languages, indef-
inite articles have the following properties (cf. Heine ):
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
() a. They have the same or a similar form as the numeral ‘one’, and
can occupy the same position in a clause as the numeral. But if
so, they are usually at the same time shorter than the numeral.
b. They are also usually restricted to marking singular nouns; cf.
E. a rabbit, but *a rabbits.
c. If they can mark mass nouns, they can also mark plural nouns. If
they can mark plural nouns, they can also mark singular nouns.
None of the properties in () is of any logical necessity for being an
indefinite article. For example, there are languages where the indefinite
article does not resemble the numeral ‘one,’ and with definite articles
one can see that there is no necessary restriction with respect to number
marking (e.g. the rabbit, the rabbits). There is no syntactic or semantic
or morphological rule that would cause, and therefore explain, the
properties in ().
It turns out that the reason for these properties is that the indefinite
articles with the properties in () have grammaticalized from the numeral
‘one’. This is a process that proceeds in stages. It starts out as a numeral (the
number ‘one’), then proceeds to the stage of a presentative marker, as in
(), then a specific marker as in (), and a non-specific marker as in ().
English a(n) has undergone all these stages. The language of the examples
is Modern English for the sake of convenience, but there is no lack of
indefinite markers in many languages at these stages. They are cited from
Heine (a: –), where a more detailed discussion can be found.
() Once upon the time, there was an [or: one] old woman.
() A man came in yesterday and started talking to me.
() Buy me a newspaper!
At the stage of non-specific marker, the indefinite article cannot yet
mark plural nouns, since it has still retained the semantic properties of
the numeral ‘one’. However, there are languages that have progressed a
step beyond English, and the indefinite article is now so generalized
that it can also mark plural nouns and mass nouns. This is the case in
Spanish, for example, as in ():
() Spanish (Bradley and Mackenzie : )
Unas vacaciones en Italia.
vacations in Italy
‘A holiday in Italy.’
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. GRAMMATICALIZATION AS EXPLANATION
Vacaciones ‘vacations’ is an inherently plural noun which even requires
the indefinite article.
To summarize, grammaticalization from the numeral ‘one’ explains
the properties in () for all those ‘one’-derived indefinite articles that
have not yet reached the last stage, as Spanish has: The ‘one’-derived
indefinite articles resemble the numeral because they are derived from
it, they are often shorter because they have undergone erosion (pho-
nological reduction) in the process of grammaticalization (Section .),
they cannot mark plural because they have retained the semantic
properties of the numeral ‘one,’ and marking mass nouns is the last
hurdle in the extension of their usage, because masses are not count-
able. As mentioned initially, this explains a remarkable uniformity in
properties of indefinite markers in languages across the globe, because,
as Heine and Kuteva (: ) have shown, in roughly percent of all
languages that have indefinite articles, thay are derived from the
numeral ‘one’.
Grammaticalization can also be the cause of cross-linguistic varia-
tion in the expression of grammatical categories. For example, Heine
(, , and elsewhere) has shown that in the vast majority of
languages, constructions for the expression of possession (e.g. I have
two cell phones) have grammaticalized from the source schemas listed
in ().
() Formula Source schema
X takes Y Action
Y is located at X Location
X’s Y exists Genitive
Y exists for/to X Goal
X is with Y Companion
As for X, Y exists Topic
The variation in source schemas is due to the fact that possession as
such is a fairly abstract concept, and the relationship between the
possessor and the possessed can therefore be conceptualized in several
more concrete ways.
Grammaticalization emerging from these schemas can explain both
the cross-linguistic differences in the expression of possession, and the
features that a possession construction has in a particular language. The
source of the ‘have’ possessives found in many European languages is
the ‘Action’ schema, since the original lexical meaning of ‘have’ in
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
Indo-European is ‘take,’ ‘seize.’ This explains why the possessor is
construed as the subject and the possession as the object of the main
predicate. This is by no means a logically necessary way to conceptu-
alize a relationship of possession, since a possessor is not necessarily an
agent that acts on her or his possessions. The possessor may as well be
conceptualized more passively, for example, as a location for the
possession. Indeed, in Russian, which has a possessive construction
based on the ‘location schema,’ the possessor is a location marked by a
locative preposition, and the possession is the subject as in ().
() Russian (Heine a: )
U menja kniga
at me book
‘I have a book.’
In Swahili, as in many other languages as well, the possession is
conceptualized as accompanying the possessor, and possession
is accordingly grammaticalized by means of the ‘companion schema,’
as in () with the preposition na ‘with.’
() Swahili (Heine a: )
ni-na gari.
I-be.with car
‘I have a car.’
As in the case of the indefinite articles, a possessive construction in a
particular language is not only characterized by which source schema
was chosen, but also by how far grammaticalization has advanced. The
possessive verb have in English has all but lost the ability of a regular
transitive verb to be passivized, as shown in ().
() a. I have two cars.
b. *Two cars are had by me.
Furthermore, grammaticalization beyond possession towards other
categories such as modality in the construction have to (), and
tense-aspect in the have + past participle (present perfect) construction
() has taken place.
() I had to leave early this morning. (necessity)
() Those guys have spoilt our party. (present perfect)
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. GRAMMATICALIZATION AS EXPLANATION
Similar grammaticalization beyond possession has taken place in other
languages, and is also the case with schemas other than the European
‘action schema.’ For a more ‘exotic’ example, there is the grammatical-
ization of a possessive construction to express necessity (modality) as in
the Austronesian language Rapanui of Easter Island (example ()).
This is a development not uncommon in the languages of the world (cf.
Narrog c: –).
() Rapanui (Du Feu : )
a Nua te runu i te pipi
Nua collect shells
‘Nua had to collect the shells’ (lit., ‘Nua’s collecting of the shells’)
Finally, grammaticalization has also been adduced to explain the poly-
semy of specific grammatical constructions. Recall the future derived
from a motion verb discussed in Section .. Across languages, motion
verbs are one of four common sources for future constructions. The
others are modal verbs of volition, obligation or ability, aspectual
constructions (mostly ‘imperfective’), and adverbs (cf. Bybee et al.
, ). Each source has a path that connects its original meaning
to the function of future. For example, English will originally indicated
volition, and relates to future through the path of development pro-
posed by Bybee et al. (: ) as in ().
() desire > willingness> intention > prediction
According to the Oxford English Dictionary (nd edition on CD-Rom,
will v.) English will had a ‘desire’ meaning up to the nineteenth
century, both as a main verb () and as an auxiliary ().
() Whan that thynge can not be done that thou woldest, woll that
thou cannest. (Taverner, )
‘When the thing that you desire cannot be done, desire what can
be done.’
() He . . . examines the dinner-card . . . ; points . . . to the dishes
which he will have served. (Thackeray, )
‘He examines the diner menu . . . [and] points . . . to the dishes
that he wants to get served.’
‘Willingness,’ ‘intention,’ and ‘prediction’ are still present in the Mod-
ern English use of will, as in (), (), and ().
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
() I’m sure he will help if you ask. (willingness)
() I will soon be back. (intention)
() Tomorrow everything will be fine. (prediction)
In this manner, the polysemy of will in Modern English and earlier
stages of the language can be explained with reference to its grammat-
icalization from a verb indicating desire and willingness to forming the
future in a construction with a main verb. As mentioned in Section .,
the development has even proceeded beyond future into epistemic
modality with present time reference, as in ().
() [There’s knocking on the door] That will be the postman.
Further grammaticalization of future constructions in various lan-
guages includes the development of an imperative sense (‘You will be
here at tonight!’) or use as a marker of complement clauses. In an
early paper, Bybee () suggested that there are as many as six
aspects of future markers that can be explained through a grammati-
calization approach: () it explains why it is difficult to find a single
abstract meaning for a polysemous future morpheme; () it explains the
cross-linguistic similarities of grammatical meanings by similar paths
of development and principles of historical change; () it explains
differences between morphemes in different languages with reference
to different lexical sources and different extent of change along the
universal paths of change; () it predicts possible combinations of
meanings; () it allows reconstruction of the lexical sources of grams;
and () it suggests processes in synchrony that are probably instru-
mental in explaining the changes and the nature of meaning.
The above areas mentioned as examples for grammaticalization as an
explanation are by no means intended to be exhaustive. They were
merely chosen with a focus on cross-linguistic comparison. Surely,
many more can be found in any individual language, especially when
it has a written history.
Conversely, we do not want to suggest that grammaticalization is the
appropriate means of explanation for any phenomenon in grammar. In
general, one speaks of ‘internal’ and ‘external’ explanations for gram-
matical phenomena. ‘Internal’ explanations refer to rules or features of
grammar itself, while ‘external’ explanations refer to factors located
outside of grammar. We believe that the latter are more powerful, since
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
the former are prone to be circular, that is, rules are formulated based
on empirical observations of grammar, and then used again to explain
certain features of grammar. ‘External’ explanations may refer not only
to language change, but also to phylogenetic evolution, language acqui-
sition, language processing, cognitive structure, discourse pragmatics,
or language contact. Within external explanations, grammaticalization
as such is not the ultimate explanation, because grammaticalization is
necessarily motivated by further cognitive and pragmatic principles. It
is therefore rather an intermediate layer of explanation. However, it is a
good level to look for immediately accessible explanations, since cog-
nitive and pragmatic principles lie at a higher level of abstraction. If
they are not solidly based on non-linguistic research, and depend on
language data, we run into the problem of circularity again.
Further reading
Most relevant literature has already been mentioned in the above section.
Narrog (a) provides an overview of the topic of grammaticalization and
explanation from a typological perspective. The explanatory potential of
grammaticalization is also discussed in Bybee (), Heine (), Kuteva
and Heine (), and Mithun (). On the nature of internal explana-
tions, see Kiparsky (); external, functional explanations are discussed, for
example, in Keller ().
1.6 How to study grammaticalization
There are essentially three ways in which the concept of grammatical-
ization can be used: First, diachronically, for the study of actual lan-
guage change based on the historical records of the language involved.
Secondly, synchronically, in terms of comparing degrees of grammat-
icalization between related constructions within one language or across
several languages. And thirdly, in terms of a reconstruction of gram-
matical markers and constructions at earlier stages of their develop-
ment for which no historical data is available. The first and the third
way are similar in that they both relate to diachrony. The second and
the third ways are similar in that they crucially involve criteria that
characterize steps in grammaticalization.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
1.6.1 Diachronic grammaticalization studies
Studying grammaticalization diachronically is in principle no different
to studying any other kind of language change in diachrony. It relies on
the analysis of written records of the language at different points in
time. For instance, for English substantial written documentation can
be found from the ninth century onward. Therefore, in principle one
can track back the development of some grammatical constructions or
markers for years. This is possible on the basis of the availability of
written documents and does not require electronic data. However, the
advent of historical corpora has dramatically enhanced the accessibility
of diachronic data for studies of language change. For English, for
example, there is the Helsinki Corpus that covers a period from the
ninth to the eighteenth century; there is the ARCHER (A Representa-
tive Corpus of Historical English Registers) corpus that covers a period
from the seventeeth to the twentieth century; and there are a variety of
specialized corpora (see McEnery and Hardie : –).⁶
Kranich’s (a) study of the development of the English progressive
is a good example for a historical corpus study on grammaticalization.
Kranich aimed to provide a more detailed account of the semantic
development and contexts of change of the progressive than pre-corpus
studies had achieved. The result would also help to explain features of the
progressive in Modern English, since “often apparent irregularities of
present-day linguistic forms are only explicable with reference to their
diachronic development” (Kranich a: ), in accord with what we
said in Section .. The corpus used was the British English part of the
ARCHER-, which covers a period from to .
In Modern English, the progressive has mainly aspectual and sub-
jective uses. The basic aspectual use is presenting a state-of-affairs as
ongoing, of limited duration, and usually dynamic (cf. Huddleston
and Pullum : ). The subjective meaning is often the expression
of a negative evaluation of a state-of-affairs, but sometimes also down-
toning (cf. Kranich a: –). The source of the progressive is
probably the merger between the constructions beon/wesan + V-(i)
ende, that is, a ‘be’ verb and the Old English present participle, and
⁶ See CLARIN (https://www.clarin.eu/resource-families/historical-corpora) for a list of
historical corpora in languages of the world.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
beon/wesan + on ~ in ~ a + V-ing, that is, a ‘be’ verb, a preposition, and
the Old English verbal noun. Both constructions are already documen-
ted in Old English. The participle and the noun formally merged
around the twelfth–thirteenth century, accelerating the merger of the
constructions. () is an early example, in which the ‘be’-verb and the
participle are already analyzed as a syntactic and semantic unit, as their
replacement with a ‘do’-verb in the second clause shows.
() Old English (HomS (BlHom ) (Traugott : ; Kranich
a: )
Þonne beo we sittende be þæm wege, swa se
then will.be we sitting at the way as the
blinda dyde
blind.man did
‘Then we will be sitting at the way-side, as the blind man did.’
The presence of a fully grammaticalized progressive construction in the
Celtic languages of the British islands probably helped this construction
to spread in contrast to continental Germanic languages (see Chapter
on contact-induced grammatical change). In Middle English, frequency
dropped, but the construction nevertheless continued to develop,
spreading to uses with present and past verbs and modals, to perfect
and pluperfect verbs.
The meaning and use of the progressive in Old and Middle English
did not differ much from each other, but they were quite different from
those found in Modern English. The main uses of the progressive were
(a) to indicate an imperfective or durative situation, but not primarily
associated with dynamic processes as in Modern English, and (b) a
‘subjective’ use to indicate that a specific state-of-affairs was particu-
larly remarkable. Modern English also has a subjective use, but the Old
and Middle English use was apparently more frequent and more
unspecific. Example () above would be an instance where the Old
English and Modern English uses overlap. On the other hand, () is an
example for the marking of duration, and () for subjective emphasis
where in Modern English, the simple present would be used instead.
() Middle English (Gower, Confessio Amantis; th century)
(Kranich a: )
The flood is Into the grete See rennende
‘the flood is into the great Sea flowing
‘The flood flows into the ocean.’
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
() Old English (Beowulf; ca. ) (Kranich a: )
Gyf þonne Frysna hwylc . . . ðæs morþor-hetes
If yet Frisian any of.this feud
myndgiend wære
mentioning would.be
‘if on the other hand one of the Frisians would mention this
feud . . . ’
Example () does not mark a ‘limited’ duration, but an infinite
duration, where in Modern English the present would be more appro-
priate, and () marks a general emphasis, while the Modern English
progressive subjective uses are more specialized, especially on expres-
sing some kind of complaint or criticism.
Kranich’s own detailed corpus study starts in the seventeenth cen-
tury. She shows that the normalized frequency of the progressive in
English texts grew approximately twelve times up to the twentieth
century. However, she also shows that the increase in frequency is
much more pronounced in fiction and drama than in specialist expos-
itory registers such as scientific writing. This can be seen by the
numbers in Table ..
There are some remarkable facts about the data in Table .. For one
thing, there is the preponderance of use in drama and fiction vs. science
and medical discourse, which Kranich (a: ) interprets as a sign
of association of the progressive with oral rather than typical written
Table 1.1 Rise in normalized frequency of the English progressive by register, –
(excerpt from Kranich a: )
/ / / / / / / /
Drama
Fiction
News —
Science —
Medical —
Overall
* / refers to the first half of the seventeenth century, i.e. – etc.
Note that Kranich’s (a) original table contains more registers than are displayed here. The ‘overall’ number
therefore differs from the mean of those listed in the rows above.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
Table 1.2 Rise in normalized frequency of the English progressive by use, –
(excerpt from Kranich a: )
/ / / / / / / /
Aspectual uses
Percentage of total % % % % % % % %
Subjective uses
Percentage of total % % % % % % % %
* As in Table ., / refers to the first half of the seventeenth century, i.e. – etc.
language. Secondly, from a methodological point of view, it is remark-
able that in no single register a linear increase in frequency is found,
although in the overall data, we do get a linear increase. This has two
possible implications. First, empirical language data often do not give
the smooth and linear results that one would theoretically or ideally
expect. Secondly, even in a large corpus, frequencies may not be
entirely reliable, as they depend on the texts chosen and the preferences
of their authors. Take for example, the exponential rise of the progres-
sive in medical writing in the first half of the nineteenth century and its
ensuing decline. This may accurately reflect the development of the
progressive in medical writing, but it may also be a function of the fact
that medical writing only makes up a small part of the corpus and the
actual numbers are rather small. Table . shows the change in fre-
quency by meaning, divided roughly into ‘aspectual’ and ‘subjective.’
The data imply that during the rapid rise of the progressive up to the
nineteenth century the subjective uses lost share to the aspectual uses but
very recently have risen again. The latter is mostly due to the rise of a very
specific subjective use, labeled as ‘interpretive’ by Kranich (a: ). In a
complex sentence, the clause with the progressive aspect “provides the
speaker’s subjective interpretation of what [the state-of-affairs in the first
clause] would mean for him personally”. () is an example.
() You are helping me, darling, you’re being an angel (Cower, The
Vortex)
The progressive in the first clause, are helping, is an aspectual progres-
sive, describing an ongoing situation, while are being in the second
clause is an instance of the interpretive subjective progressive.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
Besides the general change in frequency, the focus of Kranich’s study
is also on change of contexts of use and functions of the construction.
With respect to its functions, the most important point may be that the
progressive shifted from a marker of a durative situation and the
‘remarkableness’ of a situation, to marking progressive aspect, that is,
mainly to a situation that is dynamically in progress, while also further
developing the non-aspectual subjective meanings. Kranich (a:
) dates this change to the period between and . On the
other hand, the accompanying extension of contexts is much less
dramatic than one might expect. For example, the change in situation
types, or subjects (animate vs. inanimate), or adverbs used with the
progressive is not very significant. There are also only two clear-cut
instances of contexts that were added. First is the passive progressive as
in (), which shows one of the earliest instances of documented use.
() I have received the speech and address of the House of Lords;
probably, that of the House of Commons was being debated
when the post went out (, A Series of Letters of the First Earl
of Malmesbury, Letter from Mr. Harris to his mother, first cited
by Warner : )
Secondly, the progressive with be and have as the main verb emerged in
the eighteenth and nineteenth centuries.
() . . . but this is being wicked, for wickedness sake ( Johnston,
Chrysal II . x., example from Denison : )
With be and have, the progressive appears to trigger a dynamic reading
of the stative predicate, as in (), or a non-aspectual subjective read-
ing. This extension of context, together with the passive progressive, is
significant, because it means that the progressive can now be used with
practically all kinds of predicates in declarative clauses.
Lastly, another significant development is a shift from subordinate
clause use of the progressive as late as at the beginning of the eighteenth
century to prevalent use in main clauses in the twentieth century,
especially in the spoken registers.
Kranich’s (a) study is but one example of how research on
grammaticalization can be conducted fruitfully through corpora.
There are a number of things that only a corpus study can achieve,
and these pertain mainly to frequency. As Mair () points out, it is
only possible with corpora to show whether a rise in frequency is
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
involved in an instance of grammaticalization, and what kind of
frequency this is. Also, low-frequency grammaticalizations are very
hard to track without large corpora. Furthermore, only with corpora,
the relationship between registers and grammaticalization can be
shown clearly. Lastly, and no less importantly, the use of corpora and
concordances makes it easier to identify contexts of change and the
emergence of patterns. Search results can even bring patterns to light
that the researcher was not aware of in the first place.
On the other hand, even a large corpus study often has to rely on
data and observations from outside the corpus, and qualitative obser-
vations may be more important than quantitative ones. Note that many
of the examples cited above from Kranich’s study were not from the
corpus study itself but from previous research. More generally, while
corpora can make the diachronic study of grammaticalization much
easier, they do not offer a magic solution for fundamental problems
associated with the historical study of language based on written docu-
ments. These can include documentation gaps and a reliance on spe-
cific (especially written) registers. Also, the data may be much less
clear-cut than expected. For example, in neither Table . nor
Table . is there a clear continuous rise in frequencies for any single
criterion except in the ‘overall’ column. Likewise, the frequency of use
with specific situation types or subjects did not change significantly. Or, it
may be surprising that there was hardly any detectable change in fre-
quency and use in the progressive construction in Old English and
Middle English, that is, over a very long period of time. Lastly, highly
grammaticalized constructions are often too old to allow tracking of the
whole history of grammaticalization with actual historical data. The
progressive construction was already grammaticalized by Old English.
The situation is even more extreme with highly grammaticalized items
such as inflectional endings. Their grammaticalization usually takes mil-
lennia rather than centuries, and they can rarely be reliably tracked
through historical records to their very beginnings (cf. Haig ). But
this is a general problem that exists independently of the use of corpora.
1.6.2 Grammaticalization as a synchronic concept
It is quite common to say that a certain construction or word or
morpheme is ‘grammaticalized’ without any mention of its actual
history. In this case, the term ‘grammaticalized’ refers to a synchronic
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
state as the result of a diachronic process that we may or may not have
knowledge of. For example, one can legitimately say that the English
progressive construction is ‘grammaticalized’ as a progressive without
knowing any of the historical details presented in the previous section.
This can be done with reference to parameters of grammaticalization
that are discussed in detail in Chapters and of this book. Most
fundamentally, the progressive construction is the exponent of the
grammatical category of ‘progressive,’ which in turn is in contrast to
unmarked aspect and perfective aspect across all verb classes.
It is also possible to say that a certain construction or marker is
‘more’ or ‘less’ grammaticalized than a functionally similar marker or
construction. Indeed, it is often within the context of a comparison of
two or more markers of the same category in a synchronic state of one
or more languages, that the terms ‘more’ or ‘less’ grammaticalized are
used. The following comparisons seem to be particularly common:
. Two or more markers or constructions of the same category in
one language (e.g. “In language X, past tense is fully grammati-
calized but future tense is not”).
. Two or more markers or constructions of different categories in
one language (e.g. “In language X, markers of tense are more
grammaticalized than markers of aspect”).
. Two or more markers or constructions of the same category in
two or more languages (e.g. “In language X, the progressive aspect
is more grammaticalized than in language Y”).
The closer the functional and the genetic relationship between the
markers or constructions compared, the more reliable such a compar-
ison can be. Conditions of grammaticalization and grammatical struc-
tures may even differ for different categories in one language, and
usually differ strongly in unrelated languages.
A good example for a synchronic comparison of one category across
related languages is ‘threaten’ verbs in Indo-European languages. Com-
pare examples () and (). () is the literal and ungrammaticalized
use of threaten in English, while threaten in () has apparently under-
gone some grammaticalization.
() The bouncer threatened us to call security if we didn’t leave.
() The Australian dollar threatens to fall below cents.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
In (), in contrast to (), clearly, no actual ‘threatening’ is taking
place, because the sentence has an inanimate subject that is incapable to
commit the act of threatening. This fact as such merely indicates a
metaphorical use of the verb threaten, where an inanimate subject
referent is conceived metaphorically as an agent. But there is also no
possible experiencer, corresponding to us in (), and the meaning of
threaten here is ‘immediacy,’ that is, an aspectual category. () could
be paraphrased as ‘the Australian dollar is about to fall below cents’
with the negative evaluative connotation. This kind of usage can be
observed in languages across Europe.
In the case of these ‘threaten’ verbs, stages of grammaticalization can be
identified with specific constructions. The constructions are as follows:
C: Source construction: ‘Threaten’ has an animate subject, an
optional experiencer object, and an optional undergoer object
or complement clause. The meaning is literal: ‘threaten’.
C: The subject is inanimate instead of animate, and either the
subject or the object refer to an imminent event.
C: The subject is inanimate, the ‘threaten’ verb accompanies an
infinitival complement specifying the imminent event.
C: The nature of the subject (inanimate vs. animate) becomes
unrestricted.
() may represent C, and () C. The following examples from
Portuguese, a language that has developed all four constructions, show
C and C.
() C: Portuguese (de Lima , cited in Heine and Miyashita
: )
A firma ameaça falência.
the firm threatens bankruptcy
‘The company is threatened by bankruptcy.’
() C: Portuguese (de Lima , cited in Heine and Miyashita
: )
Um gordo e rubicundo merceeiro [ . . . ] ameaçava
a fat and reddish merchant threatened
estalar tôdas as costuras da farda.
to.tear all the seams of costume
‘A fat, reddish trader was about to burst out of all seams of his attire.’
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
Table 1.3 ‘Threaten’-constructions in some European
languages (Heine and Miyashita : )
C C C C
Portuguese + + + +
German + + + +
English + (+) + +
Danish + + +
Estonian + + +
Romanian + + (+)
Russian + +
Greek + +
Table . shows some modern European languages, and the type of
‘threaten’ construction that can be used in each language, marked
by “+”. Brackets on the “+” sign indicate some restrictions on the
construction.
The languages in Table . are intentionally arranged in the order of
which constructions they have. All European languages that were inves-
tigated had C and C. These are obviously less grammaticalized uses. C
and C are more grammaticalized, and the data suggest that in turn C is
more grammaticalized than C. The order between C and C is deter-
mined by the fact that C is based on a non-literal meaning that pre-
supposes extension of context from animate to inanimate subjects.
In this manner, the fact that a construction in a specific language is
more grammaticalized than the related construction in a related lan-
guage can be inferred from comparative synchronic data. It can also
be derived from the application of criteria of grammaticalization
(Chapter ): The meanings and functions of the constructions in C,
C, and C are increasingly more abstract, based on metaphor, and
presuppose an extension of contexts of use. Finally, in the case of
European languages, the historical facts are also available to confirm the
hypothesis based on synchronic data. Historical data from German show
that in this language, there was only C before the sixteenth century. C
rose in the sixteenth century, C in the seventeenth century, and C in the
late eighteenth century (see Heine and Miyashita ).
The semi-modals in English provide another, short example of gram-
maticalization of several exponents of one category in one language.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
While the modals (e.g. must, can, may) are generally considered as fully
grammaticalized exponents of modality, the so-called semi-modals (e.g.
had better, would rather, have got to, be to) and lexico-modals (e.g. have
to, need to, want to) are not as clearly grammaticalized and exhibit variant
degrees of grammaticalization among them.
Collins (: –) argues that among the semi-modals, have got
to and had better are more grammaticalized than would rather and be
to. The claim is based on the following evidence:
(i) Phonological reduction: have and had are not only often re-
duced to /v/ and /d/, respectively, but are sometimes even
entirely deleted (gotta, better).
(ii) In the case of have got to, the form gotta obscures the infinitive
marker to, resulting in a bare infinitival complement, similar to
the modals.
(iii) The meaning component of ‘comparison’ is already bleached
out in the case of had better but still present in the case of would
rather. Thus, the following clause pattern is infelicitous with
had better but possible with would rather in Modern English:
[Someone] would rather/*had better [do X] than [do Y].
(iv) Have got to has clearly developed an epistemic sense, while
would rather clearly has not. (With had better and be to, the
situation is less clear.)
(v) Would rather can take a finite clause complement (I would
rather that you went with someone else), and be to exhibits
subject agreement (I am to go), both properties indicating a
low degree of grammaticalization.
Note that all these arguments can be stated independently of actual
diachronic evidence.
In conclusion then, we have seen some examples of how the concept
of grammaticalization can be applied to synchronic data. This is some-
times done simply in order to state whether some morpheme or
construction is a grammaticalized part of synchronic grammar. But
more commonly it is done to compare the grammatical status of two or
more morphemes or constructions. To achieve a valid comparison, it is
important to narrowly delimit the range of morphemes and construc-
tions to items that are semantically/functionally related, and to one
language or a structurally comparable set of languages. As we have seen
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
for the English semi-modals, even when the comparison is restricted to
one language and one category, there may be criteria that cannot be
fully applied to all items under comparison, or the results may be
ambiguous.
Lastly but no less importantly, let it be noted that there is another
way to study grammaticalization synchronically, namely in terms of
synchronic variation leading to grammatical change. This is a topic we
will take up again in Section ..
1.6.3 Reconstruction
Grammaticalization applied to language reconstruction combines
properties of grammaticalization studies on diachronic data
(Section ..) and synchronic data (Section ..). While the purpose
of study is diachronic (i.e. historical reconstruction), the data as the
basis for the reconstruction may be mainly synchronic.
Traditionally, two methodologies are distinguished in reconstruc-
tion, namely, (a) internal reconstruction, and (b) the comparative
method. While (a) is normally restricted to the diachrony of a single
language, (b) deals with a set of genetically related languages. Likewise,
both approaches are about variation in linguistic expressions, namely,
(a) typically pertains to variation within a single language, and (b) to
variation among genetically related languages.
The comparative method is concerned with the reconstruction of ear-
lier states of a language or of languages based on regular sound corre-
spondences between related languages or dialects. On the basis of such
correspondences it is possible to reconstruct earlier lexical and gram-
matical forms, which in turn enable students of grammaticalization to
test, refine, or strengthen their hypotheses on grammatical change (see
Mithun for examples from the Yup’ik and Cherokee languages).
Internal reconstruction relies on the exploitation of patterns in the
synchronic grammar of a language or dialect to recover information
about its prehistory (Ringe : ). A major concern of internal
reconstruction is “with facts that make no sense, facts that do not
cohere, most commonly irregularities in an otherwise regular para-
digm” and it is hypothesized that regularity or “coherence must have
been there at some earlier evolutionary or diachronic stage” (Givón
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
: ). How internal reconstruction can contribute to reconstruct-
ing grammaticalization is demonstrated in detail by Givón () with
an example of clitics in the Athabaskan language Tolowa. His general-
izations guiding the application of internal reconstruction to those
clitics include the ones listed in ().
() Principles of cliticization (Givón : )
a. Functional relevance: Morphemes tend to cliticize to stems to
which they were functionally relevant at the time of
grammaticalization.
b. Serial position: Morphemes tend to cliticize at the syntactic
position they occupied—as words—at the time of
grammaticalization.
c. Clausal context: Grammatical and morphological innovation
tends to occur in the most neutral clause type (main, declar-
ative, affirmative, active). Conversely, non-neutral clause
types often preserve frozen relics of older grammatical stages,
in both morphology and word-order.
d. Supra-segmental effects: Morphemes cliticize to large lexical
stems because they are de-stressed and thus cannot stand
alone as phonological words.
e. Bleaching [= erosion]: Once a morpheme is de-stressed and
cliticized, its phonetic erosion is accelerated.
The way the two approaches can be jointly applied with principles of
grammaticalization to reconstruct grammatical change can be illus-
trated with the following example (see also Heine a: ). The
Bantu language Swahili of Eastern Africa has a future tense prefix -ta-
(), illustrated in (a), and the latter has an allomorph -taka- in
relative clauses, cf. (b).
() Swahili (Bantu, Niger-Congo; own data)
a. a-ta-ku-ja kesho.
---come tomorrow
‘S/he wants to come tomorrow.’
b. a-taka-ye ku-ja kesho
--. -come tomorrow
‘s/he who will come tomorrow’
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
Internal reconstruction suggests that of the two allomorphs -ta- and
-taka-, the latter is the older one and that at some earlier stage the future
marker had the form *-taka-. This reconstruction is based, on the one
hand, on the following typological observation: Grammatical forms
tend to lose rather than to gain phonological substance (see the param-
eter of erosion in Section .). On the other hand, it is based on the
observation that—other things being equal—neutral clause types such
as main clauses tend to be more innovative than non-neutral clause
types, such as subordinate clauses (Givón : ; see (c) above).
Accordingly, grammatical forms occurring in relative clauses are likely
to be more conservative than corresponding forms in main clauses.
If we assume that the earlier form was *-taka-, typological observations
on grammaticalization suggest then that this form may be related to the
Swahili volition verb -taka ‘want’, illustrated in (()). Concretely, in
many languages across the globe verbs for ‘want’ and ‘desire’ have
developed into future tense markers, the English future marker will
being a case in point (Bybee et al. ; Bybee et al. ; Kuteva et al.
, > ). Thus, there is reason to hypothesize that the verb
-taka ‘want’ and the future tense marker *-taka- are historically related.
() Swahili (Bantu, Niger-Congo; own data)
a-taka ku-ja kesho.7
-want -come tomorrow
‘S/e wants to come tomorrow.’
This raises the question of which of the two, the verb or the tense
marker, is older. For example, could the verb not have been derived
from the tense marker? The question can be answered in the negative
by means of the comparative method, which allows establishing that
the verb must be older than the future tense marker. The application of
this method shows that the verb -taka- can be reconstructed back to the
verb *-càk-a ‘desire’ of Proto-Bantu, which is the hypothetically recon-
structed ancestor of all the -plus modern Bantu languages, includ-
ing Swahili (Guthrie –). The future tense marker, by contrast,
cannot be traced back to Proto-Bantu, which suggests that it must have
arisen after the split of Proto-Bantu and, hence, must be younger.
⁷ The item a- in () is a portmanteau morpheme consisting of the noun class marker
a- plus the tense marker -a-.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. HOW TO STUDY GRAMMATICALIZATION
On the basis of these methodological tools it is possible to formulate
a strong hypothesis to the effect that the Swahili future tense marker
-ta- is the result of a common grammaticalization process. Using the
parameters that will be described in Chapter , this process can be
reconstructed as follows:
(a) Context extension made it possible for the verb -taka ‘want’ to be
extended to allow for non-human, or inanimate subject referents.
(b) In such contexts, the lexical meaning of volition was lost (dese-
manticization), giving way to an interpretation with reference to
future events.
(c) Decategorialization had the effect that the verb lost salient mor-
phosyntactic features. On the one hand, it lost its ability to occur
in the past tense or to take nominal complements, being
restricted to infinitival verbs as complements. On the other
hand, it also lost its status as a free form and turned into a prefix
of the infinitival main verb.
(d) While the full form -taka was retained in relative clauses, in main
clauses the form underwent erosion, and was reduced to -ta-.
In accordance with these parameters, the history of the Swahili future
tense marker -ta- can, on the one hand, be understood as one of
‘attrition,’ being characterized by losses of semantic, morphosyntactic,
and phonological substance. On the other hand, it was also one of
gains, in that the language was enriched by a new grammatical category
and new means of expressing temporal distinctions.
To conclude, it is always desirable to draw on historical records. But
the above example may have served to show that such records are not a
requirement for the reconstruction of grammatical change as long as all
the methodological tools that are potentially useful for supplementing
and supporting reconstruction work are exploited. Among these tools,
the comparative method and internal reconstruction play an important
role, especially since they also allow the reconstruction of earlier states
of language use without having to rely on written or other historical
documents and, hence, can readily be applied in languages without any
tradition of writing. These are not the only approaches that are useful
for sound reconstruction. For example, in some cases, contact linguis-
tics can provide an additional tool for reconstruction. We will return to
this issue in Chapter . Note further that the methodology discussed
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
INTRODUCTION
was not meant to ‘prove’ that grammatical change has taken place but
rather to search for evidence that makes it possible to formulate as
strong a hypothesis of grammatical change as possible.
Further reading
There are many individual studies of grammaticalization with the help of
diachronic corpora. Mair () and López-Couso (a, b) may
provide convenient entrance points. For a corpus-based reconstruction of
future tenses in Germanic languages, combining text data with Construction
Grammar, see Hilpert ().
With respect to grammaticalization as a synchronic concept, or degrees of
grammaticalization, there is a classic paper by Lehmann () comparing,
among others, prepositions at different stages of grammaticalization. La-
miroy and De Mulder (), Carlier et al. (), and Fagard and Mardale
() present intriguing hypotheses about degrees of grammaticalization
across Romance languages.
1.7 Overview of the remainder of the book
Chapter has provided an introduction to the most fundamental
aspects of grammaticalization. The theme of ‘fundamentals’ is contin-
ued in the following four chapters. While Chapter reviews the para-
meters and criteria of grammaticalization that have been suggested in
the previous literature, Chapter offers, in some detail, a discussion of
the most important parameters of grammaticalization proposed in this
book. Chapter deals with the semantic directionality of grammatical-
ization. In most approaches to grammaticalization, semantics and
pragmatics are understood as the drivers of this change, so this is also
an essential issue. Chapter discusses steps and phases in grammati-
calization, including the concept of cyclical change.
Chapters to discuss the mechanisms and motivations for gram-
maticalization. While Chapter deals with mechanisms and motiva-
tions in general, Chapter elaborates on grammaticalization and
language contact. While in some countries there is a relatively stable
monolingualism, in many if not most of the countries of the world,
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DISCUSSION POINTS
people live and use language under conditions of language contact. This
also impacts grammaticalization. Chapter deals with the influence of
typological features of language on grammaticalization. But we will see
that there is also some influence in the opposite direction, namely from
grammaticalization to structural typological features of languages.
Chapter is a standalone chapter discussing a number of theory-
specific approaches to grammaticalization. While most of the content
of this book is not theory-specific, or could be described as ‘theory-
neutral,’ there are a number of linguistic theories whose application
leaves a distinct imprint on the study of grammaticalization. Chapters
and deal with issues ‘beyond’ grammaticalization. These include
the well-discussed processes of lexicalization, degrammaticalization,
and exaptation, besides a number of changes less related to grammat-
icalization in Chapter , and the development of discourse markers in
Chapter . The latter has also been known under the label of ‘prag-
maticalization’ and has led to questioning the boundaries of grammat-
icalization. Chapter concludes the book.
Discussion points
() Few concepts or categories in the study of language are predefined by the
phenomenon itself. Researchers are in principle free to define their con-
cepts and categories one way or the other, but not all definitions will be
equally useful. What are the ramifications of defining grammaticalization
differently, for example, in terms of form only, or function only or their
combination, or discourse or usage etc., for the study of grammaticalization?
() Make a list of phenomena in language that you think may be amenable to
an analysis in terms of grammaticalization and a list of phenomena that
are probably not useful. Compare and discuss the ambivalent cases.
() Take the example of a well-known case of grammaticalization, such as a
go-future (English be going to) or the progressive, and discuss how a
synchronic analysis in terms of grammaticalization and a diachronic anal-
ysis may lead to the same or a different result.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
2
Criteria, parameters,
and other variables
2.1 An overview
2.2 Discussion
2.3 Conclusions
Discussion points
In Chapter , we addressed the question of what grammaticalization is.
When looking in more detail at questions such as how far a linguistic
item has grammaticalized at a specific point in history, or in which
respect one can say it has grammaticalized, we need to set up criteria of
grammaticalization. These criteria of grammaticalization also serve as
clues for the reconstruction of the history of grammatical forms in the
absence of written documents. The criteria, and their application to
reconstruction, are the main issues we look into in the following two
chapters.
The present chapter is restricted to an overview of studies that have
been proposed to analyze processes of grammaticalization. Our focus
will be on terms and concepts distinguished to identify the main com-
ponents of change. These components are most commonly described by
means of notions like ‘criteria,’ ‘parameters,’ and ‘principles.’
Section . introduces the main approaches that have been proposed.
In doing so, we have to be restricted to the main lines of previous
work, ignoring many other useful distinctions that have been made,
while Section . is devoted to a comparative appraisal of the various
approaches.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. AN OVERVIEW
2.1 An overview
The most influential approach, taking structural synchronic notions as a
basis for analyzing grammatical change is that of Lehmann ( [];
see also Norde ).¹ For Lehmann, grammaticalization is essentially a
decrease of the ‘autonomy’ of a ‘linguistic sign.’ Distinguishing a para-
digmatic and a syntagmatic aspect of grammaticalization, and juxtapos-
ing them with the parameters ‘weight,’ ‘cohesion,’ and ‘variability,’
Lehmann ( []: ) arrives at the six parameters or criteria
listed in Table . and described below.
Table 2.1 The parameters of grammaticalization (Lehmann []: )
Axis
Parameter
Paradigmatic Syntagmatic
Weight Integrity Structural scope
Cohesion Paradigmaticity Bondedness
Variability Paradigmatic variability Syntagmatic variability
Integrity, or paradigmatic weight, means that a given sign has a certain
amount of semantic and phonological substance which allows it to
maintain its identity. From a diachronic perspective, this parameter has
two components, namely decrease in semantic integrity (desemanticiza-
tion) and decrease in the phonological integrity (phonological attrition)
of a sign. These components can be illustrated with the following
example (Lehmann []: ): The Latin preposition dē ‘down
from (the top)’ developed into a genitive marker in French and other
Romance languages, for example, French de (‘of ’). Decrease in semantic
integrity had the effect that both the delative meaning ‘down from (the
top)’ and the motion component were lost, while decrease in phonolog-
ical integrity meant that dē was reduced in some contexts to d.
Paradigmaticity, or paradigmatic cohesion, is defined as “the formal
and semantic integration both of a paradigm as a whole and of a single
subcategory into the paradigm of its generic category.” Diachronically,
¹ This approach was already presented in Lehmann () but since it was republished
with some corrections in later versions, we are referring here exclusively to the latest
version of .
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CRITERIA, PARAMETERS, AND OTHER VARIABLES
it means essentially that with increasing grammaticalization a form
tends to shift from a paradigm with a larger number of members to one
with a smaller number of members. For example, English secondary
local prepositions such as beyond, before, within, amidst, etc. form a
larger paradigm than primary local prepositions such as in, on, at, from,
or to. On the other hand, the parameter refers to the integration of a
sign into an inflectional paradigm.
Paradigmatic variability refers to the freedom of choosing a sign,
typically either from the same paradigm (intraparadigmatic variability)
or choosing no member of that paradigm (transparadigmatic variabil-
ity). The latter relates to relative degrees of obligatoriness of the sign.
That grammaticalization leads to obligatorification, that is, to decrease
in paradigmatic variability can be shown with the development of
articles from Latin to the modern Romance languages, which was
characterized by a decrease in freedom and an increase in the obliga-
tory use of a demonstrative (Latin ille ‘that (masculine)’) and a numeral
(Latin unus ‘one (masculine)’) as definite and indefinite articles, respec-
tively (Lehmann []: ).
Structural scope, or syntagmatic weight of a linguistic sign, is defined
as the structural size of the construction which it helps to form. The
structural size in turn reflects its level of grammatical structure, which is
determined by the syntagmatic morphosyntactic relation(s) which it
contracts. A full verb, for example, may contract a morphosyntactic
relation with a VP as a complement, while a tense ending only contracts
a relationship with the verb stem to which it is attached. In this manner,
if a linguistic sign undergoes a typical morphological process of gram-
maticalization, from verb to auxiliary to clitic or affix to inflectional
ending, its structural size (“scope”) continues to shrink.
While this is a valid concept in consonance with Lehmann’s
overall concept of grammaticalization as loss of autonomy and reduction
of a linguistic sign, it happens to be a fundamentally different concept than
‘scope’ in linguistics outside Lehmann’s parameters of grammaticalization
(see Section . below). Therefore, ‘structural scope’ is rather misleading as
a label and should be replaced by ‘structural size’ or ‘constituent size,’
corresponding to the actual concept behind it. Another problem with this
notion is that it does not lead to the hypothesized result (condensation/
reduction) if applied to linguistic signs that do not both undergo morpho-
logical reduction and enter morphological paradigms. Examples include
manner adverbs becoming sentence adverbs (cf. Traugott ), nouns
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. AN OVERVIEW
becoming conjunctions (cf. Heine and Kuteva : –), or preposi-
tions becoming conjunctions (cf. Heine and Kuteva : –). In each
of these cases, the linguistic unit with which the grammaticalizing linguistic
sign combines is of a higher level (e.g. clause in the case of sentence adverbs
and conjunctions) than before. The only way to deny these systematic
exceptions would be to claim that the concept of structural size (“structural
scope”) does not apply to them (cf. Narrog for a more detailed
discussion of Lehmann’s ‘structural scope’ vs. scope in general linguistics).
Bondedness, or syntagmatic cohesion is defined as the intimacy with
which a sign is connected with another sign to which it bears a syntag-
matic relation (Lehmann []: –). Increase in bondedness,
involving univerbation (boundary loss) and coalescence (increase in
morphophonological integration), is said to proceed along a scale from
(a) cliticization, via (b) agglutination (affixation), to (c) fusion or merger,
where the sign loses its morphological identity (Lehmann []:
). In the grammaticalization of the Latin demonstrative ille, the
proclitic French definite article illustrates (a) while the suffixal Romanian
definite article instantiates (b). One result of (c) is seen in suppletion
patterns such as English tooth vs. teeth, or sing vs. sang.
Syntagmatic variability is defined as “the ease with which a sign can
be shifted around in its context.” It decreases with increasing gram-
maticalization (Lehmann []: –). A verb enjoys some
positional freedom vis-à-is the VP with which it combines. Once the
verb turns into an auxiliary, it will lose much or all of the freedom it had
before grammaticalization.
Rather than resting on synchronic notions and morphosyntactic cate-
gories, the parameters proposed by Heine and Kuteva (: –;
see also Heine and Kuteva ) capture salient features associated
with grammatical change. They rest on the assumption that grammat-
icalization simultaneously affects all major components of the linguistic
expressions concerned, extending from sound to meaning and from
pragmatics to syntax. The four parameters distinguished are briefly
sketched below; for a more detailed discussion, see Chapter .
Context extension (or, in short, extension) refers to the rise of new
meanings when linguistic expressions are extended to new contexts,
leading to context-induced reinterpretation.
Desemanticization (‘semantic bleaching’) leads to the loss or gener-
alization of meaning content or functions.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CRITERIA, PARAMETERS, AND OTHER VARIABLES
Decategorialization, leading to the loss of morphosyntactic proper-
ties characteristic of lexical or other less grammaticalized forms.
Erosion (‘phonetic reduction’) leads to the loss of phonetic, includ-
ing prosodic, substance.
The ordering of the four parameters reflects the diachronic sequence in
which they typically apply: Grammaticalization tends to start out with
context extension, which triggers desemanticization, and subsequently
decategorialization and erosion. Erosion is the last parameter to be
involved, and in many cases of grammaticalization it is not, or not
yet, a relevant parameter.
Hopper (: –) proposed five principles for identifying potential
instances of grammaticalization, or ‘grammaticization’ in his terminology.
They first and foremost refer to the incipient stages of grammaticalization
and the more variable patterns of language. These principles are:
Layering means that new layers of structure arising via grammati-
calization may coexist with older layers within a broad functional
domain of a language, with the two, or more kinds of layers interacting
with one another.
Divergence obtains when a lexical form undergoes grammaticalization
but survives as an autonomous lexical element, occurring side by side
with its grammaticalized variant. For example, a verb developing into an
auxiliary may still retain its status as a lexical element in other uses.
Specialization refers “to the narrowing of choices that characterizes an
emergent grammatical construction.” Thus, at one stage within a func-
tional domain there may be a variety of choices which narrows and the
smaller number of forms assume more general grammatical meanings.
Persistence “relates the meaning and function of a grammatical form
to its history as a lexical morpheme.” For example, with the change
from lexical to grammatical function, traces of the lexical meaning can
adhere to the form and be reflected in constraints on its grammatical
distribution.
Decategorialization leads towards the loss of discourse autonomy
of a form, where loss of autonomy also includes functional-semantic
shift.² Forms undergoing grammaticalization tend to lose or neutralize
the morphological markers and syntactic privileges characteristic of full
categories and assume attributes characteristic of secondary categories.
² For a slightly different use of ‘decategorialization’, see Section ..
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. AN OVERVIEW
Comrie () suggests that the terms ‘grammatical’ and ‘grammati-
calized’ have the same referents, differing only in whether the phenom-
enon is viewed purely synchronically or diachronically. He proposes a
catalog of three essentially synchronic criteria to distinguish what is or
is not grammatical and for describing grammaticalization. These are
the following:
Obligatoriness: A particular expression can be required to differing
extents, being required in some environments but not in others.
Boundness: A particular expression may be more or less bound. For
example, affixes are more bound than clitics.
Interaction with (the rest of) the grammar: This applies, for exam-
ple, when a feature of nouns interacts with the rest of the grammar
through agreement.
In clear cases of grammaticalization all three criteria are met. The
English distinction of nominal number (dog vs. dogs) is presented as
such a clear case of grammaticalization: First, English nouns are oblig-
atorily marked for number, since every noun is marked for singular or
plural (obligatoriness). Second, the plural marker -s is bound (bound-
ness), and, third, the number distinction in nouns interacts with the
rest of the grammar through agreement (this dog vs. these dogs).
Last but not not least, Harris and Campbell (: –, ) pro-
posed the following three what they call ‘mechanisms’ of change in
their theory of syntactic change. These are not strictly speaking criteria
or principles, but have nevertheless been discussed in the same vein:
Reanalysis: A process that results in changes in the underlying
structure of a syntactic pattern but does not involve any modification
of its surface manifestation.
Extension: A process that results in changes in the surface manifes-
tation of a syntactic pattern but does not involve immediate or intrinsic
modification of underlying structure.
Borrowing: Replication of a syntactic pattern which is incorporated
into the borrowing language through the influence of a host pattern
found in a contact language.
The use of the term ‘reanalysis’ in the theory of these authors overlaps
with that found in some studies of grammaticalization (see in particular
Hopper and Traugott ). However, in the latter, distinctions between
syntactic concepts such as ‘surface manifestation’ and ‘underlying
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CRITERIA, PARAMETERS, AND OTHER VARIABLES
structure’ are not generally made. The term ‘extension,’ as used by Harris
and Campbell (: ), also shows some overlaps with the way the
term is used in some studies of grammaticalization, but there are also
differences (see, e.g., Chapter .). Harris and Campbell discuss a range
of grammaticalization processes and also propose some alternative ana-
lyses for such processes. They offer the only framework among those
discussed in this section that deals with borrowing, or more precisely,
with grammatical replication. As we will see in Chapter , grammatical
replication constitutes in fact an important factor in shaping grammat-
ical change.
Further reading
The studies looked at in this section present only a spectrum of the terms and
generalizations that exist. Heine and Reh (: –) represent an early
effort to bring together the terms figuring above, as well as a number of
others. The significance of the criteria proposed by Lehmann ( [])
has been discussed in a number of studies, such as Norde (). Rather than
relying on established parameters such as those of Lehmann ( []) or
Heine and Kuteva (), Norde and Beijering () propose a clustering
approach involving a set of what they call ‘primitive changes.’
2.2 Discussion
The notions and terms discussed above to describe regularities in
grammatical change differ from one another in a number of ways.
On the one hand, differences may arise with respect to terminology,
in that, for example, regularities are referred to as ‘principles’ by
Hopper (), ‘parameters’ by Lehmann ( []) and Heine
and Kuteva (), ‘criteria’ by Comrie (), and ‘mechanisms’ by
Harris and Campbell ().
On the other hand, the notions proposed by the various authors
also differ greatly with regard to whether they define synchronic or
diachronic phenomena. Comrie () proposes strictly synchronic
terms, even if grammatical change is implied: Obligatoriness, boundness,
and interaction with the grammar refer to synchronic states of a given
language—hence, there is no need in his approach to draw on diachronic
evidence and reconstruction. Much the same applies to Lehmann’s
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCUSSION
parameters, which can be understood as capturing synchronic states.
However, Lehmann ( []: table .) also proposes processes
correlating with his parameters. For example, his parameter of integrity
entails attrition, paradigmatic variability entails obligatorification, struc-
tural size (“structural scope”) entails condensation, etc.
The situation is more complex in the case of Hopper’s ()
principles. While being couched in terms of grammatical change,
layering, divergence, and persistence refer essentially to synchronic
states reached by grammaticalized forms. The remaining two princi-
ples, specialization and decategorialization, by contrast, are essentially
diachronic in that they capture processes of grammatical change.
A different perspective surfaces in the approach proposed by Heine
and Kuteva (): All four parameters distinguished, namely context
extension, desemanticization, decategorialization, and erosion, are
intrinsically diachronic in nature in that they focus on changes from
one state of historical development to another. In this respect, this
approach resembles that of Harris and Campbell (: –), even if
the latter is more generally concerned with syntactic change than with
grammaticalization.
In spite of such differences in perspective there is considerable
overlap between the various approaches. However, there are some
substantial mismatches that the reader needs to take into account.
One kind of mismatch concerns cases where different authors use one
and the same term for different purposes or with a different denotation.
For example, Heine and Kuteva () employ the term ‘decategorializa-
tion’ strictly for morphosyntactic phenomena only, whereas Hopper
(: ) extends the term to also include functional-semantic shift
(see also Hopper and Thompson () on the use of this term). Another
example can be seen in the use of the term ‘extension’, which for Harris
and Campbell (: , ) constitutes one of the three basic mechan-
isms of syntactic change, having the effect that a condition on an existing
rule is removed. For Heine and Kuteva (: ), by contrast, ‘exten-
sion’ is essentially a text-pragmatic notion involving the use of a given
linguistic expression in new contexts which invite novel semantic inter-
pretations (see Section .).
The term ‘structural scope’ of Lehmann ( []) is an espe-
cially severe example of mismatch. As noted in Section ., Lehmann’s
‘structural scope’ refers to something fundamentally different than
‘scope’ in linguistics outside the discussion of his parameters of
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CRITERIA, PARAMETERS, AND OTHER VARIABLES
grammaticalization, and must be considered an idiosyncratic if not
misleading labeling. Lehmann’s concept is based on morphosyntactic
relations between linguistic signs, and on an older endocentric constit-
uent structure model whose details are not provided. In contrast, scope
in general linguistics is a logical-semantic concept, it is defined by an
asymmetric (exocentric) relationship between an operator and an
operand, and not bound by, or even related to morphology.³ The result
when applied to grammaticalization is just the reverse: Reduction in the
case of Lehmann’s structural size (“structural scope”; Lehmann
[]: ), vs. extension in the case of scope in the general linguistic
logical sense (e.g. Hengeveld : , : ; Matasović : ;
Roberts : ; Narrog a: ; Nicolle : –; cf. Narrog
for a discussion of both concepts).
A second kind of mismatch involves cases where two authors each use
a different term for what is essentially the same phenomenon. For
example, Hopper’s () term ‘divergence’ captures the same phenom-
enon that Heine and Reh (: –) had described earlier as ‘split.’
A third, and a more common kind of mismatch relates to cases
where one author proposes a terminological distinction not made in
that form by other authors. For example, the parameter of decategor-
ialization of Heine and Kuteva (), whereby a linguistic expression
loses morphosyntactic features characteristic of its categorial status,
includes changes that relate to four of Lehmann’s ( []) para-
meters, namely paradigmatic variability, structural size (mislabeled as
“scope”), bondedness, and syntagmatic variability.
Conversely, there are cases where Heine and Kuteva () propose
a terminological and conceptual discrimination not made by Lehmann
( []). In the framework of the former, a strict distinction
is made between morphosyntactic and phonological phenomena, the
former being covered by the term ‘decategorialization’ and the latter by
‘erosion.’ This distinction is not maintained in this form by Lehmann
( [], whose parameters are designed to be applied to all levels
of grammar. For example, his parameter of bondedness combines
both morphosyntactic and phonological phenomena, and the same
³ Lyons (: ): “By the scope of any operator, whether it is a connective or a
quantifier, is meant that part of the formula which is within its domain of operation; and
this is normally indicated by brackets.”
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCUSSION
applies to his parameter of integrity. Or, he suggests that “phonological
attrition and desemanticization go hand in hand.”
It would seem, however, that there is a need to separate desemanti-
cization and erosion since the two contrast in their diachronic behav-
ior. First, desemanticization marks most of all the early stages of
grammaticalization; erosion, by contrast, takes place as a rule at later
stages of the process. And second, many processes of grammaticaliza-
tion have been documented where there was desemanticization but
apparently no erosion (see Heine b for discussion). For example,
the High German perfect (or past tense) auxiliary haben, as in (b), is
phonologically essentially indistinguishable from the possessive verb
haben ‘have’ in (a), even though the former developed out of the
latter—in other words, in its history of roughly a thousand years the
auxiliary did not undergo erosion (Heine and Kuteva : ); see
Section . for more examples.
() Modern German
a. Paul hat viel Arbeit.
Paul has much work
‘Paul has a lot of work.’
b. Paul hat viel gearbeitet.
Paul has much worked
‘Paul has worked a lot.’
In sum, the two parameters need to be separated, even if their effect is
in an abstract sense the same: Both lead to a loss of linguistic features
or, in Lehmann’s terms, to decrease in integrity.
A second case relates to what Lehmann ( []: –)
describes as paradigmatic variability. This parameter seems to involve
two kinds of change and, hence, is captured by two different parameters
in Heine and Kuteva (: –). On the one hand, it involves effects
of context extension: The more a linguistic form is extended to new
contexts the more its use becomes an obligatory part of the host class to
which it attaches. On the other hand, it also involves decategorializa-
tion, in that the form increasingly loses morphosyntactic freedom or
variability to occur independently of its host class.
A third case relates to Lehmann’s ( []: –) parameter
of bondedness. This parameter can be interpreted in particular as
involving an amalgam of decategorialization and erosion, that is, of
loss of the morphosyntactic status as a word, a clitic, or an affix on the
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CRITERIA, PARAMETERS, AND OTHER VARIABLES
one hand, and loss of phonological substance at the boundary of a form
on the other. It would seem, however, that decategorialization and
erosion need to be separated. For example, Lehmann ( []:
) illustrates his parameter of paradigmatic variability with the
development of the demonstrative ille()/illa() ‘that’ of Latin
or some post-classical variety of Latin to the definite article le()/la
() in French. On the one hand, there was a gradual transition from
word to proclitic. On the other hand, there was loss of phonetic
substance, in that a disyllabic form (ille) was reduced to a monosyllabic
article (le). Before an initial vowel in the following word, syllabicity was
even lost entirely (l’). There is no evidence to suggest that the two
changes coincided in time.
A fourth kind of mismatch occurs when parameters or criteria pro-
posed by one author are ignored by other authors. One example is
provided by Comrie’s () notion ‘interaction with (the rest of) the
grammar’, for which there is no clear equivalent in any of the approaches
looked at in Section . (but see Hopper and Traugott : – for a
detailed discussion of pragmatic factors). Another example can be seen
in the fact that grammaticalization very frequently involves a gradual
category shift from a paradigm with a larger number of members to one
with a smaller number of members, most pronouncedly in the shift from
lexical to grammatical categories. This generalization is covered by
Lehmann’s ( []: –) parameter of paradigmaticity but is
not discussed by other authors reviewed here in the way it may deserve
(for divergent views see the last paragraph of this section and Traugott
a: ). A third example relates to the role played by pragmatics in
grammatical change, which is with one exception not treated in the
approaches sketched above. This exception is Heine and Kuteva
(), where the parameter of (context) extension is devoted to the
interaction between context pragmatics and semantics (see Section .).
A fourth example of parameters or criteria proposed by one author
but ignored by other authors relates to the role of discourse. Most
studies of grammaticalization in the past took sentence grammar as
their frame of analysis, and this fact is reflected in the nature of the
criteria surveyed in Section .. More recent research, especially
research in the tradition of Traugott (a), has been extended to
grammatical phenomena beyond the sentence, focusing on discourse at
large for generalizations on grammatical change. An example of this
line of research can be found in the theory of discourse prominence
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. CONCLUSIONS
proposed by Boye and Harder () and their distinction between
discursively primary and discursively secondary linguistic items (see
Section .). With this distinction they provide an additional parameter
for reconstructing grammaticalization processes.
Finally, there is a fifth kind of ‘mismatch’ which can be seen in the
fact that, unlike other authors, a given author may attribute special
importance to one particular parameter or concept compared to
other parameters or concepts. A case in point can be seen in Lehmann’s
treatment of the concept of paradigm (Lehmann []: ;
see also Diewald , a, b): Not only are three of his six
parameters classified in accordance with this concept. Two of these
parameters even specifically address the concept, namely paradigmati-
city and paradigmatic variability. Note that the role played by para-
digms in language use leading to grammaticalization is not entirely
clear or even controversial, and Himmelmann (: ) concludes
that “[t]here is no evidence for viewing paradigmatization as a neces-
sary factor [of grammaticalization] at all.” Note further that even a
basic distinction such as that between open class and closed class
paradigms is not without problems (Boye and Harder : –).
More research is needed on this issue.
2.3 Conclusions
The regularities of grammaticalization proposed in the frameworks dis-
cussed in Section . do not lay claim to explanatory significance.
Instead, they have various purposes, such as to distinguish grammatical
(or functional) expressions from lexical ones, or to describe typical stages
and phenomena accompanying grammaticalization, or to provide direc-
tives or guidelines for reconstructing grammatical change, or a combi-
nation of these. Despite their diversity, taken together they provide a
solid foundation for understanding what grammaticalization is about.
Newmeyer () claimed that grammaticalization is not a distinct
process. Subsequent research has shown, however, that this claim must
be taken with care (see also Chapter ). To be sure, the term ‘grammat-
icalization’ subumes a wide range grammatical changes, not all of which
turned out to be regular to the extent that earlier researchers had
assumed them to be. Nevertheless, there is by now fairly wide agreement
that the definition of grammaticalization as a process leading from lexical
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
CRITERIA, PARAMETERS, AND OTHER VARIABLES
or less grammatical to more strongly grammatical expressions is sup-
ported by a robust body of cross-linguistic data.
But in spite of all the regularities that exist, it has so far not been
possible to reduce all the manifestations of grammaticalization to one
general formula that would allow identifying instances of the process in
a uniform way. Pragmatic, semantic, morphosyntactic, and phonogical
factors do not all proceed exactly the same way and at the same time
(see, e.g., Heine, Kuteva, and Narrog ). Therefore such factors need
to be analyzed separately, as we will do in Chapter .
Further reading
For questions relating to terminological issues, such as the use of the terms
‘principle,’ ‘parameter,’ ‘criterion,’ or ‘mechanism,’ see Hopper (), Com-
rie (), Heine and Kuteva (: –), Harris and Campbell (),
and Lehmann ( []).
The question of whether the development of meaning and form in gram-
maticalization coincides or follows one another is discussed especially in
Bisang (, ), Ansaldo and Lim (), Narrog (c), and Heine
(b).
Discussion points
() Do you know of other examples from the study of language where the
same term refers to different phenomena depending on the author or the
framework, or where the same term is used more narrowly or more
broadly depending on author or framework? What are the consequences
for studying language and researching in linguistics?
() In your opinion, which of the parameters and criteria discussed in this
chapter are purely synchronic, which are purely diachronic, and which do
you think would lend themselves both to a synchronic and a diachronic
perspective?
() Which perspective do you believe to be more important for the way you
want to study language? Give reasons for your evaluation.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
3
Four parameters of
grammaticalization
3.1 Context extension
3.2 Desemanticization
3.3 Decategorialization
3.4 Erosion
3.5 Conclusions
Discussion points
The main goal of the present chapter is to propose the following four
parameters for identifying and reconstructing grammaticalization:
(a) (Context) extension;
(b) Desemanticization;
(c) Decategorialization;
(d) Erosion.
These parameters were already mentioned in Section .. In the sec-
tions to follow they will be dealt with in greater detail (see also Heine
and Kuteva : –). The reasons for opting for these parameters
are the following. First, unlike most other criteria or parameters dealt
with in Chapter , all four parameters distinguished are intrinsically
diachronic in nature, focusing on grammatical change. They therefore
seem to be best suited in a book that is concerned with grammatical
change. Second, the relevance of the four parameters can be tested by
means of standard techniques of linguistic analysis. Third, most of the
phenomena addressed in alternative frameworks can be reduced to
effects of the four parameters.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
And fourth, the parameters take care of most manifestations of
language use and language structure, ranging from semantics to pho-
nology and from pragmatics to morphosyntax. In fact, the four para-
meters are each devoted to a different component of language structure
and language use: Context extension is pragmatic in nature, deseman-
ticization relates to semantics, decategorialization to morphosyntax,
and erosion to phonology and phonetics.
Three of the parameters involve loss of grammatical features. But, as
has been established abundantly in relevant studies (see especially
Hopper and Traugott ), grammaticalization cannot be reduced
to decrease of structural richness, attrition, or ‘degeneration’: In the
same way as there are losses there are also gains (cf. the loss-and-gain
model; Heine et al. : ), captured most of all by the context
extension parameter: Linguistic items and constructions undergoing
grammaticalization lose in semantic, morphosyntactic, and phonetic
substance, but they also gain in means in new contexts for the “inter-
action with (the rest of ) the grammar”—as Comrie (: ) puts it,
processing discourse in novel ways and providing tools for restructur-
ing grammar.
There is good evidence to suggest that the four parameters apply in
the order listed (cf. Bisang et al. ). This implies in particular that
meaning change (desemanticization) precedes form a change (decate-
gorialization) in grammaticalization. This is in accordance with the
hypothesis proposed by Narrog (c) that formal changes in gram-
maticalization are secondary to functional changes and that formal
changes may or may not occur, and also in accordance with the
“meaning-first hypothesis” of Heine (b). In an alternative hypoth-
esis, called the “parallel reduction hypothesis,” meaning and form
change proceed in parallel (e.g. Bybee et al. : ), that is, dese-
manticization and decategorialization occur simultaneously. However,
the results of Bisang et al.’s () research on a cross-linguistic
database of grammaticalization lead to the conclusion that “meaning/
form covariation is rarely observed” (Bisang et al. : ). Instead,
functional change often precedes, and presumably conditions formal
change, and phonetic reduction (erosion) is the last to be involved.
However, the actual relationship between semantic and formal changes
requires a more fine-grained analysis than the two overarching hypoth-
eses suggest.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONTEXT EXTENSION
3.1 Context extension
One of the assumptions made in this book is that grammaticalization
rests on changes in the use of discourse options and norms, and one
major mechanism serving this purpose can be seen in the manipulation
of contexts for specific communicative goals (Ariel ; see Section
..). Context extension, on the one hand, covers essentially the defini-
tion of grammaticalization as proposed by Hopper and Traugott (:
), cited in example () of the introductory Chapter . On the other
hand, it also covers to a large extent what Hopper and Traugott (:
) treat as analogy or rule generalization. It means, first, that the use of a
linguistic item is extended to novel contexts, leading to the rise of new
meanings, and eventually also of new functional categories.¹ And second,
context extension also entails increase in the frequency of use: With each
new context, the probability increases that the item will be used more
frequently. Thus, relative frequency of use is, at least to a considerable
extent, derivative of context extension.
Context extension is also the most complex of all the parameters, for
the following reasons: First, it has a sociolinguistic, a text-pragmatic, and a
semantic component. The sociolinguistic component refers to the fact
that grammaticalization starts with innovation (or activation) as an
individual act, whereby some speaker (or a small group of speakers)
proposes a new use for an existing form or construction, which is
subsequently adopted by other speakers, ideally diffusing throughout an
entire speech community (propagation; see, e.g., Croft : –). The
text-pragmatic component involves the extension from a usual context to
a new context or set of contexts, and the gradual spread to more general
contexts. The semantic component finally leads from an existing meaning
to another meaning that is evoked or supported by the new context. Thus,
text-pragmatic and semantic extension are complementary sides of one
and the same general process characterizing context extension.
Second, the term ‘extension’ is applied in a variety of different ways
and in different frameworks. Harris and Campbell () use it as a
¹ ‘Context extension’ was referred to in earlier studies simply as ‘extension’ or ‘context
generalization’ (e.g. Heine and Kuteva : –; Heine a). The reason for adding
‘context’ is that the term ‘extension’ is also used in a number of different ways (e.g. Harris
and Campbell ), some of which are not entirely in accordance with the use in this book.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
technical term for one of their three basic mechanisms of syntactic
change (see Section .). In their theory, extension has the effect that
a condition on an existing rule is removed. While this use differs in
a number of respects from the way it is applied here, there is considerable
overlap, as is suggested by the examples given by Harris and Campbell.
Some of those would also satisfy our criteria of context extension.
However, there are also differences. Context extension is framed here
in terms of semantic-pragmatic notions rather than of syntactic rules.
Also, context extension entails some change in meaning, however minute
that change may be, while meaning plays a less central role in the theory
of Harris and Campbell ().
In Section .. a sketch of the model that will be employed in the
chapters to follow is provided while Section .. illustrates the model
with examples.
3.1.1 The context extension model
A number of approaches have been used to deal with phenomena
relating to context extension (see, e.g., Bybee et al. ; Diewald ;
Traugott and Dasher : –). In the present book, the four-stage
model of context-induced reinterpretation depicted in Table . is
adopted to describe salient characteristics of context extension. Note
that the model is not meant to explain what grammaticalization is about.
It merely captures specific effects of the speakers’ motivations and goals
(see Chapter ). Note further that grammaticalization is a gradual proc-
ess and the four stages distinguished here merely represent prototypical
instances along a chain of grammaticalization.
Table . suggests that the transition from the lexical or less gramma-
ticalized source meaning of stage I to the more grammaticalized target
meaning of stage IV does not proceed straight from one to the other;
rather, it involves two intermediate stages, namely stages II and III.
The initial stage I relates to the use of a given linguistic form and the
construction of which it is a part prior to grammaticalization, and
provides the input for grammaticalization. Typically, the input form
is a lexical expression but this is not a requirement.
The bridging context II involves the extension of that form to a new
context which invites a new meaning (the target meaning) co-existing
with the old meaning (the source meaning). It largely corresponds to
the ‘bridging context’ of Evans and Wilkins (: ) in their analysis of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONTEXT EXTENSION
Table 3.1 The context extension model (Heine )
Stage Context Resulting meaning
I Initial stage Unconstrained Source meaning
II Bridging context There is a new context triggering a new meaning Target meaning
foregrounded
III Switch context There is a new context which is incompatible Source meaning
with the source meaning backgrounded
IV Conventionalization The target meaning no longer needs to be Target meaning
supported by the context that gave rise to it; it only
may now be used in new contexts
semantic change, or to what Diewald () in her analysis of German
modals calls the ‘critical context.’² Bridging contexts are based on what in
the literature since Grice () has been described in terms of invited
inferences (Geis and Zwicky ) or conversational implicatures, or
what Miller and Johnson-Laird () call ‘construal rules’:
Given the schema of a verb and a particular context of use, a construal rule modifies
the schema to provide an interpretation of the word appropriate to that context.
(Miller and Johnson-Laird : )
Inferences, conversational implicatures or construal rules may already
be there occasionally at stage I. But it is only now that they become a
feature that is regularly associated with some specific set of contexts.
Bridging contexts have the properties listed in ().
() Properties of bridging contexts
a. They trigger a regular inferential mechanism to the effect that,
rather than the source meaning, another meaning, the target
meaning, is invited by the context, offering an equally plausi-
ble, or even a more plausible interpretation of the utterance.
b. The target meaning is the one most likely to be inferred but it is
still cancelable (see Grice ), that is, an interpretation in
terms of the source meaning is always possible.
² A critical context is characterized by multiple structural and semantic ambiguity,
inviting different interpretations, including the target meaning (Diewald ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
c. Within a given construction, some linguistic form may be associated
with a number of different bridging contexts but, as a rule, most of
them do not give rise to conventional grammatical meanings.
The switch context III refers to the use of a linguistic form in a new
context which highlights a new meaning (the target meaning) and is
incompatible with the earlier meaning of that form (the source mean-
ing). It corresponds to some extent to what Diewald () calls
‘isolating contexts,’ where the target meaning “is isolated as a separate
meaning from the older, more lexical meaning.” But isolating contexts
mark “the completion of the grammaticalization process,” hence they
are also suggestive of the conventionalization stage discussed below.
Switch contexts have the properties listed in ().
() Properties of switch contexts
a. They are incompatible, or in conflict with some salient prop-
erty of the source meaning of stage I.
b. Hence, an interpretation in terms of the source meaning is
implausible.
c. The target meaning now provides the only reasonable
interpretation.
d. Unlike the conventionalized target meaning of stage IV, the
target meaning appearing in switch contexts must be sup-
ported by a specific context or cluster of contexts.
The conventionalization stage IV refers to the use of a linguistic form
which arose via context extension and no longer needs to be supported
by the context which gave rise to it.
This stage marks a new quality in grammaticalization. At this stage,
speakers may no longer conceive the target meaning as being related to
the source meaning. The target meaning no longer needs to be sup-
ported by context, and it can be used in new contexts other than the
bridging and switch contexts. It now is the ‘normal’ or ‘inherent’
meaning of the form (cf. Hopper and Traugott : –). This
means that the form and the construction to which it belongs now
have the properties in () or part thereof.
() Properties of conventionalization commonly observed within a
grammaticalizing construction
a. The target meaning is no longer context-dependent.
b. The boundaries between parts of the construction are redefined.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONTEXT EXTENSION
c. The dependency relations between the parts change.
d. The target form is decategorialized.
e. The target form undergoes erosion.
The transition from stage III to stage IV can be conceived as a process
of ‘automation’ (De Smet : ), whereby a new construction
gradually evolves which, through repeated use, drifts off from its earlier
uses and becomes more and more firmly represented as a self-
consistent ‘chunk’ of discourse.
3.1.2 Discussion
A few examples may illustrate the significance of the context extension
model. It is a commonplace that a volition schema of the kind [X wants
Y] involving verbs for ‘want’ forms one of the three main sources for
developing future tense categories (Bybee et al. ; Bybee et al. ;
Kuteva et al. , > ). But this schema may also give rise
to a different grammatical function, namely that of a proximative aspect.
The function of this aspect is to define a temporal phase immediately
preceding the initial boundary of the situation described by the main
verb. Proximative aspects are commonly translated by means of ‘be about
to,’ ‘be on the verge of,’ and the like (see Heine c; Romaine ).
The examples in () are taken from the East African language Swahili
(see Heine for more details). (a) illustrates stage I, showing the
lexical use of the volition verb -taka ‘want.’ In (b), the construction is
extended to take ku-fa ‘to die’ as an infinitival complement. Since dying is
a process that one reasonably does not want to happen, (b) invites an
inference to the effect that the subject referent may not ‘want’ to die
but rather is ‘is about to’ die. Thus, complement verbs expressing an
undesirable event, such as ‘die,’ ‘fall down,’ or ‘be hurt,’ evoke a new
meaning, namely that of a proximative aspect marker (‘be about to,’
‘be on the verge of ’). At the bridging stage II, the new meaning co-exists
side by side with the earlier meaning of volition (‘want’); hence (b) is
ambiguous.
At stage III, a situation arises in which the volition verb is extended
to inanimate subject referents, such as mti ‘tree’ in (c), that—except
for metaphorical or culture-specific interpretations—are not conceived
as being capable of willful actions. In such contexts, the source meaning
of volition does not make sense. It is backgrounded or suppressed and
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
the meaning of a proximative aspect provides the only reasonable
interpretation.
() Swahili (Bantu, Niger-Congo)
a. Baba y-angu a-na-taka ku-ja. I
father -my --want -come
‘My father wants to come.’
b. Baba y-angu a-na-taka ku-fa. II
father -my --want -die
(i) ‘My father wants to die.’
(ii) ‘My father is about to die.’
c. M-ti h-uu u-na-taka ku-anguka. III
-tree :- --want -fall
‘This tree is about to fall.’
Grammaticalization processes of this kind from verb of volition to prox-
imative marker have been described in some detail (Heine , c;
Kuteva a; Romaine ; Kuteva et al. , > ).
Swahili has not proceeded beyond Stage III. This means, on the one hand,
that examples instantiating Stage III, such as (c), tend to be understood
by speakers as context-dependent variants of the lexical use of the volition
verb -taka ‘want,’ rather than as presenting a new meaning. On the other
hand, it also means that the Swahili proximative has so far not been
recognized by grammarians as a distinct grammatical category. We are
not aware of any Swahili grammar that has taken notice of it.
The situation is different in languages that have proceeded further to
the final Stage IV of conventionalization. In such languages, the voli-
tion verb (‘want’) has turned unambiguously into a proximative marker
(‘be about to,’ ‘be on the verge of ’) and the construction has ‘emanci-
pated’ itself to be recognized as a new grammatical category conform-
ing to the properties listed in (). There are a number of languages of
this kind (see Heine c for examples). Chamus is a dialect of the
Eastern Nilotic language Maa (Maasai) spoken in Southern Kenya.³
Like Swahili, it is an East African language but the two are presumably
genetically unrelated. Chamus appears to have gone through the same
process as Swahili, namely the grammaticalization of its ‘want’ verb
-yyéú to a proximative marker. The examples in () illustrate the three
³ Example () is taken from Heine (), which contains a more detailed description.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONTEXT EXTENSION
stages, corresponding to the Swahili examples in () with examples of
the initial stage (a), the bridging stage (b), and the switch stage (c).
The main structural difference between the two languages is that the
verbal complement of the volition verb is encoded as an infinitival form
in Swahili, whereas in Chamus it is encoded as a clause constructed in
the narrative tense.⁴
() Chamus (Maa dialect, Eastern Nilotic; Heine : –)
a. k-á-yyéú nánʊ n-a-ló n-kaŋ. I
k--want - --go -home.
‘I want to go home.’
b. k-é-yyéú l-páyyan n-é-rriá. II
k--want -elder. --fall
‘The old man nearly fell.’ or ‘The old man wanted to fall.’
c. k-é-yyéú l-cáni n-é-uróri. III
k--want -tree. --fall
‘The tree nearly fell.’
But Chamus speakers have even gone one step further: The verb -yyéú
in its third person singular form has grammaticalized to an invariable
aspect particle (k)eyyéú whose meaning can be paraphrased as ‘be about
to, almost,’ and which may no longer be inflected for person or aspect,
as example () shows.
() Chamus (Maa dialect, Eastern Nilotic; Heine : )
a. (k)eyyéú a-ók nánʊ kʊlɛ́. IV
-drink . milk.
‘I was about to drink milk.’
The old construction illustrated in () was replaced by a new construc-
tion, which exhibits all the hallmarks of conventionalization listed in ():
(a) The boundaries between the parts of the construction are rede-
fined in that the particle (k)eyyéú, originally the main verb, is
detached from the rest of the sentence.
⁴ A further difference relates to the fact that the proximative examples in Swahili are
constructed in the present tense but in the past tense in Chamus. This is not a trivial
difference, considering that past tense uses of the proximative are commonly called avertive
rather than proximative (Kuteva a).
⁵ The function of initial is k- unclear.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
(b) No longer being the main verb, it is the former complement of
-yyéú ‘want’ that is now the new main verb (a-ók ‘I drink’), and
rather in the narrative, the new verb is constructed in the
unmarked imperfective form. Accordingly, the subject (nánʊ ‘I’)
must follow rather than precede the new main verb since Chamus
is a verb-initial language.
(c) The erstwhile verb form k-é-yyéú (k--want) has undergone
complete internal decategorialization in that all internal morpheme
boundaries have disappeared, the erstwhile verb form (k)eyyéú
being an unanalyzable particle at the left periphery of the sentence.
(d) And finally, there also is optional erosion in that the initial
consonant k of the particle can be omitted.
The following example illustrates the context extension model with a
case of clause combining, namely with the English conjunction since
(Traugott and König : –). The evolution of since from tem-
poral to causal conjunction is sketched in ().
() The evolution of English since (based on Traugott and König
: –)
Constructed examples Meaning Stage
a. I have done quite a bit of writing since Temporal I
we last met.
b. Since Susan left him, John has been very Temporal, II
miserable. causal
c. Since you are not coming with me, I will Causal III
have to go alone.
Example (a) is a clear instance of the initial stage I, where the sentence
can be interpreted only with reference to a temporal relation between
the two clauses. In contexts like that in (b), a causal meaning is
triggered in addition, resulting in a bridging stage II situation. And
finally, in certain contexts where one clause refers to a non-past event
or to a state, the temporal meaning may no longer make sense and is
backgrounded. The result is that a causal interpretation provides the
only reasonable interpretation, as in (c), which appears to be sugges-
tive of the switch stage III.
Stage I is commonly found in Old English texts (ca. –). But
already in Old English, since occurred in certain contexts suggestive of
the bridging stage II, where a causal meaning can be foregrounded,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONTEXT EXTENSION
offering a ‘causal implicature’ (Traugott and König : ). Clear
instances of the switch stage III arose in the fifteenth century, where a
temporal interpretation is backgrounded or even blocked.
But causal since has not attained the final Stage IV of conventiona-
lization.⁶ This is suggested by the fact that it does not conform to the
properties listed in () of Section ... Thus, its use has remained
context-dependent, neither the boundaries nor the dependency rela-
tions within the construction to which it belongs are redefined, and it
shows neither decategorialization nor erosion.
To conclude, the conventionalization stage IV marks a gradual, but
at the same time also a fairly dramatic change in grammaticalization.
Note, however, that stage IV does not necessarily mark the end of
grammaticalization. Once the target meaning has been conventiona-
lized, it is open for further rounds of context extension. This may be
demonstrated with the Present-Day English suffix -hood (Hopper and
Traugott : –; Traugott and Trousdale : –). The suffix
is traced back to the Old English noun hād, which had a wide range of
meanings, including ‘status, office, rank, character, nature, state.’ The
noun formed the head of a modifying noun–noun compound where
the modifier typically referred to human beings and the head (hād) was
usually interpreted as having the restricted meaning ‘rank, status of the
person denoted by the modifier.’ Conventionalization of -hood appears
to have been concluded already in Old English with the reanalysis of the
noun as a derivational suffix. But subsequently, a second round of
context extension took place in Middle English (–) when
-hood experienced further extension to new environments, giving rise
uses as in falsehood (see Hopper and Traugott : for details).
3.1.3 On the differential behavior of stages
Once a grammaticalization process has taken place it is normally not
reversible. The grammaticalized item may fall into disuse and may
subsequently be lost, but the development leading to the rise of a gram-
matical category is unlikely to recede. This generalization, however, is in
need of qualification: It seems to apply only to grammatical forms having
reached the conventionalization stage IV but not necessarily those at
⁶ Note that our use of the term ‘conventionalization’ is not the same as that of Hopper
and Traugott (: ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
Stages II or III. At these stages ‘retraction’ can take place, which is not a
development in the reverse direction of grammaticalization (‘degramma-
ticalization’) but a decline of grammaticalized uses, leaving behind the
original lexical or less grammaticalized uses (cf. Section .).
To give an example, Kashubian is a West Slavic language spoken
in Pomerelia. Up until , Pomerelia was part of Germany, and
German was the main second language of Kashubians. After World
War II, Pomerelia was included in Poland and now Polish became the
main second language of Kashubians. German has fully conventialized
definite articles while neither Kashubian nor Polish have. But under the
influence of German, Kashubians grammaticalized their demonstrative
pronouns ten/ta/to and nen/na/no ‘that’ into stage II and III definite
articles, that is, weakly grammaticalized articles, cf. () (for contact-
induced grammaticalization, see Chapter ). But now being exposed to
Polish rather than German speakers, Kashubian speakers have largely
given up the use of their demonstratives as articles. Thus, the function
of an associative article expressed by ta in () is nowadays normally
unmarked, as it is in Polish.
() Kashubian (West Slavic; late nineteenth–twentieth-century text;
Nomachi )⁷
[ . . . ] Ta bieda bëła wiôlgô.
that.. poverty.. be.. big..
‘The poverty was enormous.’
() Kashubian (West Slavic; Present-day text; Nomachi : ;
Ø = zero)
[...] Ø Biéda bëła wiôlgô.
poverty.. be.. big..
‘The poverty was enormous.’
The data provided by Nomachi () in fact suggest that, given an
appropriate pragmatic situation, the effects of a grammaticalization
process can be undone as long as that process has not reached the final
stage IV of conventionalization, but more research is needed on this
issue.
⁷ The sentences in this and the next example are excerpts from a larger text which is
presented in Nomachi ().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DESEMANTICIZATION
3.2 Desemanticization
The term desemanticization is defined as the loss of semantic properties
that a linguistic expression undergoes in the course of grammaticaliza-
tion. It is an immediate consequence of context extension from the
bridging stage II onward.
Desemanticization is described by Lehmann ( []: table .)
as one of the manifestations of integrity leading to attrition, whereby a
bundle of semantic features associated with a linguistic expression is
reduced to fewer semantic features. The term refers to what is also
called ‘semantic bleaching’ (see Section .).
Hopper and Traugott () suggest that,
[t]here is no doubt that, over time, meanings tend to become weakened during the process
of grammaticalization. Nevertheless, all the evidence for early stages is that initially there is
a redistribution or shift, not a loss, of meaning. (Hopper and Traugott : )
‘Weakening of meaning’ apparently refers to desemanticization, and
‘redistribution or shift of meaning’ was covered in Section . within
the parameter of context extension, which also relates to what Hopper
and Traugott (, ch. ) call ‘pragmatic enrichment’ and ‘pragmatic
strengthening.’
On the basis of the context extension model in Table ., the dese-
manticization of a grammaticalizing form can be described as involving
several stages:
(a) Desemanticization arises at the bridging stage II: An expression is
placed in a context inviting a new interpretation of the meaning of that
expression. This new interpretation, called the target meaning, differs
from the earlier meaning (the source meaning) in that semantic prop-
erties of the source meaning which do not make sense in that context
are backgrounded. Thus, compared to the source meaning, the target
meaning arising in that context lacks semantic properties—that is, it is
desemanticized. However, the non-desemanticized source meaning is
still the normal one in other contexts.
For example, in a number of verb serializing languages of West Africa
or of !Xun in Southwestern Africa, the use of verbs for ‘descend’ or ‘fall
down’ was extended to contexts where the semantic property of physical
motion made little sense and was backgrounded, the resulting meaning
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
being ‘down’ in such contexts. Thus, in the W dialect of !Xun, spoken in
Northern Namibia, the verb n!!haò̀̀ -ā ‘descend’ (!! is a retroflex affricate
click), illustrated in (a), can be used in a context where it forms the
second verb in a series of two verbs. In this context, the meaning of the
action verb of physical motion is backgrounded and n!!haò̀̀ -ā is dese-
manticized to the locative target meaning ‘down’; cf. (b).
() !Xun (W dialect, Kx’a, “Khoisan”; own data)
a. tc’ámmà má n!!hà ò̀ -ā kx’à.
bird descend:- earth
‘The bird lands on the ground.’
b. mí m-é gù n!!hà ò̀ kù.
- take: down. milk
‘I took the milk down.’
(b) With the backgrounding of the source meaning in switch contexts
(stage III), the desemanticized target meaning is virtually the only
meaning surviving. Such a situation can also be illustrated with another
example from !Xun. The verb ú ‘go’ expresses physical motion, as in
(a). But when it is placed in a context where it is the main verb and its
complement is a stative verb, as gǀx’úí ‘be tired’ is in (b), then a new
meaning is foregrounded, namely that of ‘completed change of state’,
that is, an inchoative function, as in (c).⁸ In this context, the meaning
of physical motion is incompatible with that of stative verbs, and (c)
is suggestive of a switch stage III—that is, ú in (c) can no longer be
interpreted with reference to its lexical meaning.
() !Xun (W dialect, Kx’a, “Khoisan”; own data)
a. Cālò má ú kē gè’è.
Calo go sing
‘Calo sang while going.’
b. Cālò má gǀx’úí.
Calo be.tired
‘Calo is tired.’
⁸ Since ú ‘go’ is an intransitive verb it requires the transitivizing suffix -ā (TRA) to take a
complement. The grammaticalization from verbs for ‘go’ to change-of-state markers is
attested in a number of languages, see Kuteva et al. (, > --); cf.
English to go mad.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DESEMANTICIZATION
c. Cālò má ú-á gǀx’úí
Calo go- be.tired
‘Calo has become tired.’
Nevertheless, the lexical meaning of ú ‘go’ is retained in other contexts,
such as that of (a).
(c) At the conventionalization stage IV, the desemanticized target form
is generalized, no longer in need of being supported by context.
The !Xun language also provides an example to illustrate this stage.
The verb ú ‘go’, which we saw already in the preceding example, has also
given rise to a tense marker: Jointly with the transitivizing suffix -ā (),
taking the following verb as its complement, it was grammaticalized to a
future marker (*ú-ā [go-] > óá; Heine et al. : ); thus, the
meaning of the example in () can be reconstructed diachronically as
‘he goes to come later.’ The future marker óá in () is strongly gram-
maticalized: It underwent internal decategorialization by losing its inter-
nal morpheme boundary (see Section ., ()), and it changed its vowel
u to o. Note that there is an optional vowel lowering rule whereby u is
lowered to o when there is a non-high vowel in the following syllable.
Thus, the construction illustrated in () is fully conventionalized, to
the extent that it shows little resemblance to its lexical source, recon-
structed as *ú-ā.⁹
() !Xun (W dialect, Kx’a, “Khoisan”; Heine et al. , ())
hà ̀ má nǀǀà ǹ óá gǀè.
later come
‘He’ll come later.’
Desemanticization does not necessarily mean that the entire meaning
of a lexical source disappears; rather, there may be some semantic
features of the source form that survive in the grammaticalized target
form (cf. Hopper’s () principle of persistence, Section .). But as
the example in () suggests, there may be little, if anything, left of the
original source meaning once that form is fully conventionalized.
Furthermore, the degree to which desemanticization takes place
correlates with the degree that context extension has advanced: The
⁹ The process sketched is strikingly similar to that of the English be-going-to future tense,
both of which can be hypothesized to be instances of the grammaticalization pathway
> (Kuteva et al. ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
more contexts a given form acquires, the more properties of its source
meaning will be lost. For example, the !Xun verb ú ‘go’ is largely
restricted to animate subject referents. The future tense marker óá, by
contrast, does not show any context restrictions on subject selection.
Desemanticization can be the result of conceptual processes that have
been described variously as involving invited inferencing, subjectification
(Traugott ; Hopper and Traugott ), metonymy and/or meta-
phorical processes (Claudi and Heine ; Heine et al. ; Stolz
a; Heine a, b). For example, a paradigm case of grammati-
calization involves a process whereby body part terms (‘back,’ ‘head,’ etc.)
are reinterpreted metaphorically as locative concepts (‘behind,’ ‘on top of,’
‘in front of’) in specific contexts, and this may eventually result in the rise
of adpositions, cf. English in back of, in front of. Via metaphorical transfer,
concepts of the domain of physical objects, such as body parts, are used as
vehicles to express concepts of the domain of spatial orientation.
Once this happens, there will be desemanticization in the target
forms: The lexical meaning of the body part terms is gradually lost in
favor of locative meaning. In a similar fashion, when action verbs, such
as English keep, use, or go (to), were reinterpreted in specific contexts as
expressing tense or aspect functions (kept V-ing, used to, be going to; see
Section ..), this can be understood to have involved metaphorical
processes whereby a concept of the domain of physical actions was
transferred to the more abstract domain of temporal and aspectual
relations.¹⁰ Once again, this led to the gradual desemanticization of the
lexical meaning in the target forms, in that the meaning of the action
verbs was suppressed.
But desemanticization is in no way restricted to processes leading
from lexical to grammatical meaning (primary grammaticalization).
Instead, it can be found in much the same way in the grammaticaliza-
tion from grammatical to even more grammatical meanings (secondary
grammaticalization). A few examples may suffice to illustrate these
kinds of changes. The examples relate to markers of tense or aspect
on the one hand, and case markers on the other.
A fairly common process is that of future tense markers grammaticaliz-
ing further into markers of epistemic necessity or probability. This process
¹⁰ This is a simplified rendering of the process. See our example of German drohen in
Section .., which gives a more detailed description of the development from lexical verb
to auxiliary.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DESEMANTICIZATION
has happened with the English future marker will (e.g. Jane will visit me
tomorrow), which was extended to contexts where a future meaning made
no sense and gave way to an interpretation in terms of probability (That
will be Jane (said on hearing the doorbell); see Kuteva et al. , >
-). Desemanticization in this case had the effect that the future
meaning of prediction was lost in favor of a modal function. () is an
example of the Swahili future tense marker -ta-, which is suggestive of a
process similar to that of English will: Whereas -ta in () expresses future
tense, the context in () rules out an interpretation in terms of future tense
in favor of one in terms of the modal concept of probability.
() Swahili (Bantu, Niger-Congo; own data)
a. Kipchoge a-ta-ku-ja.
Kipchoge ---come
‘Kipchoge will come.’
b. A-ta-ku-wa nyumba-ni sasa.
---be house- now
‘He will be at home by now.’
Another, presumably equally common process is that from markers
for perfect aspect to past tense markers that can be found, for example,
in Modern Colloquial German (Bybee et al. : ; Thieroff ; see
Kuteva et al. , > ). Perfects (also called anteriors)
signal a past action that is relevant to the current moment, while past
tenses signal only a past action. In this case, desemanticization led to
the loss of the feature ‘current relevance,’ while the meaning of past
action was retained (Bybee et al. : ).
An example involving case markers is provided by comitative
adpositions or affixes (‘(together) with’) grammaticalizing into instru-
mental markers (‘with, by means of’) (Luraghi ; Heine and Kuteva
: –; Narrog , Kuteva et al. , >
). For example, the Latin preposition cum was originally
a comitative marker. The contexts in which it was used gradually
extended, and in Classical Latin marginal uses of cum as an instrumen-
tal marker emerge. They become fairly common in Late Latin, with
clear attestations of switch contexts since the fourth century (Michela
Cennamo, p.c.). A more recent case is provided by the comitative
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
preposition dauve ‘(together) with’ in the Norman French dialect
Guernésiais of the Channel Island Guernsey.¹¹ Its use expanded to
also express an instrumental function, eventually ousting the earlier
instrumental preposition atou ‘with, by means of’ (Jones ).
The process from comitative to instrumental function has been
described as involving a metaphorical transfer whereby an instrument
is conceived as a companion. At the same time, its effect is also one of
desemanticization in that in specific contexts the comitative meaning
‘together with’ declines and is gradually lost in favor of an instrumental
interpretation.
3.3 Decategorialization
The term decategorialization refers to the loss of morphological and
syntactic features that a linguistic item experiences in the course of
grammaticalization.¹²
For some scholars, decategorialization constitutes the most essential
parameter to reconstruct grammaticalization processes. Thus, all three
criteria proposed by Comrie (), and all three mechanisms of Harris
and Campbell (: –; ), as far as they relate to grammaticali-
zation, have to do with decategorialization, and in Lehmann’s (
[]) framework, five of the six parameters proposed are devoted to
aspects of decategorialization.
Once an item has been desemanticized, for example from lexical to
grammatical meaning, it tends to lose morphological and syntactic
features characterizing its earlier use that are no longer relevant to its
new use. In this way, a number of English verbs were decategorialized
in their gerundival form (-ing) and assumed prepositional functions,
for example, barring, concerning, considering, etc. Consequently, they
lost most of their verbal features, such as inflection for tense and
aspect, ability to take auxiliaries, etc. (see König and Kortmann ;
Kortmann and König ).
¹¹ Guernésiais (or Guernsey) was spoken on the island of Guernsey of the Channel
Islands archipelago for more than a thousand years but is now moribund: After World War
II, when many island inhabitants who had been evacuated to England during the war
returned back home, English gradually began to replace this Norman dialect, a process that
appears to be ongoing (Ramisch ; Jones : ).
¹² This definition is slightly different from that of Hopper (), but see also uses of the
term in Hopper and Thompson ().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DECATEGORIALIZATION
Instances of decategorialization thus can be identified with reference
to a set of changes that we will discuss in the remainder of this section.
First, decategorialization may be internal or external, even if the bound-
ary between the two is fluid and each of the two has also implications
for the other.
Internal decategorialization refers to the changes listed in ().
() Salient changes associated with internal decategorialization
a. Loss of the ability to be inflected and to take on derivational
morphology.
b. Loss of the ability to take modifiers.
c. Loss of internal morphological boundaries.
d. Morphosyntactic shift towards a paradigm with a smaller set
of members.
Not all of the changes in () can be observed in every given case of
internal decategorialization, for the following reasons. First, some forms
undergoing grammaticalization take neither inflectional nor derivational
affixes or clitics. This applies, for example, to adverbs in many languages,
and it applies more generally to the isolating languages of West Arica or
Mainland Southeast Asia, which dispose of little or no inflectional or
derivational morphology. Obviously, in such cases, (a) and (b) do
not apply. And second, the changes listed in () may but need not all
apply in a given case of grammaticalization (see below).
In accordance with (a) and (b), nouns undergoing decategor-
ialization tend to lose morphological distinctions of number, gender,
case, etc., and to take modifiers or determiners, while verbs may no
longer be inflected or derived, or take adverbial modifiers.
For example, German nouns like Kraft ‘strength, power’ or Zeit
‘time’ gave rise to the prepositions kraft ‘in virtue of, by dint of’ and
zeit ‘throughout,’ respectively. But on the way to developing into pre-
positions, most of the features were lost, as in (): kraft and zeit are
invariable prepositions that can neither be inflected nor derived, nor
can they be modified or determined (see Kortmann and König for
more examples from European languages).
In a similar fashion, the Swahili noun m-paka ‘boundary, border’ was
grammaticalized to the preposition and subordinating conjunction
mpaka ‘until’ and in the course of this process it lost essentially all
the features it had as a noun. Thus, whereas the noun in (a) can be
inflected for number (cf. mi-paka ‘boundaries, borders’) and triggers
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
noun class agreement, the conjunction mpaka in (b) lacks all these
features, being an invariable particle.
() Swahili (Bantu, Niger-Congo; own data)
a. M-paka w-a Kenya u-po karibu.
-boundary - Kenya - near
‘The border of Kenya is nearby.’
b. A-li-subiri kutoka asubuhi mpaka usiku.
--be.patient from morning until evening
‘He waited from morning till evening.’
In a similar fashion, Swahili ku-toka (-come.from) ‘to come from’ is
a verb that is inflected for person, tense, aspect, modality, and negation,
it can take a range of derivational suffixes, and it can be modified by
adverbs. But in its grammaticalized form as a locative and temporal
preposition (kutoka ‘from’) it is an invariable, decategorialized item
that lacks all these features, as (b) shows. And much the same applies
to English participial verb forms developing into prepositions, such as
during, pending, concerning, or considering (see König and Kortmann
; Kortmann and König for details).
When a demonstrative develops into a clause subordinator, as has
happened in many languages of the world (see Heine and Kuteva :
–), it may no longer be inflected. For example, the English demon-
strative that is inflected for number, having those as its plural form. In its
grammaticalized form as a complementizer or relative clause marker,
however, it is no longer inflected (The books that/*those I know).
(c) is commonly referred to as univerbation, that is, a process
whereby a collocation of two or more words or morphemes loses its
internal morphological boundaries and turns into a new, invariable
word (cf. Lehmann []: ). This was the case, for example,
when the English collocation in front of turned into an unanalyzable
preposition, or the Latin phrase qua re ‘for which reason’ via Middle
French quar/quer to the connective car ‘because, for’ of Present-Day
French. And this also applies to the Latin expression of obligation in
(a) turning into a future marker in (b), also involving erosion
(Section .) in addition to decategorialization.¹³
¹³ See also Section ., (e) for phonetic boundary loss.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DECATEGORIALIZATION
() Latin (cf. Hopper and Traugott : –)
a. canta-re habe-o
sing- have-
‘I have to sing’
b. canta-b-o
sing--
‘I will sing’
(d) has received detailed treatment in Lehmann’s discussion of
paradigmaticity (Lehmann []: ; see also Diewald ,
a): Grammaticalization leads as a rule from items belonging to a
large morphosyntactic paradigm to a paradigm having only a restricted
set of members. As pointed out by some authors, however, the notion
of paradigmaticity is not without problems (e.g. Himmelmann :
; Boye and Harder : –); nevertheless, (d) provides a con-
venient diagnostic tool for identifying features of decategorialization.
For example, categories such as nouns and verbs—however one may
wish to define them—form distinctly larger paradigms than the gram-
matical categories derived from them.
In a more general sense, (d) also relates to Hopper’s (: )
principle of specialization, which applies, for example, when at one
stage within a functional domain there was a variety of choices which
gets narrowed down, and the smaller number of forms consequently
assume more general grammatical meanings.
External decategorialization relates to the status of the grammati-
calizing form vis-à-vis other parts of discourse. It includes but is not
restricted to what Lehmann ( []: –) describes as syntag-
matic variability. Changes commonly observed in external decategor-
ialization are listed in ().
() Salient changes associated with external decategorialization
a. Loss of independence as an autonomous form.
b. Increasing dependence on some other form.
c. Increasing obligatoriness of use.
d. Loss of the ability to be moved from its canonical position to
other positions in the sentence in ways that are characteristic
of the corresponding non-grammaticalized source item.
e. Loss of ability to be addressed, e.g. to be referred to anaphori-
cally or to be focalized.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
There are some strong correlations between the various changes listed
in (), and some may be interpreted as specific manifestations of other
changes. For example, (a) is likely to entail (b), and (c) may
imply both (a) and (b). Only (e) appears to be relatively inde-
pendent of the other changes.
(a) and (b) relate to Comrie’s (: ) criterion of ‘boundness’
and Lehmann’s ( []: , ) parameter of ‘bondedness’ and
the process of ‘coalescence.’ Gradual loss of morphosyntactic inde-
pendence of the form undergoing grammaticalization typically pro-
ceeds along the scale described in (). The ultimate outcome can be
loss of a morphological boundary between the grammaticalizing form
and its host or some other form.
() Free form (> clitic) > affix
(c) relates to Comrie’s (: ) term ‘obligatoriness’ and to Lehmann’s
( []: , ) parameter of ‘paradigmatic variability’ and the
process of ‘obligatorification,’ leading to a situation where the use of a given
form may be required in certain contexts, that is, its use tends to become
predictable in certain morphosyntactic slots of a word or a sentence.¹⁴
(d) corresponds to Lehmann’s (: ) parameter of syntag-
matic variability and his process of fixation. The criterion is based on
the observation that lexical or less grammaticalized items may be found
to occur in different positons of a sentence whereas fully grammatica-
lized forms are likely to be restricted to one particular syntactic slot.
Nouns undergoing external decategorialization tend to lose the syntac-
tic freedom of lexical nouns, such as the ability to be topicalized or
focalized. In the Yoruba language of southwestern Nigeria, object nouns
can be moved from their position after the verb to the clause-initial
position. Now, the noun ara (+ possessive pronoun) ‘body’ has been
grammaticalized to a reflexive marker, cf. (a), and in this capacity it
can no longer be placed clause-initially, (b) therefore is not well-formed.
() Yoruba (Kwa, Niger-Congo; Awolaye : )
a. Nwosu rí ara rɛ̀.
Nwosu saw body his
‘Nwosu saw himself.’
¹⁴ A useful distinction between ‘language internal obligatoriness’ and ‘communicative
obligatoriness’ has been proposed by Diewald (: ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DECATEGORIALIZATION
b. *Ara rɛ̀ rí Nwosu.
(body his saw Nwosu.)
Verbs tend to lose their ability to be moved around in the sentence like
full-fledged lexical verbs, or to conjoin with other verbs.
(e) is of a somewhat different nature, referring to a change that is
essentially semantic-pragmatic in nature, leading from discursively
primary to discursively secondary items (Boye and Harder ; see
Section . above). Paradigm examples are provided by nouns and
verbs, which tend to lose their ability to be focalized or referred to
anaphorically. This change is not restricted to open-class items such as
nouns and verbs, it affects some closed class items in the same way. For
example, demonstratives can be focalized (e.g. It is this that I want) and
referred to anaphorically (e.g. Look at that.—What?). But once they are
grammaticalized to markers of relative or complement clauses, as has
happened with English that, they tend to lose these abilities—in the
wording of Boye and Harder (), they shift from discursively
primary to discursively secondary forms.
The generalizations presented in this section were hedged by ‘may’
or ‘tend to’ for the following reason: The changes listed in () and ()
are not necessarily all present in a given case of decategorialization. For
example, the German verb werden ‘become’ was grammaticalized to a
future tense auxiliary () but it did not lose its inflections of person
marking, as can be seen in (). Furthermore, the German definite
articles der (), die (), and das () did not lose their inflections
of number, gender, and case in their development from demonstrative
to definite article.
() Modern German future
a. Ich werde kündig-en.
I . resign-
‘I’m going to resign.’
b. Er wird kündig-en.
he . resign-
‘He’s going to resign.’
There are a number of potential causes if in a given case of grammat-
icalization an expected decategorialization has not taken place. Some
have to do with language-internal factors: There may be specific struc-
tural constraints that prevent the loss of some categorial feature. For
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
example, in a number of languages verbal auxiliaries are the only
grammatical category in the clause where distinctions of personal
deixis, tense, aspect, or negation are encoded. Accordingly, giving up
these encodings might have dramatic consequences for the information
structure of the clause. Another cause pertains to the age of the
grammaticalization process involved: Decategorialization does not
happen overnight, that is, it takes some time to materialize, and the
younger a process is, the lower will be the degree of decategorialization
reached.
3.4 Erosion
Erosion means that a linguistic expression undergoing grammaticali-
zation and being used frequently loses part of its phonetic substance,
typically at its left or right boundary.
In some approaches, erosion is treated jointly with other phenom-
ena. For example, Lehmann’s (: ) parameter of integrity in-
cludes not only erosion, which in his terminology relates to ‘decrease in
phonological integrity’ or ‘phonological attrition,’ but also desemanti-
cization, referred to as ‘decrease in semantic integrity.’
Erosion, frequently labeled as ‘phonetic reduction,’ does as a rule not
occur in the initial stage of grammaticalization (Norde : ). As
was noted in Section ., this parameter is usually the last to apply in
grammaticalization processes. Evidence is provided by a number of
documented cases of grammatical change. The following are just a few
of the examples in support of this hypothesis:
• The development of the English be-going-to future is sketched by
Hopper and Traugott (: –) in terms of a four-stage sce-
nario. On this scenario, there was reanalysis at stage II and analogy
at stage III, but the erosion of going to to gonna marks the final
stage IV after the expression had been conventionalized as a future
tense marker.¹⁵
¹⁵ Note that erosion is treated by Hopper and Traugott (: ) as ‘reanalysis.’ Hence,
the development from going to to gonna is interpreted by them as a second instance of
reanalysis.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. EROSION
• The demonstrative disi ‘this’ (< English this) of the English-based
creole Sranan of Suriname developed into a relative clause marker
around the end of the seventeenth century. Subsequently, it devel-
oped further into a subordinator of temporal, causal, and concessive
clauses in or before the second half of the eighteenth century. But
erosion from disi to di occurred only in the course of the nineteenth
century (Bruyn a, b, ; see also Section .).
Furthermore, erosion is not a requirement in grammaticalization. For
example, we noted in Section . that, in spite of its history of nearly
one millennium as a tense-aspect marker, the High German auxiliary
haben did not undergo erosion. It is still phonetically indistinguishable
from its lexical source, the possessive verb haben ‘to have, possess.’ And
much the same applies to its Latin equivalent habēre ‘to hold, possess’,
cf. (a), which has also given rise to the perfect aspect auxiliary avoir
in French, cf. (b). Nevertheless, its form as a verb of possession and as
an auxiliary in French is the same.¹⁶ In a similar way, the French negation
marker pas ‘not’ has retained the phonological form of the noun pas
‘step, pace’ from which it is historically derived (Hopper : ).
() Modern French
a. Nous avons faim.
we have. hunger
‘We are hungry.’
b. Nous avons dormi bien.
we have. slept well
‘We have slept well.’
According to Heine and Kuteva (: ) there are two basic types of
erosion. The first type, called morphological erosion, involves entire
morphological units. For example, when the Old English adverbial
phrase ϸa hwile ϸe (that. time. that) ‘that time that’ or any of
its variants was grammaticalized to the temporal and concessive sub-
ordinator while in Modern English (Traugott and König ), this
meant that morphological segments were lost. And much the same
¹⁶ This does not apply to the development of Latin habēre as a future tense auxiliary,
which has been affected by all parameters of grammaticalization, including massive ero-
sion, in almost all of the Western Romance languages.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
happened in the case of its Old High German counterpart al di wila daz
‘all the time that,’ which was grammaticalized to the causal subordina-
tor weil ‘because’ in Present-Day German, which is also characterized
by loss of morphological elements.
The second type, let us call it phonological erosion, is distinctly
more widespread. It is restricted to phonological, or phonetic features
that are lost in the process of grammaticalization. Phonological erosion
most commonly involves the changes listed in (), or any combination
thereof.
() Kinds of phonological erosion
a. Loss of phonetic segments, including full syllables.
b. Loss of suprasegmental features, such as stress, tonal distinc-
tions, or intonation.
c. Loss of phonological autonomy and adaptation to adjacent
phonological units.
d. Phonological simplification.
e. Boundary loss.
The development in English from because to colloquial English coz is
an instance of (a): It entailed both loss of phonetic segments and
reduction from a disyllabic to a monosyllabic unit. Since the eroded
form coz does not occur in all varieties of English, there is reason to
assume that erosion was a process that took place subsequent to
desemanticization and decategorialization.
The grammaticalization of the English adjective full to the derivational
suffix -ful illustrates (b), as it entails a loss of the ability to be stressed.
Loss of distinctive stress is in fact a fairly widespread effect of grammat-
icalization, for example in the process from demonstrative modifier to
definite article. Thus, Greenberg (: ) notes that in a number of
languages, demonstratives and definite articles have the same or almost
the same form. However, they differ in particular in that the former may
carry stress while the latter are unstressed. Examples include definite
articles in German (der), Hungarian (az), and in the Uto-Aztecan lan-
guage Papago (he?g) (see also de Mulder and Carlier ).¹⁷
¹⁷ There may be other differences in addition. For example, the Hungarian demonstra-
tive form az loses not only stress as an article but also final -z if the next word begins with a
consonant (Greenberg : ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. EROSION
(c) means that the phonetic shape of a grammaticalizing form is
modified and adapted to adjacent segments. It refers to what Hopper
and Traugott (: ) treat as one of the characteristics of ‘reanalysis,’
namely change in degree of cohesiveness, for example when the English
future form be going to changed optionally into be gonna.
Paradigm cases of (d) involve phonetically complex sound units
that in the process of grammaticalization lose part of their phonetic
features. For example, many West and central African languages have
labial-velar consonants, such as kp and gb, which are phonetically more
complex than the corresponding labial (p, b) or velar consonants (k, g).
Accordingly, in the Ewe language of Ghana and Togo, gb was “simpli-
fied” to a corresponding velar g in some forms that underwent gram-
maticalization. Thus, the noun gbé ‘location, direction’ gave rise to the
ingressive aspect marker -gé, and the verb gbɔ ‘to return’ to the repet-
itive aspect prefix -ga (Heine : ).
In quite a number of cases, both morphological and phonetic erosion
are involved. For example, the grammaticalization of the phrase by the
side of to the preposition beside in Modern English appears to have
involved both morphological and phonological erosion. Similarly, the
development of the Latin phrase in casa ‘in the house (of)’ via Old
French (en) chies or chiése to the Modern French locative preposition
chez ‘at’ involved loss of both morphological and phonological features
(see Section .).
Boundary loss (e) means that a phonetic boundary disappears.
Such a boundary can either be one between different parts of the
grammaticalizing element, or one separating a grammaticalizing ele-
ment from an adjacent word. An example of the former is provided by
the English future marker (be) going to when it is optionally reduced to
(be) gonna. The latter can be observed, for instance, when a plural
marker fuses with the noun to which it belongs, becoming an insepa-
rable part of the noun, as has happened in the case of English plural
noun forms like children, oxen, etc.
Loss of a phonetic boundary tends to be paralleled by loss of a
morphological boundary (see Section ., (c)), but the two need not
coincide.
In accordance with the definition provided above, ‘erosion’ is a
technical term of grammaticalization theory which should be distin-
guished from ‘phonetic reduction’ and other similar terms, for the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
following reasons: On the one hand, phonetic reduction is understood
as a more general kind of change which need not involve grammati-
calization. For example, the English copula verb ’s in She’s a teacher is a
phonetically reduced version of is, but this is not due to grammatical-
ization. And second, erosion also differs from phonetic reduction in
that it includes loss of morphological elements, as we saw above.
Erosion can have dramatic consequences for the affected forms, to
the extent that a grammaticalized form may no longer be conceived as
being historically related to its lexical or other source. For example, not
every speaker of English may be aware that the indefinite article a in a
car is historically derived from the numeral one. In the Chadic language
Lamang of Northeastern Nigeria, the forms ghvà and -và do not look
strikingly similar, the former being a noun, also serving as a reflexive
marker, whereas -và is a suffix serving verbal reflexive and middle
functions (Wolff : –). But there is common grammaticaliza-
tion process not only found in Africa but also in other parts of the
world leading from lexical nouns for ‘body’ or ‘head’ to reflexive, and
even to middle and passive markers (Heine b; Schladt ;
Wiemer : ; Kuteva et al. , > ).
At the same time, erosion can be restricted to subtle effects which are
hard to identify without detailed phonetic analysis. For example, iso-
lating languages of Mainland Southeast Asia, such as Burmese, Thai,
Khmer, or Vietnamese, exhibit grammaticalization on a massive scale,
leading, for example, from nouns and verbs to a range of functional
markers. Quite commonly, the phonological forms of the grammatica-
lized markers and their lexical sources are ostensively the same. As
phonetic experiments carried out by Ansaldo and Lim (: –)
suggest, however, even in such languages grammaticalizing words may
undergo subtle changes in pronunciation. The authors demonstrate
that in two morphologically strongly isolating and tonally complex
Sinitic languages spoken in Singapore, namely Cantonese and Hokkien,
function words show vowel/syllable reduction, that is, erosion typical of
grammaticalization.
Erosion is on the whole a fairly stable unidirectional change. Never-
theless, a number of cases of opposite changes have been documented,
refererred to as phonetic strengthening; see Norde (: ) for
discussion.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONCLUSIONS
3.5 Conclusions
In spite of the fact that there are many correlations between the various
parameters, we have treated each of them independently. There are two
main reasons for this procedure. The first is that each parameter is
devoted to a distinct component of language structure: Context exten-
sion deals with text pragmatics, desemanticization with semantics,
decategorialization with morphosyntax, and erosion with phonology
and phonetics. It is therefore possible to analyze the contribution of
each component in its own right. The second reason is that despite the
large amount of research that has been devoted to this subject matter,
the nature of the correlations between the parameters is still to some
extent unclear. What seems to be clear, however, is that the effects of
the four parameters do not, as a rule, coincide in time.
None of the parameters is sufficient to define grammaticalization as a
process; rather, each of them applies also to other kinds of change, in
particular to lexical change, or to lexicalization. But taken in conjunc-
tion, they provide a reasonable tool for identifying such processes, and
they are also instrumental to accounting for directionality in grammat-
ical change (see, e.g., Diewald : ).
And in fact, many pathways of grammaticalization bear witness to
the joint effect of parameters. One fairly widespread pathway is that
from demonstrative to definite article, for example in many European
languages (Hawkins , ch. ; Heine and Kuteva : –). First,
there was context extension in that the demonstratives were used in a
broader range of contexts, extending for example from visible referents
(e.g. Mind the step!) to non-visible ones (The king has abdicated.).
Second, there was loss of the meaning as exophoric spatial deictics
(desemanticization). Third, the demonstratives also experienced loss of
part of their morphosyntactic independence, including increase in ob-
ligatoriness of use (decategorialization) and, fourth, the demonstratives
also lost part of their phonetic substance (erosion). These parameters did
not all apply at the same time, and their effects were not exactly the same
in all languages concerned. For example, erosion in French led to the loss
of segmental phonetic substance in the development of Latin ille ‘that
()’ to French l(e) ‘the ().’ In German, by contrast, the most
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
FOUR PARAMETERS OF GRAMMATICALIZATION
conspicuous phonetic change appears to have been the loss of supraseg-
mental substance, namely loss of distinctive stress on the way from
demonstrative dér ‘this ()’ to article der ‘the ()’ (Greenberg
: ). But in spite of such differences, the joint effect of the
parameters allows treating them as manifestations of one distinct proc-
ess, namely one that leads to the rise of new functional categories and is
similar across languages—that is, a process that can be distinguished
from other kinds of linguistic change, such as lexical change.
Discussion points
() Is there any of the four parameters discussed in this chapter that is more
essential than others in your view? How will emphasizing one or the other
parameter correspond to a different view of grammaticalization overall?
() Which of the parameters is the easiest to observe empirically? What tools
or materials would be needed to observe each of them empirically?
() How would the parameters apply to one of those examples of grammat-
icalization in Chapter that were discussed in some detail (Spanish ana-
lytic future, possessive construction, English progressive, etc.)?
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
4
Directionality of semantic
change in grammaticalization
4.1 Bleaching, generalization, abstraction
4.2 Discourse orientation
Discussion points
Semantics is a long-standing topic in grammaticalization studies since
in many approaches to grammaticalization semantics is assigned a
central role. In this sense, discussion of semantics is pervasively repre-
sented in this book.
This chapter will nevertheless pick up one issue in particular that has
attracted much attention in the research literature, namely the question
of directionality of semantic change in grammaticalization. First we
discuss whether it is useful to think of the basic semantic change in
grammaticalization in terms of ‘bleaching,’ ‘generalization,’ or ‘abstrac-
tion’ (Section .), and then we address the question of ‘subjectifica-
tion’or ‘discourse orientation’ in grammaticalization (Section .).
4.1 Bleaching, generalization, abstraction
In Section . the term ‘desemanticization’ was introduced, defined as
the loss of semantic properties that a linguistic expression experiences
in grammaticalization, both from lexical to grammatical and from
grammatical to more grammatical meaning. An alternative term that
is widely used is ‘bleaching,’ which is an interesting concept for two
main reasons. First, it has been commonly used in grammaticalization
studies. Nevertheless, it has been controversial. Some researchers
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
consider it as the basic mechanism of semantic change, but others reject
it. Adoption and rejection reflect rather straightforwardly some of the
basic theoretical stances towards grammaticalization discussed in
Chapters and . Secondly, it is interesting because ‘bleaching’ is an
everyday word that has been used metaphorically as a technical term
mostly without an explicit definition. That is, for many scholars, its use
presupposes a tacit understanding of what it should mean. But precisely
those different intuitive understandings of the word are another reason
for its adoption vs. rejection, as we will see.
The earliest use of the term is attributed to von Gabelentz (:
), who mentions it in the context of new ‘expressive’ grammatical
forms that emerge to replace older ones that are bleaching out (ver-
blassen, verbleichen). In modern grammaticalization-related research it
has been brought up again by Givón (a) and has been present ever
since. Terms that are mentioned as having similar meaning but whose
actual denotation in contrast to each other is not always clear either are
‘weakening,’ ‘degradation,’ ‘desemanticization,’ ‘depletion,’ and ‘loss’ (cf.
Sweetser ; Heine et al. : ; Hopper and Traugott : ).
A simple dictionary definition of ‘bleach’ that may be the smallest
common denominator is ‘to grow white or lose color’ (Merriam Web-
ster online, bleach intransitive verb; https://www.merriam-webster.
com/dictionary/bleach). There are three main aspects to this meaning.
First is a property or substance that is being lost, namely that of color.
Second is a substance that undergoes the loss of a property or attached
substance but remains largely unaffected by the change. This is the
thing that ‘grows white or loses color,’ for example, a cloth. Thirdly, we
understand that in a natural setting, without decisive human interven-
tion, the loss of color is not sudden but gradual, for example through
long-term exposure to sunlight or through repeated exposure to water
and detergent. We could thus define ‘bleaching’ with respect to lan-
guage change as a ‘process in which a specific part of the meaning of a
linguistic form is gradually weakened and eventually disappears, while
another part of it remains unaffected.’
Now it seems that most differences in thinking about ‘bleaching’ as a
metaphor for meaning change emerge from whether the fact that there
is a substance largely unaffected by the process and that remains is
acknowledged or not. If the ‘bleaching’ is understood merely as loss
of color without reference to the remaining substance, ‘bleaching’
is simply another word for ‘loss.’ On the other hand, if ‘bleaching’ is
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. BLEACHING, GENERALIZATION, ABSTRACTION
understood from the perspective of the substance undergoing the
process, a specific part of the semantic structure gets lost while another
one remains. Secondly, it is sometimes ignored that ‘bleaching’ as a
metaphor from everyday language typically implies a gradual process
and not a sudden one.
Proponents of the term have come mainly from three camps of
grammaticalization research, namely the language processing-oriented
research (cf. Section .), formal approaches to grammaticalization (cf.
Section .), and Cognitive Grammar (cf. Section .). For the language
processing-oriented approach, as represented by Bybee (, ,
b, ), ‘bleaching’ is a process inexorably linked to the way gram-
maticalization proceeds, namely through repetition and habituation.
Repetition itself diminishes the force of a word, phrase or construction. . . . In gram-
maticization, the generalization or bleaching of the meaning of a construction is
caused by frequency, but it also contributes to additional frequency, since a generalized
construction can be used in more contexts, leading to an almost inexorable movement
along a grammaticization path. Constructions that have been bleached of their more
specific lexical meaning are more likely to pick up inferential meaning from the
context, that is, grammatical meaning. (Bybee : )
As seen in this quote, Bybee uses the terms ‘bleaching’ and ‘general-
ization’ interchangeably. This process of bleaching or generalization
occurs at the very beginning of grammaticalization, caused by repeti-
tion, and enables further grammaticalization. The reason is that “the
loss of specificities of meaning makes a morpheme applicable in a more
general range of contexts. For example, if will [, which grammaticalized
from a verb of volition to a future modal] does not signal volition, it can
be used with a wider range of subjects, including inanimate objects”
(Bybee b: ). However, the process does not stop here:
Generalization seems to characterize the entire grammaticization continuum—we note
that as the process unfolds, grams always become more general and more abstract in their
meaning, more widely applicable and more frequently used. (Bybee a: )
We can see that for Bybee, ‘bleaching’ is not an indiscriminate loss of
meaning but rather that specific aspects of meaning are lost (“the lexical
meaning is bleached of specificities of meaning, or generalized as
specific components of meaning are lost” (Bybee : )). This is
a faithful application of the ordinary language term ‘bleaching’ as a
metaphor for semantic change. It does not preclude retention of
meaning (“grams encode a meaning that is at once abstract and general,
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
but in addition contains traces of its former lexical meaning” (Bybee
: )). It also does not preclude acquisition of other meanings
(“Constructions that have been bleached of their more specific lexical
meaning are more likely to pick up inferential meaning from the
context, that is, grammatical meaning” (Bybee : )). And it is
pervasive in grammaticalization.
Starting out from a quite different theoretical background, some
generative grammarians analyzing grammaticalization have been
equally positive about the application of the term ‘bleaching’ to the
process. Abraham (, ), for example, similar to Bybee, takes
bleaching as a semantic process that accompanies grammaticalization
from beginning to end. Roberts (: ) likewise considers ‘bleach-
ing’ “an intuitive, useful and expressive description” of the semantic
aspect of grammaticalization. However, he suggests that the term is in
need of explication. Specifically, a certain type of meaning of a word or
morpheme is subject to ‘bleaching,’ and that is non-logical or descrip-
tive meaning, while logical meaning is retained. For example, in Indo-
European languages, words with the meaning ‘whole,’ that is, ‘being in
good shape, not being damaged, not lacking in any of the integral parts,
functioning properly’ are the most common source for quantifiers with
the meaning ‘all.’ The meaning ‘whole,’ that is, ‘being in good shape,’ is
a property of individuals, and as such a descriptive, non-logical mean-
ing. In contrast, the quantifier ‘all’ is a logical constant that denotes a
subset relation between two sets. Another case would be French noun
rien ‘thing’ that ended up as a negation word (‘nothing’) in Modern
French. Rien ‘thing’ refers to an entity in the world and its interpreta-
tion requires knowledge about the world. In contrast, to interpret rien
as a negator, it is only necessary to know the relation between two sets
(Roberts : –).
Lastly, grammaticalization is also conceptualized as ‘bleaching’ in
Cognitive Grammar, but we will leave the details on this approach to
Section ..
The concept of ‘bleaching’ has come under fire on a number of
accounts. First of all, it is the idea of a one-sided weakening or loss of
meaning that has been criticized (e.g. Sweetser ; Hopper and
Traugott : –). However, as we have seen above, the idea of
bleaching as a mere equivalent of weakening and loss of meaning is
neither warranted by the meaning of the original everyday word nor by
the actual use of the term by its proponents. According to Hopper and
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. BLEACHING, GENERALIZATION, ABSTRACTION
Traugott (: ), “[p]erhaps the most damaging evidence” against
the idea of bleaching in grammaticalization comes from the fact that
“when a form undergoes grammaticalization from a lexical to a gram-
matical item, some traces of its original lexical meanings tend to adhere
to it,” a phenomenon known as ‘persistence’ (cf. Section .). This
argument overlooks the fact that it is exactly this kind of ‘persistence’
that is definitory for bleaching in contrast to mere weakening and loss.
As the proponents of the concept emphasize, a constitutive part of
bleaching is that an abstract or schematic semantic structure that was
already part of the lexical meaning remains or is laid bare through the
process. For Abraham (, ), total loss of lexical meaning is a
late stage of grammaticalization that is hardly ever reached. More
commonly, some of the lexical meaning remains in the process of
bleaching.
Hopper and Traugott () however, do not entirely deny the
validity of a concept of bleaching. For them, as for Fischer and
Rosenbach (), bleaching is only characteristic of the last stages of
grammaticalization. This contrasts especially with Bybee (, ,
b, ), for whom bleaching is an essential part of the process
from the beginning to the end. For her, bleaching is a condition for the
development of grammatical meaning, since “constructions that have
been bleached of their more specific lexical meaning are more likely to
pick up inferential meaning from the context, that is, grammatical
meaning” (Bybee b: ).
Moving on from criticism that is mainly based on a different under-
standing of the term ‘bleaching’ to more substantial counterarguments,
Sweetser () points out a specific problem with the term ‘general-
ization.’ Recall that this term has been used as a synonym for bleaching
in Bybee’s (, b, ) work. Generalization seems a perfectly
suitable term when, for example, markers of progressive or habitual
become broad ‘imperfects,’ which encompass both the original mean-
ing and additional imperfect meanings. However, when a deontic
modal such as must (You must be back by five) develops an epistemic
sense (The Beetle must be the most loved car of all times), the epistemic
sense does not encompass the original deontic sense. But apparently the
problem here lies with the way the target category is conceptualized.
The imperfect is seen as a collection of various imperfect senses or uses,
including habitual and progressive. Likewise, if the overall meaning of a
modal that results from the extension to epistemic meaning is taken as
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
the target of change, instead of only the extension towards epistemic
meaning itself, the same holds for the modal as for the imperfect: The
overall result is a more general meaning, in this case including episte-
mic and deontic or other non-epistemic uses of the modal.
The most substantial counterarguments against the validity of
‘bleaching’ are probably (a) the development of meanings and functions
seemingly unrelated to the original lexical meaning, and (b) the devel-
opment of polysemies that seem to include unrelated meanings, as laid
out by Heine et al. (: –). That is, the new grammatical meanings
are not merely a filtered out, schematized, or shifted version of the
original lexical meaning but something genuinely new. According to
Heine et al. (: –), (a) is the case for the French negation markers
pas, personne, point, and rien, which derive from nouns denoting,
respectively, ‘step,’ ‘person,’ ‘point,’ and ‘thing,’ or in the case of body
part terms like ‘hand’ or ‘eye’ that gave rise to directional meanings such
as ‘from’ and ‘towards,’ respectively, in African languages. As for (b), it
seems hard to explain from the perspective of bleaching why general
verbs for ‘take,’ ‘leave,’ or ‘let’ can have given rise to such diverse
grammatical functions as completive focus and perfective aspect. How
can they be gained from a mere schematization of the original lexical
meaning? For Heine et al. (: ) cognitive and pragmatic mechan-
isms must decisively alter or add to the meaning in order to reach these
apparently unrelated new grammatical meanings.
However, even this seemingly compelling counterargument cannot
be left undisputed, as Roberts (: ) shows. For him, the negative
sense of rien is only seemingly unrelated to the original lexical meaning
‘thing.’ That is, it is intrinsically related through the loss of non-logical
descriptive meaning, with only logical meaning remaining.¹ This con-
stitutes a change “from λx (thing (x)) (‘the set of x such that x is a
thing’), to (taking n-words to be a kind of negative quantifier) λPλQ
￢∃x[P(x) & Q(x)] (‘the sets P and Q such that nothing is in both’)”.
Likewise, a closer investigation of the source and target meanings
mentioned by Heine et al. (: –) may reveal other semantic
relationships overlooked by the authors.
¹ Logical meanings, for Roberts (: ) are ‘permutation-invariant’ meanings, that is
“meanings which do not depend on empirical facts for their contribution to truth; such
interpretations depend purely on set-theoretic or numerical aspects of relations among
referents, not on any intrinsic properties of the referents themselves.” Non-logical mean-
ings in contrast depend on the properties of individuals.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. BLEACHING, GENERALIZATION, ABSTRACTION
Heine et al. () proposed ‘abstraction’ as a principle in the place
of ‘bleaching.’ Abstraction is considered a fundamental principle of
language by some cognitively oriented linguists. Fortescue (: ), for
example, suggests that there are two major types of abstraction, namely
‘disembodiment from specific sensory features’ and ‘simplification or
generalization across instances.’ The latter clearly lines up with ‘bleach-
ing’ and ‘generalization’ in grammaticalization as discussed above, and
as acknowledged by Fortescue himself (: ). Fortescue seeks a
synthesis of the two types of abstraction in Whitehead’s ( []:
) concept of ‘transmutation,’ in which “complex arrays of sensory
input [ . . . ] are integrated and simplified in order to be experienced as
unified ‘conceptual feelings’.” This process of abstraction underlies
practically all aspects of language including phonology, for example
recognition of successive phonemes, semantics, for example recogni-
tion of semantic fields and frames, syntactic constructions and textual/
discourse embedding (Fortescue : –), and is also at work in
grammaticalization.
Heine et al. (: –) in light of their counterargument against
‘bleaching’ are not satisfied with abstraction in terms of simplification
and generalization, since it does not seem to explain the development of
very diverse, and sometimes even seemingly unrelated new grammat-
ical meanings. They therefore turn to the concept of ‘metaphorical
abstraction’ by Schneider () that serves to relate more abstract
contents to more concrete contents across conceptual domains, and
thus introduces an element of creativity. Metaphorical abstraction can
be either ‘structure-preserving,’ which means that a lexical item
remains a lexical item but the meaning becomes more abstract, or
‘structure-changing.’ The latter is the case when, for example, a noun
not only adopts a more abstract meaning (e.g. ‘head’ becomes ‘intel-
lectual ability’) but also starts to participate in new, grammatical con-
texts, such as becoming a postposition or clause subordinator.
If we think of ‘abstraction’ in terms of the concrete example of future
be going to, which has been treated as a prototypical case of grammat-
icalization (see also Sections ., .), the process starts out with a verb
that has the concrete meaning of physical motion in space, with an
animate entity as the subject of motion and a physical goal. The
resulting future marker is bereft of all the concrete elements of mean-
ing, that is, the motion in space, the physical goal, and the entity that
moves. What is left is the abstract mental projection of the realization of
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
an event at a point of time ahead of the time of speech. The relationship
between source (motion towards a goal) and the target (future) of
grammaticalization can be described in terms of metaphor from
space to time, although the actual historical steps of development
may be more aptly described in terms of metonymy (cf. Section .).
In summary, although being fundamental to the process of gram-
maticalization, and having been discussed from early on, the question
of the role of bleaching or generalization vs. abstraction is still unre-
solved. Part of the problem has been the lack of definition of the terms,
especially bleaching, and the resulting gap in what scholars have meant
when they referred to it. Moreover, the attitude towards bleaching,
generalization and abstraction has been strongly influenced by the
basic concept of grammaticalization a scholar espouses. The terms
‘bleaching’ and ‘generalization’ seem to appeal to frameworks that
emphasize the mechanical aspect of semantic change in grammatical-
ization, the term abstraction to frameworks that emphasize a creative
aspect, and for yet other frameworks, none of these terms are particu-
larly appealing. In any case, more in-depth and detailed research,
including psycholinguistic research would be required to provide
more definite answers that are acceptable across frameworks.
4.2 Discourse orientation
‘Subjectification’ and ‘intersubjectification’ in semantic change in gen-
eral and grammaticalization in particular are concepts that have
sparked a host of research activities in these fields especially since the
s. Initially they were introduced through diachronic work by
Traugott (, b, and elsewhere) and synchronic work by Lan-
gacker (, , ; ‘subjectification’ only). In this chapter, we
take a broad perspective on the topic, labeled as ‘speech-act orientation’
in Narrog (a, b, b), and re-labeled as ‘discourse orienta-
tion’ here, to make clear that we are not referring to ‘speech acts’ in a
technical pragmatic sense. Under this perspective it is assumed that in
semantic change in general and in grammaticalization in particular,
meanings tend to become increasingly oriented towards the speech act
participants, that is, speaker and hearer, and towards organizing speech
or discourse itself. Traditional ‘subjectification’ is only a small part of
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
this overall development. In summary, the three tendencies of semantic
change are:
I increasing orientation towards the speaker’s perspective (increased
speaker orientation),
II increasing orientation towards the speech situation including the
hearer (increased hearer orientation), and
III increasing orientation towards speech (text) and discourse itself.
In the following sub-sections we first discuss each of these tendencies
(speaker orientation, Section ..; hearer orientation, Section ..;
textual discourse orientation, Section ..), then their relationship to
grammaticalization (Section ..), and lastly possible counterexamples
or divergences from these tendencies (Section ..).
4.2.1 Increase in speaker orientation
The notion of speaker orientation is anchored on the notions of
subjectivity, and its dynamicization, subjectification, as well as speaker
deixis. These are notions with a long tradition in philosophy, psychol-
ogy, and linguistics. In linguistics, there are three major concepts of
subjectivity that are still currently relevant (for details, see Narrog
a, ch. ; b).
A Subjectivity in terms of speaker commitment.
B Subjectivity in terms of speaker deixis (and construal).
C Subjectivity in terms of accessibility of information.
(A) represents the most common understanding of the term. It is, in
terms of Lyons (: ), “the locutionary agent’s expression of
himself and his own attitudes and beliefs.” Or, as put by Verstraete
(: ), it is about “taking positions of commitment with respect
to the propositional content of the utterance.” While we do not claim
here that these are identical ideas, they squarely fall within the range of
a speaker expressing themselves and their attitudes. (B) is a concept
that was mentioned by Lyons () mainly with respect to tense and
demonstratives. It has been used by Abraham (, ) as an
alternative to the concept of subjectivity, and has been developed in a
theory-specific way in Cognitive Grammar (Langacker , ). In
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
Langacker’s framework, linguistic expressions are viewed in terms of
construals involving a conceptualizer (speaker) and an object of concep-
tualization (an event). A construal is conceived of as subjective if the
conceptualizer’s perspective is reflected but not explicitly put “onstage” in
a linguistic expression. Otherwise it is objective. Unlike concepts (A) and
(C), the Cognitive Grammar concept of subjectivity does not require
reference to the context but is instead identified with specific linguistic
forms and constructions, and thus a matter of grammar. Also, in contrast
to (A), reference to the speaker is as a rule implicit instead of explicit.
(C) is a more technical understanding of ‘subjectivity’ that also goes
back to Lyons’ (: ) writings. Proponents of this concept are
Nuyts (a, b) and Portner (). For them, subjectivity means
that “[the speaker] alone knows the evidence and draws a conclusion
from it,” while in intersubjective sentences “[the speaker] indicate[s]
that the evidence is known to (or accessible by) a larger group of people
who share the conclusion based on it” (Nuyts a: ). This is an
entirely pragmatic concept.
Table . presents ‘speaker orientation’ as a synthesis of concepts of
‘subjectivity’ to the extent that they are compatible with each other. The
examples are mainly from the category of modality.
When referring to diachronic change involving speaker orientation,
we are assuming that there are cases in which it eventually becomes
associated with linguistic forms. This process of increasing association
of a form or construction with subjectivity has been called subjectifica-
tion. Here we speak of an increase in speaker orientation. In keeping
with the concept of speaker orientation developed here and represented
in Table ., an increase in speaker orientation may take place along the
following parameters:
(i) increase in subjective content or meaning associations of a
lexical item;
(ii) increasing constraints on the use of a form in terms of morpho-
syntactic combinability;
(iii) increasing use in constructions associated with subjectivity;
(iv) increasing use in contexts associated with subjectivity.
Of primary interest here is an increase in speaker orientation in
grammar (ii to iv). Loss of tense/aspect marking on a grammaticalizing
verb is a potential example of parameter (ii). Increasing use of a
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
Table 4.1 Features of speaker orientation
Feature Concept Examples
i. Inherently subjective Interpersonal Inherently subjective lexical items (e.g. idiot,
lexical meanings accessibility gorgeous)
ii. Constraints on the Speaker-deixis Tense and mood inflections, demonstratives;
use of a form in highly grammaticalized modality; e.g. in a
terms of language with tense inflection, a modal
morphosyntactic marker lacks the possibility for past inflection,
combinability and thus lacks this specific possibility of
descriptive use
iii. Actual use of a form Commitment A deontic modal marker used in the present
in a specific syntactic with a second person subject is usually
construction performative, and thus more subjective
(Coates )
Interpersonal Mental state verbs with first-person subject
accessibility usually indicate that the judgment is
personally associated with the speaker, and
neither invokes shared knowledge nor a
shared judgment (Nuyts a: –)
iv. Discourse- and extra- Commitment Modification of modal markers that indicates
linguistic context distancing, and thus lower performativity (e.g.
past tense, negation), may in fact have only a
mitigating function. In context, the utterance
is actually performative; e.g. I thought . . .
instead of I think . . . in expressing a counter-
argument
Interpersonal An evaluation can be based on a general rule
accessibility (more objective) or on the speaker’s personal
values (more subjective)
Features (i) and (ii) are part of linguistic forms, and therefore semanticized (i) and grammaticalized (ii). Features
(iii) and (iv) are essentially bound to context, and therefore inherently linked to pragmatics.
grammaticalizing verb with inanimate or impersonal subjects in met-
aphorical meaning would be a potential example of parameter (iii).
Parameter (iv) would be reflected in changing frequencies of use in
different environments, for example, increasing use in contexts associ-
ated with speaker–hearer interaction. Thus, (ii) may be conceptualized
in terms of distinct steps that can be identified formally, but (iii) and
(iv) are factors primarily associated with changes in frequency. In this
manner, increase in speaker orientation is a gradual process.
Example () illustrates a grammaticalizing construction associated
with subjectivity in sense (A), in expressing a speaker’s attitude. The
development of the English progressive was discussed in Section ..
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
and we repeat an example here. The Old English progressive in ()
marks emphasis, especially on expressing some kind of complaint or
criticism.
() Old English (Kranich a: )
Gyf þonne Frysna hwylc . . . ðæs morþor-hetes
If yet Frisian any of.this feud
myndgiend wære
mentioning would.be
‘if on the other hand one of the Frisians would mention this feud’
(Beowulf; ca. )
Another example that is often cited as a case of subjectification, and
constitutes an increase in ‘speaker orientation’ as defined here is the
functional extension of the English modals to express epistemic mean-
ings in constructions that express the speaker’s reasoning. Must in ()
(Middle English) expresses a general objective necessity based on
physical or social conditions, while () is one of the earliest examples
in which it expresses an epistemic necessity in the speaker’s world of
reasoning (Goossens : –).
() Middle English (Goossens : )
Wherto and why burieth a man his goodes by his grete avarice,
and knoweth wel that nedes moste hy dye? For deeth is the ende
of every man . . .
‘For what purpose and why does a man bury his goods because of
his great avarice, knowing well that he must necessarily die? For
death is the end of every man.’
() Early Modern English (Fryer: A New Account of East India; cited
in Goossens : )
The Portugals striving to possess themselves of Muschat, were put
to such stress, that had not their Armado come to their relief, they
must have desisted their Enterprize: . . .
‘The Portuguese, striving to acquire Muscat, were put under such
stress, that had not their fleet come to their relief, must have
abandoned their campaign: . . .
In the sense of the four criteria, the rise of epistemic uses in a modal like
must corresponds to an increasing use in constructions and contexts
associated with subjectivity. In terms of concepts of subjectivity, this
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
development would primarily constitute an increase in speaker
orientation in terms of (B) speaker deixis, since epistemic modals
refer to the estimation of likelihood by the speaker rather than a
condition on the subject.
4.2.2 Increase in hearer orientation
While speaker orientation is based on the discussion of concepts of
‘subjectivity,’ ‘hearer orientation’ is anchored on the discussion of
concepts of ‘intersubjectivity’ in previous research. And just as there
are three major notions of ‘subjectivity’ in linguistics, there are three
major notions of intersubjectivity currently in use, as listed below (cf.
Narrog b for details).
A Intersubjectivity in terms of attention towards the addressee.
B Intersubjectivity in terms of (a) hearer deixis, or (b) speaker plus
hearer deixis.
C Intersubjectivity in terms of accessibility of information (also:
‘evidentiality’).
‘Intersubjectivity’ is a concept less common and less discussed in
linguistics than subjectivity. Outside of, and prior to its use in linguis-
tics, it is mainly found in philosophy. (A), (C), and partially (B) can be
identified with specific authors. Traugott (c: ) defines intersub-
jectivity in the sense of (A) as the “speaker’s attention to addressee self-
image.” Intersubjectivity in this sense complements subjectivity and is
opposed to objectivity. It is also relatively narrow, as is clear from the
definition above.
As for (B), we are not aware of any specific scholar operating
explicitly with hearer deixis. The expression ‘hearer deixis’ (Ba) as a
counterpart to ‘speaker deixis’ seems problematic, because the ultimate
deictic center in language is arguably always the speaker, or at least
includes the speaker. However, there are expressions in language that
are centered on the addressee in an implicit manner, the same way that
there are expressions that are implicitly centered on the speaker. These
include imperatives (where no explicit ‘you’ is required), questions, or
addressee honorifics in languages that have them. So, if ‘hearer deixis’ is
a term to be avoided, at least ‘hearer centeredness’ or ‘hearer orienta-
tion’ is a useful concept to have.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
On the other hand, there is also a concept of deixis to speaker and
hearer that has been labeled with the term ‘intersubjectivity’ (Bb).
Verhagen (, ) modified Langacker’s concept of construal in
a way that intersubjectivity replaces subjectivity as the basis. While
Langacker identifies the “ground” in a construal primarily with the
speaker, Verhagen identifies it with the presence of two conceptuali-
zers, the speaker and the hearer. The ground as speaker plus hearer
then provides the intersubjective basis of a construal. Thus, intersub-
jectivity in his sense is speaker plus hearer deixis. Furthermore, inter-
subjectivity conceptualized in this way is not opposed to subjectivity
but subsumes (or, embeds) it, and is opposed to objectivity.
Lastly, intersubjectivity in terms of accessibility of information (C)
has already been mentioned as the opposite of subjectivity in
Section ... It applies when “the evidence is known to (or accessible
by) a larger group of people who share the conclusion based on it,” thus
leading to “shared responsibility” (Nuyts a: ). That is, it means
sharedness of information.
With respect to terminology, (C) and Verhagen’s (B) seem to be
more compatible than (A) with the denotation of ‘intersubjectivity’ in
disciplines outside linguistics. For example, Proudfoot and Lacey (:
) define ‘intersubjective’ as something for which “there are ways of
reaching agreement about” and state that it is “usually contrasted with
subjectivity rather than with objectivity, which it may include,” and the
psychologist Tomasello (: , ) seems to understand intersub-
jecivity in terms of ‘sharedness.’ Therefore, the use of ‘intersubjectivity’
for concept (A) is arguably idiosyncratic. However, the concept as such
is useful in its dynamicization for diachronic purposes. Intersubjectifi-
cation in sense (C) does not show a straightforward direction of
language change, as it would automatically entail a decrease in subjec-
tivity. Likewise, concept (B) in Verhagen’s specific sense does not have
a dynamicized diachronic version. Verhagen’s (, ) intersub-
jectivity refers to a consistent presupposition of communication, and as
such is essentially not subject to increase or decrease.
In conclusion, by ‘hearer orientation’ we here refer to (B) in the first
sense (a), that is, the expression of hearer centeredness. It can be taken
to subsume concept (A) of intersubjectivity as a subset. We understand
orientation towards the addressee in a general sense, not limiting it to
the attention to the addressee’s self or image needs as does Traugott
(b, d).
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
What holds for increase in speaker orientation, then also holds for
increase in hearer orientation: hearer orientation is fundamentally a
property of context. However, there are cases in which it may become
associated with linguistic forms and constructions. This can be
reflected in a change of morphosyntactic properties, or in a change in
contexts of use.
Referring to the category of modality again for the purposes of
illustration, the development of may’s use conceding the hearer’s
point of view, as in (), would be a case of a development of hearer
orientation. This use has been entrenched in certain constructions,
such as the concessive construction in (), in which may reflects an
imaginary objection from a non-specific addressee (cf. Narrog b:
–; the examples are from the London-Lund Corpus).
() B: Jake is useless, absolutely bleeding useless (A laughs). He is
feeble, he is weak. He is totally unorganized.
A: Yes, that may be so. But I’m sure he is capable of lifting . . . few
pieces of furniture from the van.
() Whatever one may think of Kenneth Williams, I thought he did a
very good rendering.
Utterance such as () and () presuppose the presence of an addressee
and reflects the speaker’s consideration towards them. To the extent
that this construction emerged at some point in history and then spread
to different uses, this is an example of increasing use in constructions
and contexts associated with hearer orientation.² Another well-known
example is the development of addressee honorifics from referent
honorifics (Traugott and Dasher : ch. ).
4.2.3 Increase in textual orientation
Increase in speaker orientation and hearer orientation are accompanied
by a third major tendency in semantic change of grammatical items
that has received less attention. The development of textual or discourse
or functions is a distinctive tendency that cannot be subsumed under
² According to Visser (: ), the first examples of this use of could date back to the
fourteenth century, as extensions of earlier uses of could.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
speaker- or hearer orientation. Typical examples for the development of
textual functions that do not seem to involve speaker- or hearer orientation
are the development of complex clause constructions out of independent
speech acts. The development of indirect questions, complement clauses
and relative clauses (cf. Herring ; Heine and Kuteva , ch. ;
Kinuhata ) out of interrogatives and concessive conditionals out of
imperatives (cf. Narrog b) are prime examples.
To start with an example from the category of modality again, may
has developed a so-called ‘concessive’ use in a number of constructions.
These constructions include use in as-concessives as in () and in
coordinated constructions as in ().
() I looked at some of my portraits and grotesque as they may be,
they capture some aspects of reality. (Coates : )
() We may have our differences from time to time, but basically we
trust another’s judgement. (Quirk et al. : )
May in () and () is speaker-oriented, in the sense that it expresses an
epistemic judgment of the speaker. It is also hearer-oriented by expres-
sing the attention of the speaker towards the addressee: the speaker
takes into account and admits to an imaginary objection or criticism by
the hearer, before presenting his or her own counter-argument to it.
However, going beyond this strong hearer-oriented component, may
here also has a discourse-building component. It marks a concessive
proposition, and thus creates textual coherence within a series of
propositions associated with different discourse participants. This tex-
tual or discourse function beyond a speaker- and/or a hearer-oriented
function may be rather peripheral with modal verbs, but is a central
function of discourse markers or modal particles (for instance, in
German).
Another example are imperatives forming conditionals or so-called
concessive conditionals in a variety of languages. () is an imperative
clause in English functioning as a conditional protasis. () is an imper-
ative clause in Japanese, a language in which imperatives are morpho-
logically marked, functioning as the condition in a concessive
conditional.
() Finish your homework and I’ll give you some ice cream. (Quirk
et al. : )
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
speaker orientation
(may: acquisition of epistemic meaning)
hearer orientation
(may: conceding to the hearer’s stance)
discourse orientation
(may: integrating parts of discourse)
Figure 4.1 A sequence of semantic changes in grammaticalization
() Modern Japanese (Narrog b: )
Neko=ni s.eyo, inu=ni s.eyo, okor.u
cat= do. dog= do. get.angry.
taimingu=ga muzukasi-i.
timing= difficult.
‘Whether cats or dogs, the right timing for getting angry with
them is difficult.’
In both cases, a clearly hearer-oriented construction has acquired a
textual use. Integrated into conditional and concessive conditional
sentences, the imperative serves to connect two or more propositions
in discourse and indicate semantic relationships between them.
This function can be labeled as ‘textual’ or ‘discourse orientation’
analogous to speaker orientation and hearer orientation. Like the latter
two concepts, it also has an obvious diachronic dimension, that is,
when linguistic forms develop more discourse-oriented meanings. In
the case of may, a historical chain of changes as represented in
Figure . can be assumed (see Narrog b for details). Note that
all these changes took place in an already grammaticalized modal verb,
that is, in secondary grammaticalization.
In the case of the imperatives as conditionals, the chain would start
with ‘hearer orientation.’ The potential sequence of changes is the topic
of Section ...
4.2.4 Discourse orientation in grammaticalization
The question of increase in discourse orientation (and previously
subjectification/intersubjectification) has been mostly raised in the
context of grammaticalization. Nevertheless, it has not been claimed
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
that this type of semantic change and grammaticalization entail each
other. Clearly, an increase in discourse orientation can also take place
in the lexicon. Examples are increase in speaker orientation of gradable
adjectives (Maat ) or of speech act verbs (Kissine ). On the
other hand, the reverse case is less trivial. The idea that grammatical-
ization always accompanies some form of discourse orientation would
not seem unreasonable. Visconti (: ), for example, writes with
respect to the concept of subjectification, “subjectification is pervasive,
as it is motivated by the very subjectivity of the speech event.” Ulti-
mately the involvement of discourse orientation in grammaticalization
is an empirical question, but it is only fair to assume that in a funda-
mental sense, by appropriating lexical items or constructions for the
use of grammar, speaker(s) integrate it into the world of speech act
participants, and ultimately the organization of speech. The creation of
grammar through grammaticalization thus can be seen as intrinsically
involving an increase in discourse orientation. Questions that arise then
are (a) whether there are structural correlates in grammaticalization to
an increase in discourse orientation (Section ...), and (b) the con-
crete relation between the sub-types of discourse orientation and stages
of grammaticalization (Section ...). Furthermore, if the increase in
discourse orientation is regular and unidirectional, the question of coun-
terexamples or counterdirectionality arises. This will be dealt with in
Section ...
4.2.4.1 Structural correlates
One of the most important aspects of semantic change involving
grammar is the question of the reanalysis of scopal relationships.
Arguably, scope increase is also indicative of increased discourse ori-
entation, as it signifies a move from propositional content, concerned
with the description of the event and its participants towards categories
which operate on the propositional content, and are deictic of the
speaker and the speech situation.
The most obvious indicator of the relationship between increased
discourse orientation and scope increase is word order change towards
clause-peripheral, or NP-/VP-peripheral, position. Examples for this
abound in the literature (e.g. Swan ; Traugott and Dasher , ch.
.; Visconti for English adverbs, particles, and connectives;
Adamson ; Davidse et al. ; Breban for the English NP;
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
López-Couso : – for examples from Spanish; and Onodera
and Shinzato for clause-level phenomena in Japanese).
An example of an increase in discourse-orientation in the NP is
provided by Davidse et al. (). According to them, the type nouns
sort of and kind of started out in head noun position, the most right-
ward in the NP-structure, then acquired further leftward positions, as
modifiers, then postdeterminers and nominal qualifiers, and finally as
quantifiers, that is, in primary determiner use, which is ‘textual’ and
‘intersubjective.’ The head noun, the nominal qualifier and the quan-
tifier use are illustrated in examples () to (), which are taken from
Davidse et al. (: –). The b. row in each example shows the
presumptive structure.
() a. A special kind of beer. (head noun)
b. [determiner] + [adjective + type noun/head [of + N]
() a. Some kind of an artist. (nominal qualifier)
b. [qualifying complex determiner] + [determiner][head noun]
() a. All sorts of trouble. (quantifier)
b. [quantifier = primary determiner] + [head noun]
As for leftward position on clause level, in Japanese, a head-final
language, expressions referring to the epistemic state of the speaker,
the hearer or their relationship, are usually found sentence-initially,
sentence-finally, or both, as in example () with the speaker-oriented
elements marked in bold.
() Modern Japanese (Minami : )
Tabun kare=mo k.uru=daroo=ga, sukosi
probably he= come.== a.little
okure.ru=ka=mo sirena.i
be.late.== don’t.know-
‘Probably he will come as well, but he might be a little late.’
However, a maximal degree of speaker orientation or hearer orienta-
tion and wide semantic scope do not necessarily correspond to clause-
peripheral position. This depends on the structural characteristics of a
language. German, a mixed word order language, has a set of intersub-
jective and textual markers, the modal particles, which are often an
expression of the speaker’s evaluation of the relationship between the
proposition and the hearer’s knowledge and expectations. Despite
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
these functions, they are syntactically relatively deeply embedded in the
so-called ‘mid-field’ of the German clause, as shown in ().
() Modern German (Abraham : )
Nur hat er ja alles nötige getan.
has he all necessary done
‘Nevertheless he did everything necessary, didn’t he?’
There are two modal particles here that relate the proposition to
discourse (nur) and emphasize the speaker’s claim of the accuracy of
the statement towards the hearer (ja), respectively. Abraham (:
ch. ), on the basis of co-occurrence restrictions of modal particles and
clause types, argues that they nevertheless occupy a high structural
position (CP position) in the Logical Form of the clause. Similarly,
adverbs in German may line up from left to right in order of decreasing
scope after the V verb, making the narrower-scope adverbs appear
closer to the periphery. Likewise, some of the so-called final particles in
Japanese, that express the speaker’s estimation between their own and
the hearer’s knowledge, also may occur sentence-internally That is,
instead of indicating the relationship between the proposition as a
whole to the speaker’s knowledge, this is done piecemeal fashion,
phrase by phrase, as with the particle ne in example ().
() Modern Japanese (Saigo : )
Ore=mo=ne bosai=wa=ne zitu=wa ussura
I== bonsai== really= faintly
kyoomi=ga at.te . . .
interest= be.
‘Me, too, I actually have a slight interest in bonsai . . . ’
These ‘anomalies’ indicate that not in all languages and under all
circumstances does a high degree of discourse orientation correspond
to clause (or sentence) periphery.
Concurrently to word order, the structural criteria for subjectivity
named in Section .. are also indicators of increased discourse ori-
entation. If a morpheme or construction originally had the potential to
be tensed, negated, modalized, used in the protasis of conditional
clauses, etc., but later loses these possibilities, this development reflects
the loss of non-performative uses and thus increased discourse orien-
tation. This observation is in accord with the assumption that if certain
items or certain categories are frequently used for discourse orientation,
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
and lose descriptive uses, this usage is imprinted on their structural
properties.
Lastly, it is important to note that degrees of grammaticalization
accompanying increased discourse orientation are not straightfor-
wardly comparable across structures and languages. The relation
between increased discourse orientation and grammaticalization only
holds for the history of individual structures and constructions. Not
every grammatical marker that is higher on the scale of discourse
orientation is necessarily more grammaticalized than a marker further
down the scale. It is entirely possible that an item is morphosyntacti-
cally highly grammaticalized, for example as an affix, but nevertheless
has narrow scope and less discourse orientation than a different item
that is morphosyntactically less grammaticalized. While it is possible to
track the individual history of the change of a main verb to an auxiliary,
it is much harder to measure its degree of grammaticalization and
evaluate it against that of an adverb, a mental state predicate or some
other construction. In this manner, it is only possible to hypothesize for
each individual marker or construction developing towards more dis-
course orientation that its development is potentially accompanied by
morphosyntactic grammaticalization. Furthermore, it cannot be
claimed that the most recently developed meaning has to be the most
discourse-oriented of the marker as a whole. It only has to be more
discourse-oriented than the meaning from which it is derived. On the
other hand, if two markers are on the same or a similar path of
morphosyntactic development, the more discourse-oriented one can
be expected to be more grammaticalized. Thus, in English we find that
modals such as must, may, and can are generally more discourse-
oriented than semi-modals such as have to and be able to.³
Beyond structural indications of discourse orientation, in general, an
increase of performative uses as opposed to descriptive uses indicates
increased discourse orientation. In the case of epistemic modal mar-
kers, for example, an increase of uses where the speaker expresses his or
her commitment in the truth of an inference, and in the case of deontic
(volitive) modality, an increase of uses involving the speaker and the
³ Of course it is also possible that the semi-modals of Modern English are not half way
on the path to becoming full modals, but rather form a category with properties on their
own. Krug () argues for such a view.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
hearer, and specifically, the speaker as a source of authority, is then
taken to be indicative of increased discourse orientation.
4.2.4.2 Discourse orientation and stages of grammaticalization
A glance at the subjectification research shows that subjectification in
the sense of (A) in Section .. of the speaker’s commitment and
explicit expression of their attitudes and beliefs has been primarily
identified with an early stage in the development of grammatical
items. Traugott (c: ) herself suggested this when writing that
“subjectification is more likely to occur in primary grammaticalization
(the shift from lexical/constructional to grammatical) than in second-
ary grammaticalization (the development of already grammatical mate-
rial into more grammatical material).” The reason has to be sought in
“the greater accessibility of the original meaning to speakers and
hearers before this has been obscured by frequent reanalysis” (Traugott
b: ). Going even further, Visconti () considers semantic
shift from propositional to non-propositional meaning (that is, in
primary grammaticalization) as definitional for subjectification.
However, the association of speaker orientation with early grammat-
icalization would only be true if we identified it with concept (A)
subjectification. Explicit subjectivity in the terms of speakers’ opinions
and evaluations is naturally more likely expressed if a linguistic item
still has some lexical substance. The association with early grammati-
calization does not hold if speaker orientation is identified with speaker
deixis. Speaker deixis, as a more abstract form of speaker orientation,
actually increases in late stages of grammaticalization, that is, in sec-
ondary grammaticalization. This happens, for example, when non-
epistemic modal verbs become epistemic, labeled as ‘extreme subjecti-
fication’ by Langacker (), or when aspect markers, which are not or
only weakly speaker deictic, eventually evolve into markers of absolute
tense (e.g. Bybee et al. : ; Kuteva et al. : ; Narrog ),
which are speaker deictic practically by definition.
Hearer orientation in the sense of intersubjectivity (A) (see
Section ..), that is, corresponding to intersubjectification in the
narrower Traugottian definition, has been primarily identified with
secondary grammaticalization. Traugott (b: ) hypothesized
that “there cannot be intersubjectification without some degree of
subjectification.” Later, she made preceding subjectification
definitional for intersubjectification in Traugott (c: ) and
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
claimed that “[i]ntersubjectification intersects less extensively with
grammaticalization. In most languages it is grammaticalized only into
some discourse markers and interjections” (d: ).
However, overall the evidence from grammaticalization research so
far is still limited, and there is also no logical necessity for this order of
changes. If hearer orientation in its diachronic dimension is conceived
of sufficiently broadly, and is not based on speaker orientation by
definition, the question of the sequence of changes is still open to
empirical inquiry.
In the case of textual discourse orientation, the evidence is in favor
that, except for the development of discourse markers (Chapter ),
textual discourse orientation is a late stage of development. The list
below shows some examples of developments in support of this
hypothesis (cf. also Section .., and for more details, see Narrog
b).
• As discussed above, for English may the textual function came last
in the modal verb’s development.
• In some languages, such as English, Russian, etc., imperatives, that
is, the most clearly hearer-oriented mood constructions in lan-
guages (see Fortuin and Boogaart ), have assumed a text-
building conditional function (e.g. Make a move and I’ll shoot).
These appear to be very late, if not final developments in the lives
of such constructions.
• Similarly, imperative constructions in some languages have devel-
oped concessive conditional functions, for example in Lithuanian
and Japanese (e.g. Ambrazas ; Narrog b).
• Subordinating markers indicating logical relations between pro-
positions, discourse markers, and final particles are often end
points of chains of grammaticalization. See, for example, Bybee
et al. (: –) for subordinating functions as end points in
the grammaticalization of modal markers, Heine and Kuteva
(: ) for subordinating functions as end points in the
grammaticalization of a variety of categories, and Abraham
() for German modal particles with discourse functions as
the end points in grammaticalization.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
Overall, then, the relationship between discourse orientation with
grammaticalization is an area that has yet to be fully explored, but we
may tentatively conclude the following:
• Increase in speaker orientation in the Traugottian sense of
subjectification as the expression of speaker’s attitudes or evaluations
is primarily associated with early stages of grammaticalization.
• Increase in speaker orientation in terms of speaker deixis is pri-
marily associated with late stages of grammaticalization. Thus,
while the expression of speaker attitudes recedes, speaker deixis
may instead increase in late grammaticalization. This is in keeping
with the overall development towards increasingly abstract mean-
ings in advanced grammaticalization (Section .).
• There is not much reliable evidence yet about where hearer ori-
entation starts, increases, or decreases. The claim for a narrow
concept of hearer orientation (Traugott’s (d) intersubjectifi-
cation) has been that it is associated with late stages, but this is a
definitional rather than empirically based claim. Hearer-oriented
grammatical expressions are found both at an early stage of gram-
maticalization (e.g. addressee honorifics), and in potentially highly
grammaticalized items and constructions (e.g. imperatives).
• Textual discourse orientation is primarily associated with late
stages of grammaticalization (acquisition of additional, more
advanced, grammatical functions).
A fixed order of changes rather than strong tendencies cannot be
established without more extensive evidence. We may assume that
the above-mentioned tendencies are not absolutes. It is also conceivable
that there is some variation in the sequence of changes depending
on the grammatical category. For modality and mood, that is, the
category that provided most examples in this section, textual discourse
orientation appears to be the last stage in the development of gram-
matical markers, following increase in speaker orientation and hearer
orientation. There is also evidence for the possibility of direct hearer
and discourse orientation without a significant stage of speaker orien-
tation. The most likely candidates for this kind of change are so-called
parentheticals or theticals (cf. Chapter ), that is, constructions which
may be co-opted instantaneously for high-level interpersonal or dis-
course functions without running through a gradual semantic and
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
grammatical development. Thus, it is premature and may in the end
even be misleading to posit a single, rigid sequence of changes.
4.2.5 Counterexamples to (inter)subjectification and their status
in discourse orientation
Proponents of (inter)subjectification and related concepts have
generally expressed confidence that this is an important, if not the
most important type of semantic change in grammaticalization. Thus,
Traugott and Dasher (: ) assert that “the main mechanism of
semantic change is subjectification (including intersubjectification).
This follows from the hypothesis that the seeds of semantic change
are to be found in SP/Ws [speakers/writers; a.n.], drawing on and
exploiting pragmatic meanings that arise in negotiated interaction.”
Likewise, Narrog (a: –) hypothesizes that “semantic change
always proceeds in the direction of the same or a higher degree of
speech act orientation.”
Indeed, there is a long trail of case studies across a large range of
typologically diverse languages and theoretical literature that under-
scores the ubiquity of speaker orientation (subjectification), although
work on hearer orientation (intersubjectification) and even more on
textual discourse orientation is still relatively scarce. A list of relevant
literature can be found in “Further reading” at the end of this section.
In contrast, arguments against (inter-)subjectification as an impor-
tant or dominant type of change, or for reverse tendencies of changes in
individual cases, have been few and far between. They include Adam-
son (), who made a case for the de-subjectivization of emphatic
deictic elements in literary prose, and Kranich (a, b), who
suggests that secondary grammaticalization of the English progressive
and other tense–aspect related categories involves de-subjectification.
Most significantly, based on her study of the history of the English
progressive, Kranich (b: ) claims that “secondary grammati-
calization processes (or later stages of grammaticalization) commonly
lead to a type of semantic change that can be labeled objectification.”
Furthermore, presumptive counterexamples to the sequence of changes
in (inter)subjectification have been brought forward. We will discuss
these counterexamples in the light of the concept of discourse orienta-
tion, starting with counterexamples to the overall directionality of
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
change (Section ...), and proceeding to problems with the sequence
of changes (Section ...).
4.2.5.1 Cases of de-subjectification/objectification
In the (inter-)subjectification literature, the general assumption is that
the overwhelming majority of changes conform to the hypothesized
unidirectionality, and counterdirectional developments may exist only
as very limited exceptions. Traugott, the original proponent of (inter-)
subjectification, left the door open for such exceptions, but only under
specific conditions:
Does objectification then occur? Indeed. In semantic change it is the highly conscious
and deliberately interventive sort of change that comes about when ordinary words are
preempted for technical or legal purposes, . . . . But these are cases of “external” “non-
natural” change not of “internal” “natural” change, which is the central concern of
historical cognitive semantics. (Traugott : )
The idea here is that changes towards more subjectivity are natural,
while counterdirectional change is the result of an artificial manipula-
tion of language material. Notwithstanding, some scholars have argued
for “naturally occuring” counterexamples to subjectification.
Two cases proposed in the literature involve question markers. In a
study on the grammaticalization of historical questions in Tamil,
Herring () proposed a chain of grammaticalization from question
markers to text cohesion markers and clause-linking markers, that is,
relativizers and adverbial conjunctions.
Example () shows the question marker -ē indicating a rhetorical
tag question in narration, and as a relative clause marker in (). The
latter is hypothesized to be derived from the former.
() Tamil (Herring : )
Inta aracan kannai mutikkittiruntan-ē
this king eye. close...-
‘The king was keeping his eyes closed, right?’
() Tamil (Herring : )
Nan poy aval ninriruntal-e anta itattil . . .
I go she stand...- that place.
‘I went and stood in the place where she had stood . . . ’
Herring’s () analysis is not strictly based on historical data, but we
are not aware of research questioning the plausibility of this scenario
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
of change. According to the author, the original development of a tag
question marker is a case of subjectification (in our terms, we would
add that it also crucially or even primarily involves hearer orientation),
while further development towards a relativizer is the opposite. She
speaks of “pragmatic unmarking” and “gradual diminishing of [the
markers’] expressive impact” (Herring : –). While the ques-
tion markers are originally used “subjectively” for rhetorical questions,
they end up as grammatical markers of conjunction and subordination
devoid of expressiveness. Herring () concludes that,
[t]hus the history of the conjunctions . . . and the relativizer may be said to involve both
subjectification—in the original extension . . . from true to rhetorical questions—and
de-subjectification, in the grammaticalization of pragmatic devices as autonomous
clause-linking elements. (Herring : )
In Japanese, a similar development has taken place. In this case, the
language is historically well-documented. Japanese originally had no
indirect question markers. The direct question marker ka developed
into an indirect one from the fifteenth century on. Kinuhata ()
sketches this development as follows. () shows an example of the
direct question use which is already documented in the early stage
(eighth century), while example () shows an example of the indirect
question use from Modern Japanese.
() Old Japanese (Kinuhata : )
Wa=ga puru sode=wo imo mi-tu=ramu=ka
I= wave sleeve= wife see-==
‘Did my wife see me wave my sleeves?’
() Modern Japanese (Kinuhata : )
Watasi=wa dare=ga paatii=ni ki.ta=ka
I= who= party= come.=
sir-ana.i
know-.
‘I don’t know who came to the party’
According to Kinuhata (), indirect questions are not only seman-
tically less expressive of the speaker’s doubt (subjectivity), but also
syntactic properties of subjectivity are lost, since the indirect questions
are now embedded in past tense, negation, etc. He concludes that the
development of the indirect question function “constitute[s] the
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
counterevidence to subjectification” and labels this as a case of ‘syntac-
ticization’ (Kinuhata : –).
Last but not least, Kranich (a, b) has argued for interpreting
the historical development of the English progressive in Early and Late
Modern English as a case of de-subjectification, counter to previous
research that had presented it as a paradigm case of subjectification
(especially Fitzmaurice in Wright ; Fitzmaurice a). The main
reason is that in the decisive period of grammaticalization of the
English progressive to a common aspect marker in the second half of
the eighteenth century, subjective uses significantly decline in number.
Interestingly, though, they rose again from the nineteenth century on,
when the aspectual uses were firmly established (cf. Kranich a: chs
., .). Specifically, certain types of subjective use, evaluative and
emphatic ones, as exemplified in (), declined, while uses as in exam-
ple (), which Kranich (a) labels as ‘interpretative,’ have been on
the rise.
() They were now able to speak to each other and consult. That
Louisa must remain where she was, however distressing to her
friends to be involving the Harvilles in such trouble, did not
admit a doubt. Her removal was impossible. (archer\-.
bre\aust.f) (Kranich a: )
() HARRY. Why, it is possible you may yet receive a valentine.
SOPHIA. Nay, now, but don’t you go to think that I am asking
for one; for that would be very wrong of me, and I know better.
(archer\-.bre\holc.d) (Kranich a: )
Thus, Kranich (a: ) observes that “the rise in the subjective uses
that one sees in the th and th centuries only affects a very specific
meaning of the progressive,” namely the interpretative function. In her
view, this function “was built on the aspectual use” (Kranich a:
). As for the temporary decline of subjective uses of the progressive,
Kranich suggests that
. . . in order for a form to be recruited for the expression of subjective meaning, the
speaker has to be free to decide whether or not to use the marker in question. . . . Now if
the use of a particular marker is felt to be appropriate (or even obligatory) in a
particular grammatical context, regardless of speaker involvement, attitude, or desire
for emphasis, then such subjective meanings can no longer be associated with this
marker in such a context. (Kranich a: )
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
This is a quite plausible explanation. If a grammatical marker becomes
extremely frequent and obligatory in certain contexts, it obviously
cannot retain its ‘subjective expressiveness.’ At each instance of use
the speaker would be forced to be ‘subjective’ if intended or not. One
can also put it reversely in more traditional grammaticalization terms,
namely that the expressiveness of a marker will ‘wear off ’ or ‘bleach off ’
through very frequent use. While the English progressive developed
new subjective uses instead, this is not necessarily the case, as Kranich
(b) argues. It seems that in the course of the development of tense–
aspect markers in other European languages as well, expressive subjec-
tive uses simply wore off, without being “replaced.”
The first counterexample to directionality in (inter)subjectification,
namely the development of clause-integrating usage of question mar-
kers, is related to the development from subjective and intersubjective
to markers of clause integration, entailing loss of subjectivity and
intersubjectivity. This actually fits well with the framework of increase
in discourse orientation presented here. Expressive speaker-oriented
and hearer-oriented uses give way to text-/discourse-oriented uses. As
stated above, it seems quite likely that textual discourse orientation is
very often a late or the last stage of discourse orientation. This is also in
line with the overall tendency of ‘abstraction’ in grammaticalization
(Section .). Meanings and functions that operate primarily or purely
for the organization of grammar and text are more abstract than earlier
stage speaker-oriented and hearer-oriented expressive meanings.
Likewise, it is no surprise to see expressive and evaluative subjective
uses of aspect markers to be primarily associated with early stages of
grammaticalization and then receding when the aspect marker becomes
an obligatory part of grammar. As we know from Bybee et al. () and
other research on grammaticalization, in a broad perspective, aspect
markers are on a trajectory towards tense marking, and thus towards
speaker deixis. That is, they shift from a more concrete and expressive
type of speaker orientation to a more abstract one.
The labeling of part of these processes as ‘de-subjectification’ is
correct in the sense of subjectivity (A) in Section ..: Expressive
subjective meanings decrease or are lost. However, the term ‘objectifi-
cation’ would be misleading since we do not see an increase in, or
return to objective description of the external world. On the contrary,
the meanings of the grammaticalized items increasingly become part of
the abstract internalized domains of grammar and speech.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
4.2.5.2 Problems with the sequence of changes
The issue here is again the mutual order of the three tendencies,
namely speaker orientation, hearer orientation, and textual discourse
orientation, as they have been discussed in the literature. We will
start here with the question of the relative position of textual orientation
vis-à-vis speaker orientation and hearer orientation since relevant ex-
amples were already presented in the preceding paragraphs. In the case
of ‘concessive’ may (cf. Section ..), we first find the development of
speaker-oriented epistemic uses, then uses that imply a concession, that
is, are hearer-oriented, and finally clearly marked concessive construc-
tions (cf. Narrog b: –). In the case of the imperatives gram-
maticalizing into conditional or concessive conditional protases, it is
clear that the hearer-oriented use formed the basis for the extension to
the text-oriented uses.
With respect to the development of question markers into textual
markers presented in Section ..., we find the same tendency. The
Japanese direct question marker, which has a clear hearer-oriented
dimension of meaning, historically unambiguously precedes the text-
and discourse-organizing indirect question use. With respect to the
Tamil rhetorical question markers developing various subordinating
functions, Herring () depicts the following scenario. The rhetorical
questions are “a means of engaging the attention of the interlocutor.”
For that purpose, “the narrator constructs a hypothetical listener with
whom he ‘interacts’, even speaking at times in this other listener’s
‘voice’” (Herring : ). In our current understanding, we would
identify such uses as hearer-oriented. From this, according to Herring
(), the ‘discourse-organizational’ and ‘textual’ function developed,
that is, textual orientation followed and eventually “replaced” hearer-
oriented uses (Herring : ). Note that Herring () labels the
hearer oriented uses as “expressive” and “subjective” because this was
the only available terminology at that point. Traugott (b: )
commented on Herring () that “these changes . . . do evidence a
shift from attention directed primarily to the hearer toward more
internally oriented, more solely speaker-based text organisation.” This
comment is interesting because Traugott seems to admit here a shift
from the intersubjective (which was not yet distinguished from the
subjective) to the textual or the subjective (if the textual is included in
the subjective). This leads to the next question, namely the ordering
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
. DISCOURSE ORIENTATION
between the speaker oriented and hearer oriented, or “subjective” and
“intersubjective” in previous terminology.
We will not deal here in any detail with alleged counterexamples that
are based on concept (C) of intersubjectivity, namely shared evidence
based on information accessible to both speaker and hearer. Intersub-
jectivity in this sense is close to objectivity. We do not expect “inter-
subjectification” in this sense to be a relevant diachronic tendency.
Quite the contrary, we would expect intersubjective meanings to give
way to subjective ones. Unsurprisingly, this has actually been shown in
studies such as Cornillie (, , ) or Ghesquière (, ).
They are therefore not examples for a different sequence of changes or
counterdirectionality. A more detailed discussion of these cases is
provided in Narrog (: –).
If we try to identify possible counterdirectional developments from
intersubjective to subjective vis-à-vis the dominant (A) concept, what
we do find is the development from intersubjective to textual-oriented
markers outlined above. This is due to the fact that in the (A) concept
the textual uses are not acknowledged as such but instead labeled as
‘subjective.’ The only other case that we are aware of relate to certain
parentheticals, that is, constructions that can be used instantaneously
(without undergoing much grammaticalization) for metatextual gram-
matical and discourse functions as illustrated in () and () (the
examples are from Kaltenböck et al. : ).
() Mary—don’t forget—is coming over to visit.
() Because John is, you say, a spy, we should be careful what we say
to him.
Don’t forget in () and you say in () are apparently not grammati-
calized but used instantaneously. Nevertheless, these phrases immedi-
ately assume intersubjective ((), ()) and evidential () functions.
It is thus reasonable to assume that also more established expressions
such as you know are intrinsically intersubjective, and do not need to go
through a phase of subjectification. Fitzmaurice (a) tried to pre-
empt arguments against the hypothesis that subjectification necessarily
precedes intersubjectification by claiming that second-person you know
(intersubjective) must be derived from first-person I know (subjective).
This is historically not really clear but cannot be excluded either.
Avoiding this problem, Brinton () chose to study a first-person
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DIRECTIONALITY OF SEMANTIC CHANGE IN GRAMMATICALIZATION
intersubjective parenthetical instead, namely, I mean. Based on the
observation of the historical data, she stated that “in the case of I
mean one cannot show that subjective meanings clearly precede inter-
subjective meanings” (Brinton : ). Traugott (c: ) argued
against this, suggesting that “parenthetical I mean has always been
pragmatically intersubjective. Over time it has been used more inter-
subjectively to express emphasis and assertion of the veracity of an
utterance . . . but it has not been intersubjectified, except in fixed
phrases like You know what I mean?.”
In terms of the concept of discourse orientation, none of these
sequences of changes presents a challenge or a counterexample to this
concept. First of all, we do not claim that there is one fixed order of
changes. We merely observed that a change from speaker-oriented to
hearer-oriented and finally textual-oriented meanings is most frequent
and most likely. Herring’s () case confirms this tendency. We also
admit the possibility that there may be hearer- and discourse-oriented
meanings without preceding speaker orientation. The development of
discourse markers broached here could be a good example for such a
development (see Chapter ).
4.2.6 Conclusion
As argued in this chapter, an increase in discourse orientation may be
fundamental for most if not all grammaticalization. Speakers appropri-
ate words from the lexicon to express meanings and functions relating
to the speech act participants, and ultimately to organize speech and
text itself. In most cases this goes hand in hand with increasing
abstraction (Section .). It is only natural therefore if expressive
speaker-oriented and hearer-oriented meanings and functions eventu-
ally give way to speaker and hearer deictic functions and finally to
functions that purely organize speech itself. The concept of discourse
orientation accounts for these changes better than ‘subjectification’ and
‘intersubjectification,’ which only cover partial developments. The term
‘objectification’ for the last, very abstract stages of grammaticalization
does not seem appropriate, since these stages do not represent a return
to meanings that objectively describe an external world, but anchor
linguistic expressions to the extreme in an internalized world of speech.
OUP CORRECTED PROOF – FINAL, 25/1/2021, SPi
DISCUSSION POINTS
Further reading
Besides the two quasi-monographic studies (Traugott and Dasher ;
Narrog a) and other research already cited above, there is a large number
of book-length paper collections dedicated to the topic. These include Stein
and Wright (eds.) (), Athanasiadou et al. (eds.) (), Cornillie and
Delbeque (eds.) (), Onodera and Suzuki (eds.) (), Davidse et al.
(eds.) (), Devos and van der Wal (eds.) (), van der Auwera and
Nuyts (eds.) (), Brems et al. (eds.) (), and van Olmen et al. ().
A monograph on intersubjectification by Shindo () deals with intersub-
jectification in the lexicon.
Other notable contributions not yet cited outside these paper collections
include Visconti () on the subjectification of conditionals in Germanic
and Romance languages, Fitzmaurice (b) on subjective discourse
markers eventually becoming intersubjective, Aaron and Torres Cacoullos
() on quantifying subjectification of a Spanish counter-expectation
marker, Breban () on subjectification of adjectives in English NPs,
López-Couso () with a theoretical consideration on the concepts of
subjectification and intersubjectification, Rhee (, , ) and Sohn
() on (inter)subjectification of sentence-final particles, discourse mar-
kers, and reportatives in Korean, Xing () and Chor () with studies
on (inter)subjectification in Sinitic languages, Evers-Vermeul et al. () on
the subjectification of Dutch and French causal connectives, Breban and
Davidse () on the (inter)subjectification of very, and Bergqvist (),
and Tantucci (, ) with theoretical considerations on the concept of
intersubjectification: Bergqvist () disusses the requirements for a lin-
guistic expression to become intersubjective, while Tantucci (, )
proposes an extended concept of intersubjectification, in which an assumed
third party has an indirect social bearing on the utterance.
Discussion points
() Describe the concepts bleaching, generalization (of meaning), and
abstraction with examples from a language that you know well. Which
concept is most suitable on the basis of your analysis?
() How do the terms discourse orientation and (inter)subjectification differ in
the way they capture meaning change in grammaticalization?
() Why do you think is speaker orientation thought to be more common
than hearer orientation? Do you think that hearer orientation presupposes
speaker orientation?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
5
Steps and phases of
grammaticalization
5.1 Gradualness
5.2 Stages
5.3 Grammaticalization chains
5.4 Grammaticalization in semantic maps
5.5 Polysemy, transcategoriality, and heterosemy
5.6 Cycles
Discussion points
In this chapter, we will explore a number of ‘nuts and bolts’ issues about
how grammaticalization proceeds: Is grammaticalization sudden or
gradual, in what kind of stages does it advance, and do the same
grammaticalizations recur again and again in the same language?
We start out with the question of gradualness (or non-gradualness)
of grammaticalization (Section .), continue with stages of grammat-
icalization (Section .), then with so-called ‘grammaticalization
chains’ (Section .), The concept of ‘semantic maps’ is analyzed in
Section ., while Section . deals with three technical terms that are
widely used in the literature on grammaticalization. Finally, we discuss
cycles of grammaticalization in Section ..
5.1 Gradualness
To begin with, we need to emphasize that whether or not grammati-
calization is a gradual process is to some extent a matter of the
perspective adopted and the model employed to describe the process.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRADUALNESS
For example, one researcher may analyze the process as consisting of a
sequence of distinct acts of innovation, to be described in terms of a
‘scale’ (Lehmann []: ). Another researcher may prefer to
look at the same process from the bird’s eye perspective of the historical
linguist interested in long-term effects of grammatical change. For him
or her it may be best described and accounted for in terms of a
‘continuum’ (Heine and Reh : ), a ‘chain’ (Claudi and Heine
; Craig : –; Heine et al. : –; Heine ;), or a
‘cline’ (Hopper and Traugott : –; see Section .). Naturally,
readers may want to choose the perspective that is best compatible with
their own approach to the study of language change.
‘Gradual’ change, as in the title of this section, refers to “change that
occurs via discrete micro-steps” (Luraghi and Bubenik : ).
Lehmann (: ) provides a mathematical formula, in which
‘gradual’ means that “given more than two elements E, E . . . En, Ei-
is to Ei as Ei is to Ei+” (where “E” stands for “element”). Of course, this
is meant as an approximation, and not literally in the sense that the
distance between each set of two functions must be exactly the same.
There are three logical possibilities with respect to the gradualness of
grammaticalization as a process, the first of them, (A), is that it is
gradual as just defined. The two others, (B) and (C), are the polar
opposites on a scale with ‘gradualness’ in the middle.
A Grammaticalization is ‘gradual’, that is, it proceeds step by step.
B Grammaticalization is ‘abrupt’, that is, it involves a saltation from
source to target category. In Lehmann’s () formula this would
mean that there are steps as in (A) but these steps are maximally large.
C Grammaticalization is ‘continuous’ or ‘gradient’, that is, proceeds
in an indistinguishable continuum. In Lehmann’s formula, this
means that “whichever two adjacent variants Ei and Ej are selected,
there is always a variant Ek such that Ei > Ek > Ej is gradual . . . ”
(Lehmann : ).
Note that the question that we are concerned with here is how change
proceeds within the grammar of a language and its speakers, and not the
spread within a community of speakers. The latter is necessarily gradual
in a natural language setting. The gradual diffusion of language changes
within communities of speakers is also well-documented in sociolinguis-
tics (e.g. Labov ), and should be considered uncontroversial.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
As for grammaticalization in the grammar of a speaker, or a
language, each of these possibilities has been claimed as a hypothesis
in previous research. However, (A) has been the majority position in
grammaticalization studies. It is also the position adopted in this book.
We will therefore briefly present positions (B) and (C) and state the
reasons why we think hypothesis (A) should be preferred.
The idea of grammatical change in general and grammaticalization
in particular being ‘abrupt’ or ‘catastrophic’, cf. (B), is mainly associated
with formal studies of syntactic change (e.g. Lightfoot , ;
Roberts and Roussau ). It is based on (a) a model of grammar
that requires rigid correspondence between linguistic forms and syn-
tactic positions, and (b) the idea that grammaticalization takes place
during first language acquisition: For example, in a Principles and
Parameters model, children set the parameters of their language,
which may be different from the parameter setting of their parent
generation. This leads to change (Lightfoot ).
The idea that grammaticalization is a gradient in accordance with
(C), in contrast, may be latent in some functional research on gram-
maticalization, but has been argued for most explicitly in Lightfoot
(), who suggests that “language change operates in a fairly con-
sistent manner, owing to human perception and cognition, working
relatively steadily along the lines of similarity and contiguity” (Lightfoot
(: ).
Stance (B) is dictated by its theoretical premises and has been
criticized for not being compatible with the gradualness observed
empirically in synchronic and diachronic language data, which exhibit
gradience and gradualness. Furthermore, it has been claimed that
language change can also be described in a model that is based on
gradual change, and thus is more compatible with fine-grained empir-
ical data (e.g. Heine et al. : –; DeLancey , ; Hopper
and Traugott : –). In fact, in formal grammar itself, there has
been a movement to model grammaticalization in a way that allows
for gradualness and better accounts for empirical data (van Gelderen
, c).
The main problem with stance (C) is that while gradience may be a
well-observed phenomenon in synchronic language data, and may
indeed have psychological reality, it is not a practical heuristic for the
study of grammatical change. As Lehmann (: ) notes, “if
differences between adjacent variants become infinitely small, no
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRADUALNESS
grades may be discerned any longer.” In order to meaningfully describe
change, we need to assume steps with some degree of distinction. With
historical data, essentially it is only possible to show steps, for example
the extension of a construction to a new set of verbs, or with a new tense
form, etc.; it is not possible to show developments that cannot be
pinned down as steps in the data. Another question is to what extent
the concept of gradience is empirically valid in language change. Even
some functional linguists suggest that change may proceed in distinct
steps for the individual speaker, even if those steps are very small (e.g.
Hopper and Traugott : ; Traugott and Trousdale : –).
In the rest of this section we provide a few examples of gradualness in
the development of grammatical categories.
The first example involves the Swedish verb sitta ‘sit,’ which is used
as a progressive in a pseudo-coordinative construction. This is illus-
trated in ():
() Swedish (Hilpert and Koops : )
Vi bara satt och pratade.
we just sat and talked
‘We were just talking.’
According to Hilpert and Koops (), whose description we follow,
the construction was already available in the oldest texts -(fourteenth
century). () is an example.
() Swedish (Hilpert and Koops : )
Ther sato nokre kompana oc drukko oc lifdho i
There sat some friends and drank and lived in
ofwerflødhlikheth.
abundance
‘There sat some friends and drank and lived in abundance.’
However, all the examples in the oldest texts “express actions that are
either conventionally carried out in a sitting posture, such as eating or
drinking, or are at least fully compatible with a sitting posture, such as
talking” (Hilpert and Koops : ). But sitta could already be used
in temporally intersecting events, cf. ().
() Swedish (Hilpert and Koops : )
Tha han kom til qwinnan oc sat och taladhe
when he came to woman and sat and talked
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
med henne, tha frestade direfwllin honom
with her then tempted devil him
‘When he came to the woman and sat and talked to her, the devil
tempted him . . . ’; alternatively ‘When he came to the woman and
was talking to her, the devil tempted him.’
Providing the temporal background of a main clause event is, according
to Hilpert and Koops (), a typical progressive and imperfective
function. Furthermore, the verb sitta in this construction was used
strikingly often without a locative phrase that indicated the place of
sitting, unlike in lexical use outside of this construction. So, we may
assume that grammaticalization was already incipient and speakers
at that time were already able to interpret sitta in the sense of
ongoingness.
In a corpus study from the oldest text to Modern Swedish, Hilpert
and Koops (: –) then show that the following features of a
grammaticalized aspect construction gradually increased in frequency,
in comparison to lexical uses outside the construction:
(a) Use without specification of location of sitting: In Modern
Swedish, this constitutes the clear majority of uses of sitta in
the pseudo-coordinating construction, in contrast to lexical sitta.
(b) Temporal adverbials have become increasingly located outside the
construction, indicating greater coherence of the construction.
(c) Extraction of objects of the second verb, which is generally not
possible in genuine coordinated constructions, is becoming
increasingly more frequent.
Note that with respect to all these criteria, a gradual increase in
numbers is observed. If anything, Hilpert and Koops () provide
an example for the continuity of change. However, at some point,
speakers of Swedish acquired an interpretation of sitta as ongoingness
instead of physical sitting, and without the assumption of such a step,
the description of the grammaticalization would be incomplete.
The second example is about the development of the English prep-
osition thanks to. According to Ahn (, ), the preposition
developed from the noun thanks (pl.) and the preposition to in the
following steps:
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRADUALNESS
(i) Semantic change of the lexical noun thank ‘thought’ (thirteenth
century) to ‘favorable thought/favor,’ then ‘gratitude,’ and then
‘thanks being given to’ (seventeenth century).
(ii) At this stage of the semantic development, thanks came to be
used in the construction [[be thanks] to someone].
(iii) It became possible to delete be in this construction: The struc-
ture now was [[thanks] [to (someone)]].
(iv) The construction was rebracketed as [[thanks to] (someone)],
and the interpretation was ‘indebted to someone’ (eighteenth
century).
(v) In the nineteenth century thanks to further acquired the mean-
ing ‘in consequence of,’ whereby the favor could also be attrib-
uted to an inanimate entity, and in Present-Day English, finally,
ironic uses evolved where it does not refer to a favorable
situation.
According to Ahn (: ), the semantic change preceded and
enabled the syntactic changes. The decisive step was the rebracketing
in the eighteenth century, but this was preceded by a number of
necessary steps, and then followed by further steps in the development.
Another example is provided by the Modern French preposition chez
‘at (with human referents),’ which has developed from the Latin noun
casa ‘house’ (Old French (en) chies or chiése). The development was
enabled by a number of conditions and steps, which we describe here
from (i) to (vi), following Harrison and Ashby ():
(i) While Latin had free word order, in Vulgar Latin and early
Romance, the dependent genitive, in this case the owner, came
to follow the noun, in this case chiése ‘house.’
(ii) In Old French, with human possessors, a preposition did not
have to precede the possessor, i.e. it did not have to intervene
between chiése ‘house’ and the possessor.
(iii) Between the ninth and the eleventh century, chiése became
obsolete as an independent noun with the meaning ‘house.’
Other nouns, such as maison, became prevalent in this meaning.
(iv) In parallel, chiése acquired the ability to indicate a location
without being preceded by a preposition of location (en or a).
(v) However, chiése continued to be used with a following possessor
in the meaning ‘house of.’ Harrison and Ashby (: )
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
consider use with professions instead of buildings, e.g. chiés un
pestor ‘house of a miller/at a miller’s,’ as the first clear preposi-
tional use. There is a metonymic connection, since professionals
were associated with the buildings in which they resided.
(vi) In later periods, chiése came to be used with a group of people
entirely dissociated from buildings, e.g. chez les perses ‘among
the Persians,’ chez ma mère ‘at/of . . . my mother,’ and finally
with works of literature and art chez Descartes ‘in Descartes’
writing.’
There is one decisive step in this chain, namely the emergence of the
possibility to indicate a location without being preceded by a locative
preposition. This was made possible by the loss of the use as an inde-
pendent noun: chiése could thus specialize in the possessive use indicat-
ing someone’s location. Still, there is no saltation since chiése retained its
lexical meaning. So, while Harrison and Ashby () do not present
corpus data, we may expect that in the transitional period in Old French
there was some measure of ambiguity between chiése as simply ‘the
house of ’ and ‘at the house of.’ Furthermore, semantic generalization
proceeded fairly slowly. Even in Modern French, chez seems to be still
restricted to use with human beings (and their products) in its preposi-
tional use. On the other hand, it is also not particularly insightful to say
that there was a continuous change, since we need to pin down steps in
the development to describe it meaningfully.
5.2 Stages
If the steps in the grammaticalization of individual items are looked at
from a larger perspective and generalized across instances, we can
identify larger steps, that is, ‘stages’ of grammaticalization from a specific
source to a specific target. The idea is that with certain sources and targets,
there is no direct change from source to target, but clearly discernible
steps in between, the ‘stages.’ An example was already mentioned in
Section ., namely the cross-linguistically observed stages in the devel-
opment of indefinite articles from the numeral ‘one.’
At stage I, the numeral is just a numeral and has no functions beyond
that. This is, for example the case in Swahili, where the numeral moja
‘one’ (ex. ()) has no grammatical function.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. STAGES
() Swahili (Heine a: )
Ni-na gari moja.
I-have car one
‘I have one car.’
At stage II, the numeral is able to function as a presentative marker,
introducing a new discourse referent, but has no further grammatical
function. This stage is ascribed to the Russian numeral odin ‘one’, as in
example ().
() Russian (Heine a: )
Zhyl da byl odin starik . . .
lived was one old.man
‘Once upon a time there was an old man . . . ’
At stage III, the numeral can function as a ‘specific marker,’ marking
any participant in discourse known to the speaker but presumed to be
unknown to the hearer. This stage can be attributed to Street Hebrew
exad, as in example ().
() Street Hebrew (Heine a: )
Ba hena ish-xad etmol ve-hitxil le-daber
came here man-one yesterday and-started to-talk
ve-hu . . .
and-he
‘A man came in yesterday and started talking and he . . . ’
Next, the numeral may become a nonspecific marker, marking a
participant whose referential identity neither the hearer nor the speaker
knows (ex. (); stage IV). The marker is still restricted to singular
reference. This is the stage into which the indefinite articles in English
and many other western European languages have developed.
() Buy me a bottle of water, please!
At the last stage, V, the erstwhile numeral ‘one’ can be used for
practically all kinds of nouns, even mass or plural nouns, although
there may be language-specific restrictions. Spanish and Italian, for
example, have reached this stage. In () (cited as ex. () in
Chapter ), the indefinite article derived from the numeral ‘one’ has
evolved to mark even plural nouns.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
() Italian (Bradley and Mackenzie : )
Unas vacaciones en Italia.
vacations in Italy
‘A holiday in Italy.’
One more example for ‘stages’ of grammaticalization is the
grammaticalizaton of posture (sit–stand–lie) auxiliaries in many
languages of the world as presented by Kuteva (, ). At stage I,
the verbs simply mark bodily posture of animate beings. Especially
relevant for the further development of the verbs is their use in a
construction where the clause with a posture verb is followed by a second
clause referring to a simultaneous event, as in () from Bulgarian.
() Bulgarian (Kuteva : )
Ana sedi na divana i piše pismo
Ana sit.. on couch.the and write.. letter
a bašta í sviri na piano.
whereas father her play.. on piano
‘Ana is sitting on the couch and is writing a letter whereas her
father is playing the piano.’
The decisive feature of stage II is that now the position is not restricted
to animate beings but is extended to inanimates. () exemplifies the
same biclausal construction in Bulgarian as (), but this time with an
inanimate subject.
() Bulgarian (Kuteva : )
Drexite sedjat v koridora i sǎbirat
clothes.the sit.. in corridor and gather..
prax.
dust
‘The clothes are in the corridor and gather dust.’
At this stage, a reinterpretation of ‘sit’ to merely indicate the continu-
ation of the event in the following clause is especially prone to taking
place because the verb ‘sit’ has lost its concrete, physical meaning when
used with inanimate subjects. Thus, () may be subject to reinterpre-
tation as ‘the clothes in the corridor are gathering dust.’ This reinter-
pretation (stage III) is confirmed in sentence structure when adverbials
do not follow the posture verb anymore but the whole verbal complex,
as in () (note that the verb here is ‘lie’, and not ‘sit’ as in the previous
examples).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION CHAINS
() Bulgarian (Kuteva : )
Trionǎt leži i rǎždjasva v mazeto
saw.the lie.. and get.rusty.. in cellar.the
‘The saw is getting rusty in the cellar.’
Lastly, at stage IV, the construction is extended to animate subjects,
where the meaning ‘sit’/‘lie’/’stand’ also gets replaced with a progressive
reading as in (). This represents the stage of conventionalization
discussed in Section ..
() Bulgarian (Kuteva : )
Sedi i se oplakva vmesto da
sit.. and complain.. instead to
se xvane za rabota.
take.. for work
‘S/he is complaining all the time instead of starting to work.’
5.3 Grammaticalization chains
In adopting a yet more general perspective on consecutive steps in
grammaticalization, one can go beyond the ‘stages’ in Section ., and
identify ‘grammaticalization chains.’
‘Grammaticalization chain’ (in short, ‘chain’) is a concept that has
been used to describe the interlocking pattern of transition from lexical
or other less grammaticalized expressions to more strongly grammati-
calized expressions (Heine ).
An example is provided by Moser (), who identified the com-
plex grammaticalization chain of the ‘give’ verb in the Nilo-Saharan
language Kabba, rendered in Figure ..
Since the language has no significant historical records, the recon-
struction is based on synchronic variation. Strictly speaking, Figure .
kàre BENEFACTIVE RECIPIENT
‘give’ (2 arguments) (1 argument)
m-ar-i k-àr-i
CAUSATIVE PURPOSIVE/ COMPLEMENTIZER SEQUENTIAL CONNECTIVE
(1 argument) RESULTATIVE (0 arguments) (0 arguments) (0 arguments)
kàr-em (0 arguments) ta kàre tà àre tà
ta kàre (tà)
Figure 5.1 Grammaticalization chain of ‘give’ in Kabba (Moser : )
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
presents two chains, which can also be conceptualized as one large
chain. The starting point is the ditransitive (-argument) verb kàre
‘give.’ Note that in various inflectional forms, the initial k- can be
dropped, and syntactically, not all arguments have to be realized in
the surface structure.
In the upper chain in Figure ., kàre first developed into a benefac-
tive marker with two arguments. The arguments are realized as prefixes
and suffixes on the verb stem.
() Kabba (Moser : )
M-ínga dèné m-ar-έ
-found wife -give-
‘I found a wife for him.’
In a literal reading, () might be translated as ‘I found a wife and I gave
[her] to him.’ However, according to Moser (), semantically ‘give’
is already a benefactive and does not imply literal ‘giving.’ The next step
to ‘recipient’ is represented in example ().
() Kabba (Moser : )
N-áw à pà tàr kàrə Dómìnic
-go to tell story give Dominic
‘He is going to tell the story to Dominic’
As a marker of ‘recipient,’ kàre may still have person endings. However,
after specific verbs, such as the one for ‘tell’ in (), it can even appear
in uninflected form. Note that ‘benefactive’ and ‘recipient’ are both
dative functions, and Moser () glosses both in () and () simply
as ‘dative.’
The second chain starts with ‘causative’ function and goes all the way
to a ‘connective.’ We will only provide examples for the first two
functions on this chain here. () illustrates a causative, and () a
purposive. We retain ‘give’ as a gloss in each case, while Moser ()
glosses each instance with the function that it instantiates, emphasizing
that she considers kàre as fully grammaticalized in each of its functions.
¹ Note that in the of Figure . (m-ar-i), the prefix is for subject and
the suffix for .
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION CHAINS
() Kabba (Moser : )
M-á kàr-έ n-óso
- give- -fall
‘I will make him fall.’
() Kabba (Moser : )
Ń-ddɔ pùrù dɔ-dé té tà kàrə ń-bbeél
-shot fire head- give -fear
‘They fired at them to make them afraid.’
Now, decisively for the concept of a chain, each following function on
the chain should be ‘interlocking’ with the previous one, that is, we
should expect some overlap and instances of use that exhibit ambiguity
between two adjacent functions. For example, ambiguity between
‘causative’ function as in () and the ‘purposive’ function as in ()
can be found in an example such as (), which could be interpreted
both as ‘causative’ and as ‘purposive.’
() Kabba (Moser : )
Ń-pà bè mbatà kùwà né gɔ kərə kàrə
-spoke like seize his foot elephant give
tà sánge màre dèw ke kété
search other person first
‘He spoke like this to excuse himself to the elephant so that he
(the elephant) could find someone else (to shave his head).’
According to Heine (, a) grammaticalization chains have the
following properties:
() Properties of grammaticalization chains
a. They are linear.
b. They are directional, extending from least to most strongly
grammaticalized expression.
c. They have an interlocking structure of the form [A–A/B–B–B/
C . . . Z], where each member of the chain, i.e. function, shares
at least one feature with adjacent members (functions).
d. They are the result of a gradual process leading from least to
most strongly grammaticalized expression.
e. They have both a synchronic and a diachronic dimension:
Being the result of a historical process, this process is likely to
be reflected in the synchronic structure of a language.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
f. They are continuous structures in the sense that there is no
general answer to the question of how they should be cut up
into segments, or whether they contain focal points or pro-
totypical instances.
We assume chains to be a ubiquitous phenomenon, relating to many
different aspects of grammatical change. Practically any small and
locally restricted sequences of changes as discussed in Section . or
the preceding section of this chapter can be conceptualized as a chain.
A minimal example is given in (), which is based on the data in (),
repeated from Section ...
() Chain of meaning change of English since from temporal to
causal conjunction (Traugott and König : –; see
Section .. above):
Temporal > temporal/causal > causal
() The evolution of English since (based on Traugott and König
: –)
Constructed examples Meaning
a. I have done quite a bit of writing since we last met.
Temporal
b. Since Susan left him, John has been very miserable.
Temporal,
causal
c. Since you are not coming with me, I will have to Causal
go alone.
On the other hand, there are what one may call ‘macro-chains,’ ex-
tending over a range of more general grammatical changes, such as the
one in Figure ., or the macro-chain from lexical noun to case affix in
(), based on Lehmann (: ).
() Macro-chain from lexical noun to case affix
Lexical > relational > secondary > primary > case
noun noun adposition adposition affix
Furthermore, chains may involve either substantive expressions, such as
since in (), or schematic categories, such as ‘noun’ or ‘adposition’ in (),
and they may involve either meaning or form, or both, as in Figure ..
Lastly, instead of ‘chain’ (Claudi and Heine ; Craig : –;
Heine ; Heine et al. : –), alternative terms such as ‘con-
tinuum’ (Heine and Reh : ), ‘scale’ (Lehmann []: ;
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION IN SEMANTIC MAPS
Heine and Claudi ), or ‘cline’ (Hopper and Traugott : –)
have been proposed. The term ‘cline’ especially is just as widely used as
‘chain.’² One reason for not adopting the latter term, or any of the other
terms, is that they do not reflect property (c), that is, the interlocking
nature of grammaticalization chains.
5.4 Grammaticalization in semantic maps
From chains of grammaticalization it is only a small step further to the
representation of grammaticalization in so-called semantic maps.
Semantic maps are two- (potentially even three-)dimensional repre-
sentations of degree of similarity of a specific range of linguistic cate-
gories. The similarity is usually conceived in terms of function and
meaning, although it may be instantiated through something different,
such as contexts of use.
To take an example from the lexicon, ‘breathe’ is not obviously
related to ‘take a rest,’ but it is related to ‘pause for breath,’ which can
be seen as an intermediary between ‘breathe’ and ‘take a rest.’ There-
fore, if we represent the similarity between these meanings in a two-
dimensional space, we want to place ‘pause for a breath’ spatially
between ‘breathe’ and ‘take a rest.’ François () presented a map
for the similarity between meanings related to ‘breathe,’ which is
rendered as here Figure ..
Note that the map in Figure . not only contains a number of hypotheses
about the similarity between certain lexical meanings related to ‘breathe,’ but
also about connections between the expressions of those meanings, repre-
sented by lines. For example, a lexeme is likely to express both the meanings
utter, speak take a vacation
whisper take a rest
[someone] blow pause for breath cease to do
[wind] blow breathe
Figure 5.2 Semantic map of ‘breathe’ (François : )
² For a different use of the term ‘cline,’ see Halliday (: ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
‘breathe’ and ‘pause for breath,’ and then may or may not additionally
express the meaning ‘take a rest.’ Furthermore, a lexeme may express
both ‘take a rest’ and ‘take a vacation,’ and in addition ‘cease to do.’ In
contrast, according to the map, we should not expect a lexeme to express
‘take a vacation’ and ‘cease to do’ without also expressing ‘take a rest.’
This is indicated by the lack of a connecting line between ‘take a vacation’
and ‘cease to do.’
The evidence for similarity and connections can be taken from an
appropriate quantity of data from one language, showing, for example,
that two related meanings are expressed by the same lexeme or are used
in the same or in similar contexts. It can also be provided by data from
many languages that show that the similarity is recurring across lan-
guages. This type of evidence is obviously even more robust, and there-
fore semantic maps are primarily associated with linguistic typology.
The concept of semantic maps has also been applied to a number of
areas in grammar. Narrog and Ito () constructed the semantic map
in Figure . based on case polysemy data in languages.
The semantic map contains fifteen meanings/functions from the
domain of case, centered on ‘companion’ (Mary walks with her child)
and ‘instrument’ (Mary stunned Jane with a gorgeous dress). We
assume here that the other meanings and functions on the map are
easily understandable or can be found in the literature or on the
internet. In any case, relationships between meanings in this domain
were posited () on the basis of simple mathematics, namely the
calculation of the—dependency of one meaning on other meanings in
the sample, and () on singular co-occurrences of one meaning with a
specific other meaning.
duration
physical
co-participant route
proximity cause/reason
clausal NP- companion instrument passive agent
coordination coordination
ergative agent
possession source
material
temporal
‘from’
Figure 5.3 Semantic map of the instrument/companion domain (Narrog and Ito )
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION IN SEMANTIC MAPS
As for (), the requirement is that each of the meanings must appear
in at least ten morphemes in the database. If the occurrence of one
meaning depends on another meaning by more than percent (.),
the existence of a connection between those two meanings is hypothe-
sized. Extending the calculation to relations between three meanings, a
dependency of more than percent (.*.) is required in order to
hypothesize a connection, with the three meanings occurring simulta-
neously in at least five morphemes. As for (), it was assumed that if a
morpheme M has only the meanings A and B, and this situation occurs
in at least three different languages, there is a direct connection between
these two meanings. Solid lines indicate the result of calculation of a
dependency-relationship between two meanings, square-dotted lines
the result of calculation for three meanings, round dotted lines the
result of the assumption (), and long-dashed lines the result of the
second assumption under less strict conditions, namely singular co-
occurrence in one morpheme in only two languages.
Now it is possible to add a diachronic dimension to the map. To this
end, we need to assume that synchronic connections between
meanings/functions are the result of semantic extensions that occur
through time. For example, given that ‘instrument’ and ‘passive agent’
are connected, it is both possible that the ‘instrument’ function
extended to a ‘passive agent’ function and vice versa. The ultimate
arbiter on the direction of the connection is diachronic evidence.
Here it is in favor of an extension from ‘instrument’ to ‘passive agent’
(see Narrog : ). When all connections on the map in Figure .
are checked for diachronic directionality, the diachronic map in
Figure . is obtained (cf. Narrog a: ).
duration
physical
co-participant route
proximity cause/reason
clausal NP- companion instrument passive agent
coordination coordination
ergative agent
possession source
material
temporal
‘from’
Figure 5.4 Semantic map of the instrument/companion domain with directionality of
meaning extension (Narrog a: )
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
The first thing to note about this map in terms of grammaticalization
is that only meanings and functions that are already grammatical are
represented. That is, Figure . is a map of ‘secondary’ grammaticaliza-
tions. This is not a necessity, however. Nothing would preclude us from
combining lexical and grammatical meanings and forms in one map. It
is undeniably more common, however, to have semantic maps exclu-
sively made up either of lexical or of grammatical meanings, for the
sake of greater coherence and comparability between the categories on
the map.
Secondly, relating back to Section . on grammaticalization chains,
it is also possible to split up a complex map like the one in Figure .
into many grammaticalization chains, say, from ‘companion’ to ‘instru-
ment,’ and further to ‘cause/reason.’ Also, instead of first building a
map such as the one in Figure . and then adding a diachronic
dimension to it, it would be perfectly fine to work in reverse—combine
several grammaticalization chains and build a larger semantic map like
that in Figure . in a bottom-up fashion. In any case, a diachronic
semantic map is the potentially largest unit to represent consecutive
and related grammaticalizations above the ‘chains’ (Section .) and
the ‘stages’ (Section .).
Further reading
Besides the papers cited in this section, Anderson (, ) and Haspelmath
(, ) are classic papers on semantic maps. Narrog and van der Auwera
() specifically address the topic of semantic maps and grammaticalization,
while Georgakopoulos and Polis () represent the state-of-the-art of
semantic map research.
5.5 Polysemy, transcategoriality, and heterosemy
When as a result of step- or stage-wise grammaticalization, as described
in the preceding sections, the meaning of a linguistic form changes,
then the semantic outcome is commonly described as polysemy: There
are now two related meanings or senses associated with one and
the same form, namely the source meaning and the target meaning,
and the two are in some sense related to one another, that is, there is
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. POLYSEMY, TRANSCATEGORIALITY, AND HETEROSEMY
significant overlap in semantic content between the two meanings
(Enfield : ). For example, when the use of the volition verb will
was extended in the Middle English period to contexts taking inanimate
subject referents incapable of volition (see the parameter of context
extension; Section .), a future tense meaning evolved, giving rise to
polysemy where the earlier meanings of volition and intention now
coexisted side by side with the future meaning (Bybee and Pagliuca ).
Polysemy thus appears to be a largely predictable effect of grammat-
icalization. But there are problems with this term. One problem is how
to establish that the different meanings of a grammaticalizing form
such as will are ‘related’: Is there uncontroversial empirical evidence to
show that there is significant overlap in semantic content between the
concepts of volition and future (tense)?
Another problem is whether the different meanings can belong to
different grammatical categories, such as different syntactic or mor-
phological classes. A third problem concerns the question of whether
diachronic information, such as etymological relationship, should play
a role in defining polysemy. For example, go in (a) is a verb of
physical motion but auxiliary denoting future tense in (b). We
know that the latter meaning is historically derived from the former
(e.g. Bybee et al. ; Hopper and Traugott ; Mair ; Hilpert
). Should such knowledge play a role in deciding whether the two
meanings expressed by go qualify as an instance of polysemy?
() a. Jane is going to Paris.
b. Jane is going to come soon.
A final problem is the following. Once a new meaning has evolved as
a result of grammaticalization, its form may change due to erosion
(Section .). For example, the future marker is going to in (b) is
optionally reduced to is gonna. Assuming that the meanings of go in
(a) and (b) are polysemous, there remains the question whether
polysemy extends also to reduced forms such as gonna?
To be sure, such problems have received detailed treatment, and
extended versions of polysemy have been proposed (e.g. Enfield ).
Still, the general question remains whether ‘polysemy’ is a relevant
concept that we need to take into account when analyzing grammatical
change.
In view of such problems, alternative perspectives and concepts have
been proposed, either to supplement or to replace the term polysemy.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
In particular, two terms have been suggested, namely transcategoriality
and heterosemy.
Transcategoriality obtains when a linguistic form has two (or more)
semantically related members, where each member is associated with a
different plane of linguistic organization and the two are part of some
regular pattern (Heine et al. ). An example of transcategoriality
can be seen in the English item round, occurring as an adjective (a
round table), a preposition (He walked round the corner), a verb (The
boat will round the buoy soon), an adverb (I went round to the shop), or
a noun (It’s my round).
‘Different plane’ on the one hand refers to the morphosyntactic
structure within a sentence, most commonly different word or mor-
pheme categories, or different constituent structures (Robert ). On
the other hand, different planes may also refer to structures of linguistic
discourse beyond the sentence (Do-Hurinville and Hancil ; Do-
Hurinville and Dao ). ‘Related members’ may be both free forms,
like nouns or verbs, or bound forms, like affixes. But one member can
also be a free form while the other is a bound form, such as English full,
which is a free form when used as an adjective (a full cup) but a bound
form when used as a derivational suffix (lawful data).
Membership in a set of transcategoriality rules out items that are
semantically and/or phonologically entirely different from one another,
but it includes cases where the different members share semantic and
phonological features to the extent that they can be conceived by
speakers as being related to one another in a meaningful way. Thus,
members of a transcategoriality set need not be identical in every
respect but must be similar enough in meaning and form to be judged
to be related.³
Unlike polysemy, transcategoriality is claimed to have explanatory
significance. Robert (a, b, c) in particular argues that
transcategoriality can be accounted for in terms of economic motiva-
tion: It provides a means for the optimization of linguistic systems in
that it allows having a maximum of functions expressed with a mini-
mum of forms. Like polysemy, transcategoriality is essentially a syn-
chronic notion, and both are faced with the same problem, namely with
how to determine what ‘related meanings’ are.
³ This criterion comes close to the criterion proposed by Persson (), according to
whom members of a heterosemy set must have a common core sense.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. POLYSEMY, TRANSCATEGORIALITY, AND HETEROSEMY
This problem does not exist in the case of heterosemy. According to
Lichtenberk (), heterosemy obtains when within a single language:
. . . two or more meanings or functions that are historically related, in the sense of
deriving from the same ultimate source, are borne by reflexes of the common source
element that belong in different morphosyntactic categories. (Lichtenberk : )
Heterosemy differs from both polysemy and transcategoriality in the
fact that historical relationship is a definitional property. Consider the
following example. The English numeral one and the indefinite article
a, differ from one another both semantically and formally to the extent
that one might hesitate to classify them as belonging to the same set of
polysemy or transcategoriality.⁴ But they clearly belong to one and the
same set of heterosemy since we know that the latter is historically
derived from the former via grammaticalization.
In a similar way, the Swahili verb ‐taka ‘want’ shares heterosemy
with the future tense prefix ‐ta- since the latter is a grammaticalized
form of the former (see Kuteva et al. , > ). In the course
of grammaticalization, ‐taka was reduced to ‐ta- via erosion (see
Section .). But the two morphemes differ from each other both in their
meaning (‘want’ vs. future tense) and their form (‐taka vs. ‐ta-)—hence,
they are not suggestive of polysemy, nor presumably of transcategoriality.
On the other hand, both of the terms transcategoriality and
heterosemy have been proposed more specifically to deal with poly-
grammaticalization, that is, with cases involving “a mulitiplicity of
grammaticalization chains that may originate in one particular lexical
morpheme” (Craig : ). The example in () from the West
African language Ewe illustrates a network of polygrammaticalization.
The morpheme le occurs on three different planes of grammar, namely
that of lexical items in (a), of verb phrase morphology in (b), and
of noun phrase morphology in (c). On the basis of principles of
grammaticalization it can be assumed that first there was the locative
copula verb, as in (a). This then developed on the one hand in the
direction of the verb phrase, contributing to the rise of a discontinuous
progressive marker le . . . ḿ, as in (b). On the other hand, it also
grammaticalized into the noun phrase, turning into the locative (and
temporal) preposition le, as in (c).
⁴ Not everybody may agree, though, that the numeral and the indefinite article are
semantically different.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
() Ewe (Kwa, Niger-Congo; Heine et al. : –)
a. Kofí le megbé. Verb
Kofi be.at behind
‘Kofi is behind.’
b. Kofí le nú ɖu-ḿ. Aspect marker
Kofi thing eat-
‘Kofi is eating.’
c. Kofí wɔ dɔ́ le Lome. Preposition
Kofi do work at Lome
‘Kofi worked in Lome.’
The terms transcategoriality and heterosemy provide convenient
means for identifying patterns of semantic and formal similarity
between linguistic items belonging to different grammatical cate-
gories. However, heterosemy is arguably the more informative term
since it not only refers to what relationship patterns exist but also to
why they do exist.
To conclude, all three terms, polysemy, transcategoriality, and het-
erosemy, capture certain features of grammaticalized forms. But for the
purposes of grammaticalization research, heterosemy might be the
most useful one since it rests on diachronic reconstruction, which is
central for grammaticalization theory.
Further reading
Polysemy in grammaticalization has been discussed in a wide range of
studies (e.g. Traugott ; Heine b). The term ‘transcategoriality,’ by
contrast, evolved only at the beginning of this century (Robert a, b,
c, ; Do-Hurinville ; Do-Hurinville and Dao ; Do-
Hurinville and Hancil ; Heine et al. ) and has so far not been widely
used in work on grammaticalization. The term ‘heterosemy’ was first pro-
posed by Persson () and subsequently modified and popularized by
Lichtenberk ().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CYCLES
5.6 Cycles
In the present section as well, grammaticalization processes are
considered from a wider perspective, that is, one that may be of help
for understanding why such processes take place. To this end, two main
mechanisms are distinguished in Section .., while Section .. deals
with a concept that has kindled considerable scholarly activity in the
course of the last decade, namely that of cyclic grammatical evolution.
5.6.1 Renewal and innovation
There are two main ways in which grammaticalization can arise,
namely either via renewal or via innovation. Strictly speaking, these
two mechanisms must be distinguished from the concept of grammat-
icalization: They are not covered by the definition of grammaticalization
(see Chapter ). Nevertheless, they are a prerequisite of grammaticali-
zation, and hence are needed for a more comprehensive understanding
of grammaticalization. For example, while presenting a detailed and a
critical account of the notion ‘renewal,’ Reinöhl and Himmelmann
() suggest that,
. . . the empirical ground covered by renewal significantly overlaps with the empirical
ground covered by grammaticalization theory, as both are concerned with the development
of grammatical markers and constructions. (Reinöhl and Himmelmann : )
We are restricted here to a concise descriptive use of the two mechan-
isms to the extent that they are of use to account for grammatical
change.
Renewal is a mechanism whereby an existing expression of a gram-
matical category is replaced by a new expression, with the result that
the new expression assumes a function that is similar to the old one
(Lehmann []: ; Hopper and Traugott : ; Vindenes
).⁵ Renewal can be achieved either by reinforcement or by substi-
tution. Figure . provides a graphic representation of the two kinds of
renewal to be discussed below.
⁵ For reasons of ‘terminological elegance,’ Lehmann ([] : ) uses ‘renovation’
instead of what is called here ‘renewal.’ Note that ‘renovation’ is also used by him
interchangeably with substitution. In order to avoid confounding the two notions, ‘reno-
vation’ is avoided here.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
a Reinforcement A => AB
b Substitution A => B
Figure 5.5 Two mechanisms leading to renewal
Note: (=> stands for ‘is replaced by’)
Reinforcement is commonly understood to be a mechanism
whereby a grammatical marker A is supported by some additional
expression B ‘strengthening’ the meaning of A, the new marker thus
being AB (e.g. Lehmann []: ; Reinöhl and Himmelmann
: ). The term is most commonly applied to processes where a
linguistic form has lost part of its phonetic, semantic, and/or pragmatic
substance and is assumed to be ‘strengthened’ by some other element.
Reinforcement applies, for example, when a case suffix is strength-
ened by an adposition, an adposition is strengthened by an adverb, a
personal affix of a verb is strenghtened by a full personal pronoun, or
a negation marker by some nominal or adverbial expression. Items
reinforcing a grammatical form most frequently follow the latter,
but they may as well precede it, or both precede and follow the
modified lexical form, as the adverb her ‘here’ does in the Norwegian
example in ().
() Present-Day Norwegian (Vindenes : )
de her arrke her
that here sheet here
‘this sheet here’
One may wonder whether ‘reinforcement’ is in fact an entirely suitable
term for the process involved. It would seem that rather than ‘reinfor-
cing’ or ‘strengthening’ an existing form, a more important reason for
introducing a new element is to provide the speaker with new discourse
options for expressing functions such as conceptual differentiation,
focusing, emphasis, anaphoric highlighting, or for narrowing down
the range of possible referents.⁶ This, however, is an issue that would
need a separate treatment (see, e.g., Hansen ).
⁶ Since the term ‘reinforcement’ addresses at best only one of the functions of the new
element, the more general term renewal is preferred in some of the literature of grammat-
icalization (e.g. Heine and Reh : ; Hopper and Traugott ; Kuteva ;
Giacalone Ramat and Mauri : –; cf. also the ‘phonological bulking’ of Matisoff
() and the ‘hermit crab processes’ of Heath ()). Presenting a detailed analysis of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CYCLES
In substitution, also referred to as ‘renovation’ (Lehmann []:
), expression A falls into disuse and is replaced by a largely equivalent
expression B. For example, the Latin conjunction nam ‘because’ was
replaced by the form quare, which developed into the reason conjunction
car ‘because’ of Modern French (Lehmann []: ).
Paradigm examples of substitution can be found in the develop-
ment of pidgin languages. Pidgins arise in situations of intense
language contact where a contact language, frequently called the
‘lexifier language,’ is used regularly by speakers whose first lan-
guages are different. In such situations, both the lexicon and gram-
matical structure of the contact language tend to be drastically
reduced (see Section .). This reduction process affects in particu-
lar bound morphemes such as inflectional and derivational affixes,
which are likely to disappear and to be replaced by free forms like
lexical items. The way this process may lead to large-scale substitu-
tion can be illustrated by the following example, where the ‘lexifier
language’ was a variety similar to Standard Swahili, cf. (), and the
resulting pidgin was Kenya Pidgin Swahili, cf. the corresponding
sentence in ().
() Standard Swahili (own data)
ni-me-ku-tafut-i-a kazi.
--.-search-- work
‘I have been looking for work for you.’
() Kenya Pidgin Swahili (Heine , ; Heine and Kuteva )
mimi kwisha tafuta kazi kwa wewe.
I finish search work at you
‘I have been looking for work for you.’
The examples illustrate four kinds of substitution that appear to have
affected the pidgin: (a) The first person subject prefix ni- of Standard
Swahili was replaced in the pidgin by the free pronoun mimi ‘I’, (b) the
second person object prefix ‐ku- was replaced by the free pronoun
the notion ‘renewal,’ Reinöhl and Himmelmann (: ) conclude that it “is not a useful
and viable concept for the analysis of linguistic change.”
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
wewe ‘you’, (c) the perfect prefix ‐me- was replaced by the verb kwisha
‘finish’, and (d) the function of the applicative suffix ‐i- is expressed by
the general locative preposition kwa ‘at, with’ in the pidgin.
Innovation is a mechanism whereby a new grammatical category is
created for which earlier there was no conventionalized category in the
language concerned (Lehmann []: ). Innovation contrasts
with renewal, which introduces new grammatical expressions by repla-
cing earlier equivalent expressions. For example, Latin had no con-
ventionalized articles whereas modern Romance languages do. The
grammaticalization of the Latin numeral unus ‘one (.sgm)’ to
indefinite articles and of the Latin distal demonstrative ille ‘that
(.sgm)’ to definite articles in the Romance languages thus consti-
tute instances of innovation rather than renewal.
The factors leading to innovation are still largely unclear. But it
seems that language contact can play an important role, where speakers
of language A replicate a grammatical category from language B for
which previously there was no equivalent category in A. A number of
innovations of this kind are reported by Aikhenvald (, , )
from the North Arawak language Tariana of northwest Amazonia in
Brazil. Tariana has been in close contact with languages of the East
Tucanoan group, and in the course of this contact it innovated a
range of new grammatical structures by replicating (‘calquing’) East
Tucanoan structures.⁷ For example, Tariana speakers innovated a per-
fective category by grammaticalizing their verb ‐sita ‘finish’ to a per-
fective aspect marker on the model of East Tucanoan languages.
The above observations relate to clear manifestations of the two
mechanisms—they ignore the fact that renewal and innovation cannot
always be clearly separated from one another (see Lehmann
[]: – for discussion). Furthermore, our focus is exclusively on
grammatical change in the early stages of its development. What on the
surface appears to be a case of substitution can actually go back to an
earlier reinforcement process, as will be shown in Section ...
Renewal and innovation have contrasting effects on the typological
profile of the language in which they occur: That profile is not signif-
icantly altered when renewal takes place. Innovation, by contrast,
means that the language is enriched with a new grammatical category
⁷ For the term ‘replication’, see Chapter (language contact); cf. also Heine and Kuteva
().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CYCLES
and, hence, its profile is no longer exactly what it was before. And the
two mechanisms differ also from one another in that only the former
normally gives rise to morphological cycles.
5.6.2 The morphological cycle
Both reinforcement and substitution can be responsible for cyclic
grammatical evolution. This can be demonstrated with the classic
example of negation marking in the history of French. In Old
French, the verbal negation proclitic non (subsequently nen and
finally ne) was frequently reinforced with post-verbal nouns such as
pas ‘step, pace’, point ‘dot, point’ or mie ‘crumb’. The nouns pas and
point were subsequently grammaticalized, giving rise to the discon-
tinous negation marker ne–pas and ne–point of Modern French. Ne–
pas is frequently reduced to pas in colloquial speech. This evolution
can be reconstructed as in (); for exemplification, see Hansen
(: ).
() Reconstruction of the evolution of reinforcement in French
negation marking (simplified)
Stage Construction Structure
I ne Verb A
II ne Verb (pas) A (B)
III ne Verb pas AB
IV (ne) Verb pas (A)B
V Verb pas B
Comparing only the initial stage I with the final stage V, the impres-
sion is conveyed that this is an instance of substitution in that the
earlier negation marker ne (A) was replaced by pas (B). As a matter of
fact, however, we are dealing with a cyclic evolution of reinforcement
that must have proceeded along the stages set up in ().
Cases like () are referred to in the relevant literature as instances of
a morphological cycle, or of cyclic grammaticalization, that is, a cyclic
form of evolution, typically involving morphosyntactic material, lead-
ing to the renewal of a grammatical category.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
Most of this work is devoted to what is commonly referred to as
Jespersen’s Cycle and the study of negation, such as that of the general
French negation marker sketched above (Jespersen ). But the
notion of cyclic grammaticalization has also been employed for a
number of other grammatical phenomena. In particular, van Gelderen
(a) discusses the following cases of cyclical morphosyntactic devel-
opment with examples from a range of languages:
(a) the subject agreement cycle;
(b) the object agreement cycle;
(c) the pronominal copula cycle;
(d) dependent marking cycles;
(e) the DP cycle;
(f) tense–mood–aspect cycles;
(g) negative cycles.
Cyclic grammaticalization can in fact be observed in a number of domains
of grammar. Examples of the cyclic renewal of demonstratives are dis-
cussed by Vindenes (), and the notion of cyclicity has also been
extended to semantic-pragmatic phenomena and the analysis of prag-
matic markers (Hansen ); see also below, and Heine et al. ().
The example of French negation in () illustrates two important
features of cyles. First, the old and the renewing element may be similar
in meaning, but this is by no means a requirement: The reinforcing noun
pas ‘step’ of stage II had semantically little in common with the negation
marker ne. And second, the renewing element need not occur in the
same slot as the old element: The cycle had the effect that negation in
French shifted from the pre-verbal to the post-verbal position.
Most work on cyclicity has been concerned with morphosyntactic
change. But cyclic development can also be found in semantic change.
For example, based on the data of Ghezzi and Molinelli (), Hansen
() discusses what she calls semantic/pragmatic cycles in Latin and
Italian, called pragmaticalization cycles by Ghezzi and Molinelli ().
For example, both the Latin speech act verbs rogo ‘I ask’ and quaeso
‘I pray/request’ and their Italian equivalents chiedo ‘I ask’ and prego
‘I pray’ gave rise to a series of cyclic developments from verb form to
politeness marker.
Cyclic grammaticalization can be restricted to one morphological
cycle, but it may as well be repeated. Thus, in the development of forms
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CYCLES
for ‘(with)in’ from Proto-Indo-European to French there were three
instances of cyclic reinforcement, as the example in () shows.
() From Proto-Indo-European to French: Repeated reinforcement
of forms for ‘(with)in’ (Lehmann []: ; “=>” stands
for reinforcement, and “>” for grammaticalization)
a. Proto-Indo- *in ‘in’ => *en-tos
European
b. Latin intus ‘within, => *de-intus ‘of/from
inside’ within’
c. French > dans ‘in’ => dedans ‘inside’
Rather than the metaphor of a ‘cycle’, it has been argued by some
scholars that a ‘spiral’ is more appropriate to describe the developments
involved. Von der Gabelentz ([] : –) was presumably the
first to propose the latter metaphor, arguing that renewed pathways are
not exactly the same as the ones preceding them. The term was taken
up by Meillet ([] ), who maintained that linguistic develop-
ment proceeds in spirals since languages add extra words to obtain an
intensified expression. Subsequently, the words weaken, wearing out and
being reduced to the level of simple grammatical tools. New words are
added again for expressive purposes, and the weakening process begins
anew, and so on without end. The term spiral has been taken up in some
recent studies of cyclic grammatical change. Hansen (: ), for exam-
ple, found that some of the morphosyntactic and semantic/pragmatic
changes identified by her in pragmatic markers of Romance languages
behave “more like spirals than cycles.” Haspelmath () speaks of an
Further reading
The following studies in particular provide accounts of cyclic grammatical
development: Jespersen (), Heine and Reh (: –), Heine et al.
(: –), Croft (), Dahl (), van Gelderen (, a, b,
), Hansen (, , ), and Ghezzi and Molinelli (). Earlier
studies of cyclicity include von der Gabelentz (), Jespersen (), Meillet
([] ), and Hodge (). Hodge () argued that cyclical develop-
ment can be applied to the evolution of entire languages (cf. also Section ..).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
STEPS AND PHASES OF GRAMMATICALIZATION
‘anasynthetic spiral’ for the change involved in the renewal of categories
that are expressed synthetically in morphosyntax by those that are ex-
pressed analytically (see Chapter ).
Discussion points
() Grammatical change has been variously described as gradual, abrupt, or
continuous. What advantages and drawbacks do you see for each of these
perspectives?
() The first three sections of this chapter have discussed steps and stages
and chains of grammaticalization. One can also find the term ‘cline’
instead of ‘chain.’ Which of these terms do you think is the most charac-
teristic for grammaticalization?
() Why would speakers of a specific language be motivated to create a ‘cycle’
or a ‘spiral’ of a certain category, and under which circumstances may they
be not motivated?
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
6
What drives grammaticalization?
Mechanisms and motivations
6.1 Mechanisms: reanalysis and analogy
6.2 The language system
6.3 Communicative needs or goals of the speaker
6.4 Inferences and contexts
6.5 Cognition and conceptualization
6.6 Frequency of use and language processing
6.7 Discourse
6.8 Conclusion
Discussion points
In this chapter we will discuss what is behind grammaticalization. Even
at this point this is still a wide-open question, since it cannot be
sufficiently answered purely on the basis of language data. Most scho-
lars would agree that an answer would require reference to fundamen-
tals of cognition and communication to which we do not have direct
access. The solutions scholars have given tend to correlate to their
theoretical backgrounds and convictions, and their general approach
to the study of grammaticalization.
Answers to what drives grammaticalization are usually given in terms
of ‘mechanisms’ and ‘motivations.’ There is no clear borderline between
the two. For some scholars, the mechanisms as such are already the
motivations, for others, motivations are the triggers for mechanisms,
and for still others, mechanisms are not needed. On the other hand,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
many processes have been labelled variously as both ‘mechanisms’ and
‘motivations’ (for example, analogy, metaphor, inferences).
Nevertheless, we make a distinction here along traditional lines.
‘Reanalysis’ and ‘analogy’ have been labelled fairly consistently as
‘mechanisms.’ They are discussed together in Section .. Sections .
to . discuss aspects of grammaticalization typically considered
as ‘motivations’ for grammaticalization, namely the language system
(Section .), communicative goals (Section .), inferences and con-
text (Section .), cognition (Section .), frequency (Section .), and
discourse (Section .). They are followed by a summary and conclu-
sion in Section ..
6.1 Mechanisms: reanalysis and analogy
‘Mechanism’ refers to possible regularities in how some meaning or
form, or combination of meaning and form changes, beyond a mere
comparison of source and target (cf. Lessau : ). This may
include enabling and motivating factors of the change (cf. De Smet
: ).
In historical linguistics, reanalysis, analogy (extension), and borrow-
ing have been proposed as the three main mechanisms of grammatical
change (cf. e.g. Harris and Campbell : ; Campbell : ).
Borrowing is part of the discussion of grammaticalization and language
contact in Chapter . It can often be observed very directly from its
result with some knowledge of the source. In contrast, whether a
change should be attributed to reanalysis or analogy, or both or neither
of them is far less obvious and has been notoriously controversial. We
will start here with a short discussion of reanalysis (Section ..),
which was identified as the main mechanism for a long time, continue
with a discussion of analogy in Section .., which has received more
attention recently, before summarizing in Section ...
6.1.1 Reanalysis
While unknown as a concept to the nineteenth- and early twentieth-
century historical linguists, reanalysis has been a mainstay of modern
grammaticalization theory from early on (see Hansen ). Lord
(, ) identified grammaticalization of serial verbs mainly with
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. MECHANISMS: REANALYSIS AND ANALOGY
reanalysis. Heine and Reh (: –) provided numerous potential
examples for grammaticalization and reanalysis in African languages,
emphasizing that grammaticalization and reanalysis were related but
not identical processes. The most frequently cited definition of reana-
lysis is by Langacker (: ), namely “as change in the structure of
an expression or class of expressions that does not involve any imme-
diate or intrinsic modification of its surface manifestation.” He adds,
however, that “reanalysis may lead to changes at the surface level . . . ,
but these surface changes can be viewed as the natural and expected
result of functionally prior modifications in rules and underlying re-
presentations” (Langacker : ). The structure affected may
include:
(i) constituency and hierarchical structure,
(ii) categorization,
(iii) grammatical relations, and
(iv) cohesion, that is, if a linguistic unit is an independent word or
part of a larger unit (cf. Harris and Campbell : –).
English be going to, grammaticalizing from verb of motion to future
tense, may serve as an example again. According to Campbell (:
), the following reanalysis took place.
() Lilly is going to marry Ron.
Structure: Lilly is going to marry Ron
() Lilly is going to marry Ron.
Structure: Lilly is going to marry Ron
In (), be going to is a verb of motion in a purposive construction, that
is, someone is physically moving with the goal of performing a specific
action at the destination. In (), be going to indicates a temporal
relationship between the time of speech and the event indicated by
the main verb. The decisive point for Campbell () is that the
surface structure of the sentence has not changed, but speakers inter-
pret it differently.
Following Hopper and Traugott (: ), we can assume that addi-
tionally a rebracketing of constituents has taken place. Speakers have
reanalyzed be going to in () as one unit in (), that is, as an auxiliary:
() Lilly is going [to marry Ron].
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
() Lilly [is going to] marry Ron.
It is important to note, though, that the essence of reanalysis is not
category change or syntactic rebracketing but semantic reinterpreta-
tion. This is clearly stated by Harris and Campbell (: ): “Gram-
maticalization is often associated with ‘semantic bleaching,’ and this
‘bleaching’ is the result of reanalysis or, perhaps better said, it is the
essence of the reanalysis itself.”
This analysis of grammaticalization in terms of reanalysis may
sound very common-sensical. Hopper and Traugott (: ) still
stated that “[r]eanalysis is the most important mechanism for gram-
maticalization, as for all change,” and in structural approaches to
grammaticalization, reanalysis has always been identified as a core mech-
anism of grammaticalization (cf. e.g. Roberts and Rosseau ; van
Gelderen c). In generative frameworks it is generally assumed that
it is the child in language acquisition that does the reanalysis.
However, the role of reanalysis in grammaticalization has increas-
ingly come under fire. In (i) to (v), we will list the most important
arguments that have been brought up against reanalysis, and offer a
short critical discussion of each argument.
(i) Grammaticalization is unidirectional, while reanalysis is not
(Heine and Reh: : ). For example, the plural -s in Old
English treowe-s was reanalyzed as part of the stem to become
Modern English truce, while in the case of peas, the -s was
originally part of the noun stem (Haspelmath a: ).
However, the few examples for the bidirectionality of reana-
lysis that have been provided are mostly from the domain of
morphology, where counterdirectionality is more commmon
than in other domains of grammar. Morphology is also the
most important source for examples of change in the opposite
direction of typical developments in grammaticalization (cf.
Section .). In fact, the reanalysis of treowe-s to truce can be
taken as the last stage of grammaticalization (obsolence/loss),
and the reanalysis of peas to pea-s, as an exceptional change in
the opposite direction.
On the other hand, it is clear that unlike grammaticalization,
reanalysis is not intrinsically unidirectional. This should be
undisputed.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. MECHANISMS: REANALYSIS AND ANALOGY
(ii) While reanalysis is abrupt categorical change, grammaticaliza-
tion is gradual (cf. Lehmann : , Haspelmath a:
). That is, reanalysis was negatively associated with a theo-
retically motivated concept of abrupt grammaticalization that
was found to be empirically not adequate (cf. Section .).
However, at least some proponents of reanalysis consider
reanalysis as ‘micro-change’ in contrast to grammaticalization
as ‘macro-change.’ Thus, one grammaticalization can consist of
a number of micro-changes, including several reanalyses (e.g.
Harris and Campbell : ). Reanalysis in the eyes of these
proponents is rather like the seed of change, the invisible
semantic reinterpretation that later may or may not lead to
more obvious and drastic syntactic and phonological change
(cf. Harris : ).
(iii) Not all grammaticalization involves reanalysis, and not all
reanalysis is grammaticalization (e.g. Heine and Reh ;
Haspelmath a).
For the latter, it is not difficult to come up with examples.
Haspelmath (a: –) provides a number of them, includ-
ing the reanalysis of the German external possessor dative to
adnominal possessor, which is not an apparent grammaticali-
zation. Lehmann (: ) shows that in ‘lateral conversions’
like the derivation of the verb butter from the noun butter,
reanalysis has taken place but not grammaticalization.
In contrast, for the former, the evidence is much trickier, and
largely hinges on the concept of reanalysis adhered to. Haspel-
math (a: –) provides a host of alleged examples of
grammaticalization without reanalysis. These include change
from serial verb to preposition and complementizer, from rela-
tional noun to adposition, from head to dependent, clause
fusion, and so on. These changes have in common that they
were presumably neither abrupt nor did they involve a change
in the hierarchical structure of the clause.
This argument rests on very specific assumptions about what
reanalysis must be like: It must be abrupt and it must imply a
change in hierarchical clause structure. However, as we have
seen above, both assumptions are neither conceptually neces-
sary nor are they held by the proponents of reanalysis them-
selves, at least not by all.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
(iv) Reanalysis, in contrast to analogy, has no psychological or
neurological reality in terms of language processing (Fischer
).
This argument is very difficult to prove or disprove with the
methods of linguistic research without resorting to circularity.
Fischer’s understanding of language processing and language
change is mainly informed by Pulvermüller’s () model of
‘neuronal grammar,’ in which analogy plays a central role (cf.
Fischer : –). This idea is intriguing in its reference to
interdisciplinary research, but it is practically impossible to
evaluate it meaningfully in a purely linguistic framework, espe-
cially given that hypotheses about grammar and the brain are
constantly subject to modification themselves.
(v) Reanalysis needs ambiguous structures. However, an ambigu-
ous structure presupposes that an alternative interpretation
already exists, that is, that some change has already taken
place (Lehmann : ; De Smet , ). De Smet
() proposes in more detail that change traditionally labeled
as ‘reanalysis’ is in fact due to either analogy, ‘automation’ or
‘category-internal change,’ whereby the latter two can be under-
stood as kinds of extension (cf. Section ..), that is, are also
analogical in nature.
This is a logical rather than an empirical argument and may
in fact be the most substantial of all counterarguments to
reanalysis in language change in general. It implies questioning
fundamental assumptions about the nature of linguistic cate-
gories. If an existing model is necessary for reanalysis to a new
category, then, indeed, either all change is essentially analogical,
or we have to assume the existence of universally available
categories (De Smet : , –). On the other hand,
one could argue that universally available pathways of reason-
ing, that is, inferences in context, enable human beings to create
new target meanings and functions through reanalysis even if
the target was not extant as a model in the language. In that
case, it would not be the categories that are universally available
but the paths leading to those categories.
Note also that according to Hansen (), reanalysis typi-
cally takes place in contexts that are not ambiguous but rather
vague or underspecified while ambiguity may be the result of
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. MECHANISMS: REANALYSIS AND ANALOGY
reanalysis. Furthermore, she makes a distinction between two
kinds of reanalysis, namely ‘neo-analysis’ and ‘re-analysis,’
according to whether or not the hearer’s mental grammar
already includes an existing analysis of the construction which
is reanalyzed.
In conclusion, it is not clear that every grammaticalization includes
reanalysis, and it is clearly not the case that every reanalysis in language
change is a grammaticalization. However, this does not exclude the
possibility that reanalysis still plays an important role in grammatical-
ization. Almost everything hinges on the concept of reanalysis applied.
In much of the discussion about reanalysis, proponents and opponents
do not share the same concept. Opponents often assume a very rigid
idea of reanalysis that one might stereotypically associate with histor-
ical research in a traditional generative framework. It only matches the
actual concept of reanalysis espoused by some of its proponents, and is
not a logical necessity either.
6.1.2 Analogy/extension
Analogy is a broad concept based on similarity. Luraghi and Bubenik
(: ) define analogical change as “[c]hange in word structure
under the influence of semantically, formally or functionally related
words. It results from an attempt to make some linguistic forms more
similar to other linguistic forms in some respect.” A typical example is
so-called ‘analogical leveling,’ in which words with a minor inflectional
pattern change their forms in accordance with a majority pattern; for
example Old English bōc/bēc (singular and plural of ‘book’) becoming
book/books in Modern English or sunne/sunnan (singular and plural of
‘son’) becoming son/sons (Gaeta : ).
The concept of analogy is believed to go back to Ancient Greece and
was already established in nineteenth-century historical linguistics. It
has been applied to practically all areas of grammar, including phonol-
ogy. Because of its age and the breadth of phenomena that can be
subsumed under it, it has been considered as a rather vague concept.
Attempts have been made to improve on it, including the introduction
of an alternative term ‘extension,’ which could be defined from
scratch and therefore more narrowly. Concretely, Harris and Campbell
(: ) define extension as “a mechanism which results in changes
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
in the surface manifestation of a pattern and which does not involve
immediate or intrinsic modification of underlying structure.” This
definition is intentionally designed to complement reanalysis as a
change not directly affecting surface structures.
Analogy was originally known to historical linguists as an irregular
type of change complementing regular change (cf. Campbell :
–; Gaeta : –). So, not surprisingly, one of the earliest
proponents of grammaticalization, Meillet ([] ), treated anal-
ogy and grammaticalization as mutually exclusive.
When grammaticalization reemerged as a topic in linguistics in the
s and s, analogy was initially treated with great skepticism as
well. Givón (: ) labeled analogy as a “pseudo-explanation” and
Heine and Reh (: ) wrote, “it is doubtful whether analogy can
be attributed any explanatory significance at all.”
In the s a different view became prevalent, according to which
change was initiated by reanalysis, and analogy/extension was the
potential key mechanism in the ‘actualization’ of the change, that is,
its spread. The basic idea came from Timberlake (: ), who
suggested that reanalysis was followed by ‘actualization,’ which is “the
gradual mapping out of the consequences.” In the model of change
proposed by Harris and Campbell (: ), “[o]nly reanalysis and
borrowing can introduce an entirely new structure into a language,”
while “by definition, extension cannot introduce an entirely novel struc-
ture.” Instead, extension followed on reanalysis and helped to alter
surface manifestations step by step, thus rendering (syntactic) change
gradual (Harris and Campbell : ). Multiple reanalyses and multi-
ple extensions might follow on each other to complete a change.
Taking the example of be going to again, Hopper and Traugott (,
) showed how a sequence of reanalysis and analogy (extension)
could come up for the process of grammaticalization. As we saw in
Section .. ((), ()), first a reanalysis, as from () to () took place.
() Lilly is going [to marry Ron].
() Lilly [is going to] marry Ron.
Then, analogy took over, as the pattern spread to verbs with which the
original construction was semantically not compatible, such as like
in () (cf. Hopper and Traugott : ).
() Lilly [is going to] like Ron.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. MECHANISMS: REANALYSIS AND ANALOGY
Most instances of analogy in this kind of model, like the one in (), can
perhaps more profitably be interpreted as manifestations of ‘context
extension,’ that is, one of the four parameters of grammaticalization
discussed in Section .. Since analogy brings about (linguistic or
sociolinguistic) ‘rule spread’ rather than ‘rule change,’ it presupposes
reanalysis in grammaticalization. For Hopper and Traugott (:
–, ), the presence of analogy even provides evidence to establish
that reanalysis has taken place.
Lastly, as we can also expect from Harris and Campbell’s ()
model, another reanalysis, namely from going to to gonna, took place in
the domain of morphology.
While in the s the idea of a collaboration between reanalysis and
analogy in grammaticalization took hold, more recently, scholars from
totally different theoretical backgrounds have emerged who suggest
that analogy is the fundamental motivation for grammaticalization,
and reanalysis is irrelevant. In (i) and (ii) we will list brief versions of
their claims.
(i) Fischer (, , , ) seeks to locate grammaticaliza-
tion in a ‘usage-based’ approach to the study of language. Based
on a model of neuronal grammar by Pulvermüller (), Fischer
assumes that not only language processing but also language
acquisition is mainly based on analogy. In this view, analogy is
not only the mechanism, but also the cause and motivation for
language change. Grammaticalization itself is not a specific type of
change, but the result of interaction of analogy with usage fre-
quency (cf. Fischer : –), and it is essentially not direc-
tional. De Smet () takes the same ‘usage-based’ approach.
However, as seen in Section .., while largely denying the
validity of ‘reanalyis,’ he does not attribute change exclusively to
analogy.
(ii) For Kiparsky (), grammaticalization is ‘grammar optimiza-
tion’ via analogical change. Since grammaticalization was origi-
nally conceptualized by Meillet () as the creation of ‘new’
categories, and analogy is based on the existence of an extant
model, there is an apparent contradiction. This is resolved by
Kiparsky () through the hypothesis that the model for
grammaticalization, that is, constraints, patterns, and categories,
are provided by Universal Grammar, that is, they need not be
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
extant in the grammar of the language. Because of the con-
straints of Universal Grammar on optimization, analogical
change is necessarily unidirectional. The type of analogical
change instantiated through grammaticalization is labeled as
‘non-exemplar based’ (Kiparsky : –), in contrast to ordi-
nary exemplar-based analogy, which can also lead to changes of
different directionality.
As can be seen, the claims in (i) and (ii) and the two concepts of
analogical change are almost diametrically opposed and have little
more in common other than their label. Fischer’s analogy is of the
commonly known exemplar-based kind, while Kiparsky’s analogy is an
extraordinary one which is non-exemplar based. Relatedly, Fischer’s
model is usage-based while Kiparsky’s is based on Universal Grammar.
And lastly, their respective concepts of ‘analogy’ lead to the assumption
of non-directional change in the case of Fischer, and unidirectional
change in the case of Kiparsky.
6.1.3 Summary
Reanalysis and analogy have been proposed by many scholars as the
main mechanisms of grammaticalization, sometimes one to the exclu-
sion of the other, and sometimes as collaborating. Here is a non-
exhaustive but hopefully representative list of stances with respect to
the relationship between grammaticalization and these mechanisms.
Some proponents are listed in brackets.
A Grammaticalization is essentially based on reanalysis (e.g. Lord
, ; Roberts and Roussau ; Eckardt ).
B Grammaticalization relies on reanalysis, but analogy may collab-
orate in the process, via extension (e.g. Hopper and Traugott
, ; Harris and Campbell Itkonen ; Traugott
and Trousdale ).
C Grammaticalization is essentially based on analogy (Fischer ,
, , ; Kiparsky ).
D Grammaticalization may sometimes but does not always involve
reanalysis and analogy (Lehmann ).
E Grammaticalization may involve reanalysis, but analogy is a
separate type of change (Heine and Reh ).
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. THE LANGUAGE SYSTEM
F Grammaticalization often involves analogy, but reanalysis is irrel-
evant (De Smet , ).
G Grammaticalization is a mechanism on its own, and analogy and
reanalysis are of little relevance to it (Meillet ; the concept of
reanalysis was not yet known at Meillet’s time; Haspelmath ,
where reanalysis is irrelevant, and analogy is not mentioned;
Heine a—reanalysis and analogy both irrelevant).
Some logical possibilities have not been claimed yet, for example, that
grammaticalization equals analogy, or equals reanalysis. Both claims
would be hampered from the start by the fact that grammaticalization
has been conceptualized and often even defined as intrinsically unidi-
rectional, unlike reanalysis and analogy. Also, analogy and reanalysis are
not confined to grammar. Furthermore, it has also not been claimed yet
that all grammaticalization must involve both reanalysis and analogy.
To summarize, in modern grammaticalization theory, reanalysis has
been the elephant in the room as the main mechanism, while propo-
nents of analogy have emerged more recently. Much of the argumen-
tation in the literature has taken place at a very abstract level, as also
mirrored in this section. This reflects the difficulty in pinning down
what is actually taking place in language change. Ultimately, in order to
know what mechanisms are at work, one would have to be able to look
into the mind of the speaker during language change, which is only a
remote possibility in the near and mid-term future. This does not
preclude that on the basis of purely linguistic arguments at some
point a consensus can be reached about the main mechanism behind
grammaticalization, but currently we are far from that.
6.2 The language system
In historical linguistics, there has been a line of thought that looks to
the language ‘system,’ or to language structures themselves, as the
motivation or cause of language change in general. For example,
according to Luraghi (: –), some scholars have claimed that
languages change in order to preserve distinctions and distances
between items in a language system, and others have claimed that
languages are goal-directed systems in which change occurs to ensure
the functionality of the system. In other words, languages are sets of
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
correlated structures, and if the structures or their correlations are for
some reason disrupted, or decline, or have been lacking in some aspect,
language change will make up for the deterioration or the lack. This
type of motivation for language change has also been called ‘internal,’
that is, inside of language and language structure. In contrast, there are
possible ‘external’ motivations for change that come from outside
language itself. Most prominent is language contact (see Chapter ),
but there are also others, such as communication, social goals of the
speaker, or cognition, which will be discussed in the following sections
of this chapter.
Internal motivations for language change in a slightly different sense
are also mentioned prominently in the classic historical linguistics
textbook by Hock and Joseph (). For example, a principle of
“one meaning–one form” may be responsible for a development in
which all word forms of the Modern English verb choose (choose, chose,
chosen) came to have a word-internal /s/ throughout, instead of /s/ in
some forms and /r/ in other forms in Old English. The process took
place as analogical leveling of /r/ to /s/ (Hock and Joseph : ).
Or, likewise, the emergence of the use of hopefully as a sentence
adverb (e.g. Hopefully, I’ll be finished by tomorrow) may have been
motivated by the parallel to the adverb happily that was previously
used in that position (Hock and Joseph : ). Note that these
types of analogical changes are not necessarily goal-directed. Neverthe-
less, they also refer to motivations for change that lie in language
structure itself.
When it comes to the phenomenon of grammaticalization, and the
main stream of grammaticalization studies, however, the language
system, or internal, structural motivations in general, have not played
a prominent role in the discussion. In contrast, some proponents of
grammaticalization have been explicitly critical. For example, Lehmann
(: ) noted that “a considerable number of grammatical changes
are quite superfluous from the point of view of the language system;
that is to say, the change leads to a state that is maximally similar to the
starting point. . . . This shows once more that system-internal explana-
tions of linguistic change do not fit. There is much change just for the
sake of change.” And Hopper and Traugott (: ) state that, “to
assume that as an older system becomes eroded it may cease to function
at an adequate level of communicative coherence and therefore must
necessarily be revitalized . . . is to posit a stage of language such as is
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. THE LANGUAGE SYSTEM
unknown, in other words it violates the uniformitarian principle and is
not empirically supported by the data. Most emphatically, languages
are not goal-oriented.” In other words, the idea that language systems
are at any stage deficient is rejected on the basis of the observation that
languages seem to perfectly meet their speakers’ needs at any historical
stage of language development. Beyond that, the underlying idea of a
‘language system’ is also sometimes questioned, especially within fra-
meworks that regard grammar as usage-based and therefore fluent or
even ‘emergent’ (cf. Hopper and Traugott : , ).
There are two exceptions to this indifference towards possible
motivations for grammaticalization in language structure or in the lan-
guage system. The most prominent one is the study of grammaticalization
in the generative framework. This framework is generally concerned
with language structure as the cause and the target of grammaticaliza-
tion and will be discussed in some detail in Section .. The other
exception in terms of a specific approach is the study of grammatical-
ization in the tradition of Danish structuralism (e.g. Andersen a,
b, ; Nørgård-Sørensen et al. ; Nørgård-Sørensen and
Heltoft ), which will be briefly discussed in this section. Other
individual scholars holding a similar view are mentioned in “Further
reading” below.
The overarching idea about change in the Danish structuralist
approach is that “all linguistic change, . . . is the product of multifarious
grammar-internal conditions plus the social conditions” (Andersen
a: ). Nørgård-Sørensen et al. () present a number of studies
about grammaticalization changes within grammatical systems, espe-
cially morphological systems. For them, “the great majority of morpho-
logical changes are changes from one morphological system to another”
(Nørgård-Sørensen et al. : xi). They coin the term ‘connecting
grammaticalization.’ This term refers to the idea that “[m]orphological,
topological and constructional paradigms very often connect to form
complex paradigms, so-called hyperparadigms . . . , and grammaticaliza-
tion processes include the formation, restructuring and dismantling of
such complex paradigms” (Nørgård-Sørensen et al. : xii).
One example is the parallel development of animacy as gender in the
nominal domain, and the perfective–imperfective aspect distinction in
the verbal domain in Russian as sketched out by Nørgård-Sørensen
(: –). The development in the nominal domain started when,
as a result of regular sound change, the nominative and the accusative
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
of masculine singular nouns merged in Common Slavic. This led to an
extended use of the genitive as the second argument (instead of an
accusative) for the domain of individual-denoting nouns. Eventually,
the choice of accusative or genitive for the second argument became an
index of the animacy value of the noun. Decisively then, in the four-
teenth century, all plural declensions of nouns, which were distinct
until then, merged into one. This extended the genitive in second
argument (accusative equivalent) function from specific masculine
declensions to the entire noun lexicon. This in turn implied the spread
of the animate/inanimate opposition to all nouns, that is, the establish-
ment of animacy as a gender distinction.
In parallel, in the eleventh century, a new suffix on verbs, -iva-/-ivaj-,
developed that obtained the broader function of marking non-action, and
began ousting other suffixes in this function. In the fourteenth century,
the distinction of action/non-action was reanalyzed as change-of-state vs.
non-change-of-state, that is, the modern perfective–imperfective dis-
tinction. This involved a conceptual reinterpretation of the action from
an indivisible whole to a complex concept with two constituents, the
activity and the state, linked by telicity, adding to the grammatical
prominence of the activity/state distinction. Overall, then, the parallel
changes in the verbal and nominal domain conspired to foreground
an activity/potential actor (animate) versus state/non-potential-actor
(inanimate) distinction in the language, and according to Nørgård-
Sørensen (), they can be seen as one integral ‘reanalysis’ of major
parts of the grammar.
While grammar or ‘the language system’ as a motivation for gram-
maticalization is rather controversial, it should be less contentious that
extant grammar serves as a constraint and a model for new gramma-
ticalizations. Unless there is a severe disruption to language structure,
most grammatical innovations adhere to the rules of extant grammar
such as word and constituent order. Even under the influence of a
contact language, when categories are ‘replicated’ from a model lan-
guage, the category is usually replicated in a manner structurally
compatible with the target language, for example by emulating already
extant minor patterns in the target language (cf. Heine and Kuteva
; Chapter ). Likewise, it is only reasonable to assume that among
many options available to express a state-of-affairs in a novel way, it is
those that are supported by the extant vocabulary and grammar of a
language that are more likely to be formed and eventually adopted.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. THE LANGUAGE SYSTEM
But beyond its ‘constraining’ influence, extant grammar most likely
also serves as an analogical model. This can be seen by the fact that
while theoretically we might expect that every language has one expo-
nent for each cross-linguistically available grammatical category, in
reality, speech communities tend to grammaticalize specific categories
over and over, “neglecting” others. Furthermore, the categories whose
grammaticalization is preferred may impact the mode of grammatical-
ization of other categories. For example, in the area of the verbal
categories tense-aspect-mood, Bhat () described how languages
may be either tense-, aspect-, or mood prominent. The not overtly
expressed categories may be indirectly expressed through the well-
grammaticalized ones. In terms of grammaticalization, speech commu-
nities normally tend to maintain and only slowly change the typological
profile of their language by renewing and extending extant categories
rather than frequently abandoning extant categories and creating new
ones instead. This might be due to the higher degree of attention
afforded by speakers to the already extant categories whose expression
is obligatory versus not yet grammaticalized categories (cf. research on
thinking for speaking; e.g. McNeill ; Lai et al. ). Chafe ()
spoke of ‘florescence’ in languages. He suggests that, “[l]ike forests,
languages may develop toward a climax stage where particular combi-
nations of features, like plant communities, may flourish to define a
particular language type. I think it is useful to think in terms of the
florescence of linguistic features in this sense—the flowering of features
that come to dominate the form a language takes” (Chafe : ). As
examples, Chafe provided extremely elaborate pronominal prefixes,
and noun incorporation in Iroquoian languages that are both the result
of successive waves of grammaticalization. It is not hard to come by
examples from other categories and languages as well (cf. Narrog ).
In conclusion, there is a view of grammaticalization in which gram-
maticalization in essence is about grammatical paradigms, and in
which the motivation for grammatical change has to be sought prima-
rily in the grammatical system itself. Also, in this view, reanalysis and
analogy as mechanisms of change neatly complement language struc-
ture as the motivation for change. Some of the major figures in gram-
maticalization studies have rejected this view at least partially, mainly
on the grounds that for speakers of a language, grammar is at any point
already completely functional. An alternative to viewing grammar as
the motivation for grammaticalization is to view it as a constraint and
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
a model. This is the view we take in this book. As Dahl (: )
put it, “the likelihood for a certain grammaticalization process to
appear is at least to some extent dependent on structural properties
of the language.”
Further reading
Vennemann () made a sophisticated case for language change as lan-
guage improvement, and Kiparsky () for grammaticalization as gram-
mar optimization. Fischer (, ) argues that the synchronic system
motivates and determines the course of grammaticalization. Diewald and
associates (Diewald , ; Diewald and Smirnova , etc.) are close to
the group of Danish structuralists in considering paradigmatization as the
essence of grammaticalization and in positing linguistic conditions as triggers
for change.
6.3 Communicative needs or goals of the speaker
One common conception about some types of language change in
general, and grammaticalization in particular, is that the latter is a
result of the communicative needs or goals of the speakers of a lan-
guage. Both of the terms ‘communicative need’ and ‘communicative
goal’ have been used. The term ‘goal’ seems to presuppose a more active
role of the speaker, but a ‘need’ will only become effective in language
activity if it is turned into a ‘goal,’ so we can consider them here as
practically synonymic. If we conceive of this idea broadly, we can
distinguish between (a) an internally oriented need for expression of
the speaker and her thoughts, and (b) an externally oriented need to
communicate successfully within the social context in which she uses
language. Ultimately, these are two sides of one coin.
First, as for (a) internal need, within the context of grammaticaliza-
tion, Lehmann (: ) formulated this idea as follows: “Every
speaker wants to give the fullest possible expression to what he means.
The received grammatical devices are notoriously insufficient to ade-
quately express what he wants to say.” Furthermore, “it is no exaggera-
tion to say that languages change because speakers want to change
them . . . they do not want to express themselves the same way they did
yesterday, and in particular not the same way as somebody else did
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. COMMUNICATIVE NEEDS OR GOALS OF THE SPEAKER
yesterday.” The roots of this idea can be traced back to Gabelentz (:
), who attributed language change to two major forces, a ‘drive
towards ease’ (Bequemlichkeitstrieb), and a ‘drive towards clarity’ (Deu-
tlichkeitstrieb). In this case, it is the latter that is referred to.
A related idea is proposed by Heine et al. (: –). They argue
that grammaticalization is the result of ‘problem solving,’ and that it is a
creative process. Invoking the ‘principle of the exploitation of old
means for novel functions’ of Werner and Kaplan (: ), Heine
et al. () suggest that,
[b]y means of this principle, concrete concepts are employed in order to understand,
explain, or describe less concrete phenomena. In this way, clearly delineated and/or
clearly structured entities are recruited to conceptualize less clearly delineated or
structured entities, and nonphysical experience is understood in terms of physical
experience, time in terms of space, cause in terms of time, or abstract relations in terms
of physical processes or spatial relations. (Heine et al. : )
For example, future tense is a fairly abstract concept and in many
languages across the globe it is therefore expressed and grammaticalized
in terms of concepts of physical motion, that is, by means of movement
verbs for ‘to go’ or ‘to come’ (see Kuteva et al. (: –) for
examples). Heine’s idea is also directly connected to cognitive motiva-
tions (Section .), and will feature there again in more detail.
Secondly, a presumptive (b) externally oriented need, namely to
communicate more successfully in a social context, has been formu-
lated by Haspelmath under the term ‘extravagance’:
The speakers’ goal is not just being understood at the lowest possible cost, but rather
being socially successful with their speech. . . . [S]ocial success can also be achieved by
being extravagant, and I propose that this is the reason why some speakers introduce
innovations. . . . [S]peakers not only want to be clear or “expressive,” sometimes they
also want their utterance to be imaginative or vivid—they want to be little “extravagant
poets” in order to be noticed . . . ” (Haspelmath : –)
Thus, speakers may choose a more ‘extravagant’ expression such as by
means of instead of a simple preposition such as with, and if enough
speakers copy this novel way of speaking, by means of will eventually be
grammaticalized.
In Haspelmath’s () model, grammaticalization is based on in-
creasing frequency and habituation of an erstwhile lexical expression
(see also Section .). While ‘extravagance’ is the motivation for start-
ing the grammaticalization process, at the tail end the process of
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
routinization leads not only to the loss of erstwhile extravagance but
also eventually to the loss of phonological and morphological weight
and semantic contents (Haspelmath : –). This will then
prompt a new round of grammaticalization (see also Section .
above on cycles).
The idea of grammatical change as the indirect result of communi-
cative goals of speakers has an intriguing theoretical background in
Keller’s (, ) theory of the ‘invisible hand’ in language change.
According to Keller, language change is neither ‘teleological,’ that is,
driven by the purpose to achieve a specific goal such as language
improvement, nor purely mechanical, that is, the result of random
variation. It is a ‘phenomenon of the third kind,’ namely the unin-
tended result of many intentional actions of individual speakers.
For example, if a speaker chooses to use the relatively novel con-
struction by means of instead of with to express an instrument or
means, the motivation may be to be clearer, more expressive, or even
to be extravagant in comparison to simply using with. This choice is
goal-directed, that is, it serves to fulfill the speaker’s communicative
purposes. On the other hand, the speaker does not have any intention
to make by means of grammatical or to eventually make it a preposi-
tion. Thus, the grammaticalization of by means of is the unintended
outcome of the speaker’s intentional actions.
The idea of communicative goals motivating grammaticalization is
in principle compatible with having ‘mechanisms’ to implement those
goals but, as we have seen in the previous section, its main proponent
Haspelmath (a) is sceptical of mechanisms.
Haspelmath’s concept (a) of ‘extravagance’ and Heine et al.’s
concept () of ‘problem solving’ have different presuppositions.
Haspelmath’ idea of ‘extravagance’ will fit well in a scenario in which
the concept to be grammaticalized is already grammaticalized but the
speaker wants to express it differently in order to garner more hearer
attention. It would also fit in particularly well with a model in which
young adults rather than small children or adults are the instigators of
grammaticalization (cf. Hopper and Traugott : ). In contrast,
Heine et al.’s () concept would fit particularly well if the concept is
not yet expressed, that is, if a new category is created.
Another well-known potential motivation of language change in
speaker-external, social communication needs is Grice’s () Coop-
erative Principle, which assumes that speakers communicate with a
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. COMMUNICATIVE NEEDS OR GOALS OF THE SPEAKER
common purpose and adjust their contributions to conversation
according to that purpose. This overall cooperative principle results
in a number of maxims (Grice : ), which have been frequently
adduced to explain language change via inferences. The whole of
Section . is dedicated to this concept as applied to grammaticaliza-
tion. Therefore, we will not go into details here. However, suffice it to
state here that the idea of grammaticalization derived from the Coop-
erative Principle and conversational maxims can also be broadly under-
stood as change based on communicative goals and needs.
Finally, Andersen (: ), based on Jakobson () suggested six
communicative needs that may result in innovations. These are, ()
reference precision, () emotive expressiveness, () aesthetic aptness,
() conative (i.e. addressee-directed) effectiveness, () channel effi-
ciency (i.e. communication as a connection), and () code conformity.
The first three are what we have labeled as ‘internally oriented,’ and the
others as ‘externally oriented.’
The idea of communicative goals and needs has not been without
critics. The clearest rejection is formulated in Hopper and Traugott
() as follows:
Clearly “communicative need” is not a plausible motivation in most cases of gram-
maticalization, since not all languages express the same grammatical functions, and
even less do they express the same grammatical functions in the same way. Further-
more, to assume that as an older system becomes eroded it may cease to function at an
adequate level of communicative coherence and therefore must necessarily be revita-
lized (as suggested by Givón : cited in Bybee : ) is to posit a stage of
language such as is unknown, in other words it violates the uniformitarian principle
and is not empirically supported by the data. Most emphatically, languages are not
goal-oriented. (Hopper and Traugott : )¹
However, this rejection seems to assume that speakers intentionally try
improve or change languages, which are insufficient in one respect or
the other. Very few proponents of language change based on the
communicative needs or goals of the speaker would seem to claim that.
Rather, the consensus goes in the direction of Keller’s theory of the
‘invisible hand.’ Language changes through the cumulative effect of
local acts of communication by speakers who try to communicate
successfully with local goals. Also, the idea of speakers trying to be
¹ Hopper and Traugott’s () critique seems to be simultaneously directed at struc-
tural motivation (Section .) and communicative goals or needs (this section).
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
‘expressive’ is corroborated by the well-known high historical turnover
in what is known as ‘boosters’ or ‘amplifiers’ such as very or terribly (cf.
Peters ). The beauty of Keller’s (, ) ‘invisible hand’ is that
it is basically agnostic with respect to more concrete motivations. In
fact, basically an infinite number of communicative goals can be ac-
commodated by the ‘invisible hand’ theory. It could be ‘extravagance,’
it could be ‘problem solving,’ but it could also be something totally
different. For example, the speaker may want to squeeze in as much
information as possible in the short time window of the hearer’s
attention, thus eventually leading to shorter word forms. In this
broader sense, the communicative needs or goals of the speaker are
hard to pass over as potential motivations for grammaticalization.
6.4 Inferences and contexts
Inferences, leading from one meaning to another, can be labeled both
as ‘mechanisms’ and as ‘motivations,’ but the latter is more common in
the grammaticalization literature. The concept of inferences is tradi-
tionally associated with the field of pragmatics (cf. Hopper and Trau-
gott : –; Huang : –). Two theories have dominated
this field in the past two decades: Neo-Gricean pragmatics and Rele-
vance Theory. With respect to grammaticalization, the former has
played a much more prominent role, especially through the work of
Traugott, which culminated in the “Invited Inferencing Theory of
Semantic Change” (IITSC) model (Traugott and Dasher ).
Accordingly, this model will be the main topic of this section. Then
we will briefly deal with contexts of grammaticalization, and an alter-
native take from the vantage point of Relevance Theory. Note that the
idea of inferences motivating grammaticalization rests on the assump-
tion that the core of grammaticalization is semantic/functional change.
Inferences operate in semantic change in general, for example also in
lexical semantic change. But here we are interested in semantic change
associated with grammaticalization.
As already mentioned, the idea that inferences drive grammaticali-
zation is mainly associated with the work of Traugott (; Traugott
and König ; Traugott and Dasher ; Hopper and Traugott
; etc.). The original idea is that new grammatical meanings of
some linguistic expressions are often due to ‘pragmatic strengthening.’
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. INFERENCES AND CONTEXTS
‘Pragmatic strengthening’ means the strengthening (enhancement) of
the ‘informativeness’ (cf. Levinson : –) of some linguistic
expression through an additional inference associated with the expres-
sion. For example, the English conjunction while was originally purely
temporal and meant ‘at the same time as.’ In contexts where the
simultaneity of the two events connected by while was surprising or
unexpected, an inference of concessivity would arise. () is an early
example of an utterance where this inference is obviously present, that
is, the hearer/reader can infer an adverse relationship that was not
explicitly expressed by the speaker/writer.
() Early Modern English (seventeenth century; quoted from the
OED by Traugott and König : )
Whill others aime at greatnes boght with blod, Not to bee great
thou stryves, bot to bee good
‘While others aim at greatness that is bought with blood, you
strive to be not great but good.’
While being repeatedly used in such contexts then presumably led to
the conventionalization of a concessive meaning of the connective.
The idea of ‘pragmatic strengthening’ was further developed into the
“Invited Inferencing Theory of Semantic Change” model. This theory
presupposes three different types of meaning and three heuristics for
speakers that guide their choice of expression (cf. Traugott and Dasher
: –). The heuristics are:
(i) The Q(uantity)-Heuristic: “Make your contribution as infor-
mative as required, and imply no more thereby,”
(ii) The R(elevance)-Heuristic: “Say/write no more than you must,
and mean more thereby,” and
(iii) The M(anner)-Heuristic: “Avoid prolixity”; or: “Specially
marked, complex expressions warn ‘marked situations’.” For
example, if I use a longer and more elaborate expression, instead
of an available shorter one, I might imply that I am conveying a
specific meaning that is not explicit with the shorter expression.
As for the type of meanings, the first type is ‘coded meanings.’ These
are the conventional meanings of a certain linguistic expression at a
given time. Their opposite extreme are ‘utterance-token meanings.’
These are meanings that arise “on the fly” in individual utterances,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
that is, these are novel, spontaneous uses of some expression. In the
middle of them are ‘utterance-type’ meanings. These are conventiona-
lizations of novel uses or ‘utterance token meanings’ that are common
in specific language communities, but can be canceled.
For example, expressions of temporal precedence are prone to be
reinterpreted, and given an additional meaning of concessivity or
causality. This is the case with Modern English since, which has in
some contexts, as in (), a purely causal reading, equivalent to ‘because.’
() Since I won’t be there to pick up my daughter tomorrow, I just
wanted to say thank you now.
No temporal reading is possible. In fact, the main clause I just wanted to
thank you now even temporally precedes the first clause with since.
However, the original meaning in Old English was a purely temporal
one, ‘after,’ with the event in the since-clause preceding the event in the
main clause. This meaning is still preserved in other contexts, especially
when the events in both clauses are in the past, as in ().
() I haven’t been able to score a single goal since the season started
last fall.
Here, a causal reading does not make sense. In contrast, a sentence
like () allows both interpretations for since, both ‘because’ and ‘after.’
() I haven’t been able to score a single goal since I had an ankle
operation this spring.
Clearly, the difference between () and () is not a structural one, but
results from additional inferences of causality between the two clauses
that can be attributed to () but not to (). As a matter of world
knowledge, we know that an ankle operation will cause a decline in
sports performance, so both speakers and hearers may impute such a
reading to the relationship between the two events. Among the heuris-
tics, it is primarily the R-Heuristic “Say no more than you must, and
mean more thereby” that drives the inference. This heuristic has stepped
in place of the ‘Principle of Informativeness’ in Traugott’s earlier work.
In the case of since, a speaker explicitly verbalizes only a temporal
relationship but may in fact mean a causal relationship beyond that.
The Q-Heuristic is a factor that delays change, while the M-heuristic
is secondarily in play when the new use is especially noticeable or
conspicuous (cf. Traugott and Dasher : ). In the case of since
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. INFERENCES AND CONTEXTS
in () this is hardly the case, since for Modern English speakers, the
causal use is already established. In terms of the “Invited Inferencing
Theory of Semantic Change” model, since has already reached the stage
of ‘coded meaning.’ () is a possible early example from a time when
the causal reading was presumably still unconventional, that is, prob-
ably an ‘utterance-token,’ and at best an ‘utterance-type’ meaning.
() Old English (King Alfred’s Orosius, as quoted in Mitchell
(: ))
ϸa, siϸϸan he ire wæs and gewundod, he
then, after/since he angry was and wounded, he
ofslog micel ϸæs folces
slaughtered much of-that troop
‘Then, since/after he was angry and wounded, he slaughtered
much of that troop.’
A case parallel to ‘since,’ but where the causal relationship is not yet
established even in Modern English, is after as in ().
() After we read your novel we felt greatly inspired. (Hopper and
Traugott : )
After in () can have a ‘because’ reading (Traugott and König :
). However, there is probably no larger community of English speak-
ers for whom the causal interpretation is a coded meaning that cannot
be canceled in specific contexts. For most communities of English
speakers, the causal reading of after is either at the stage of ‘utterance
token meaning,’ if the use is merely spontaneous, or of ‘utterance type
meaning,’ if the use is to some degree established.
There is considerable overlap between the model of change through
invited inferences as presented above, and the ‘context-induced rein-
terpretation’ or ‘context extension’ model (Heine et al. ; Heine
) discussed in Section ... In the ‘context extension’ model,
invited inferences operate at the stage of the ‘bridging’ context. They
correspond to ‘utterance-token’ and ‘utterance-type’ meanings, such as
in examples (), (), and () in contrast to the ‘coded’ meanings of
the switch context, such as example (). The main difference between
the two models is one of perspective: While the context extension
model highlights and differentiates various linguistic contexts and the
speaker’s cognitive processes, the Invited Inferences Theory highlights
speaker–hearer interaction, and differentiates between the inferences
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
involved. One can also say that the context extension model has the
main function of identifying and comparing stages of grammaticaliza-
tion across different languages, while the Invited Inferences Theory
focuses on the development of individual forms or constructions.
Despite a few quibbles with specific aspects of the Invited Inferences
Theory (cf. Traugott a: –; also Kearns ), this model has been
broadly accepted by scholars adopting the same Neo-Gricean approach to
pragmatics. In contrast, it has been viewed rather critically from the
perspective of Relevance Theory. Nicolle (, ) has levered some
criticism against both Traugott’s IITSC and Heine’s context extension
model. Decisively, for him, grammaticalization does not need a step by
step process via generalization of inferences, or bridging contexts leading
to the switch context. Nor are there different types of inferences necessary.
All that is needed for a grammatical meaning to emerge is a ‘Privileged
Interactional Interpretation’ that adds ‘procedural’ information to the
lexical item’s ‘conceptual’ information. And this can happen spontane-
ously and directly. More recently, even the whole concept of conversa-
tional implicatures has been put into question (Lepore and Stone ).
Traugott (a) herself has recently suggested further distinguishing
different types of invited inferences: local inferences, discourse-
structuring inferences, and turn-taking inferences that may be responsi-
ble for different types of semantic and functional changes, not only in
grammaticalization.
6.5 Cognition and conceptualization
While the idea of inferences being the driving force behind grammat-
icalization is mainly associated with the work of Traugott and associ-
ates, the idea that cognition is the driving force is mainly associated
with the work of Heine and associates. In contrast to Traugott as a
scholar of English language history, Heine and his associates studied a
variety of African languages from a typological perspective. The fun-
damental observation was that the same source-target relationships in
grammaticalization can be found in a variety of unrelated and other-
wise structurally dissimilar languages. This could only be explained by
common principles of human cognition driving grammaticalization
irrespective of language family, language area, or the structure of the
language where the change occurred.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. COGNITION AND CONCEPTUALIZATION
Many grammatical categories in the languages of the world are
derived via grammaticalization from a very limited set of sources
relating to basic human experience. In Section ., we have already
given the example of possessive constructions from Heine (a).
Comparative constructions are another one. Most expressions of com-
parison in the languages of the world have grammaticalized through
one of the source schemas listed in (). X is the comparee, Y the
predicate, and Z the standard of comparison (Heine a: ).
() Source schema Label
X is Y surpasses Z Action
Y is Y at Z Location
X is Y from Z Source
Y is Y to Z Goal
X is Y, Z is not Y Polarity
X is Y, then Z Sequence
X is Y (like) Z Similarity
X and Z, X is Y Topic
As can be seen in (), the schemas have to do with what one does,
where one is located, where one moves from or to, and the like. They
determine the particular linguistic shape a given comparative construc-
tion is going to take in a language. Many of them are shared with
grammatical constructions in other categories, such as possession (as
we saw in Section .) or tense-aspect (Heine : ). The first five
schemas in the list are particularly common in the languages of the
world, while the others are rather negligible.
() and () are examples for the realization of the action schema in
genetically unrelated languages. () is from a Niger-Congo (African)
language and () from Thai.
() Yoruba (Kuteva et al. : )
Ile mi kere ju tiwon
House my small exceed theirs
‘My house is smaller than theirs.’
() Thai (Kuteva et al. : )
Khaw jaj kwaa phom
he big exceed me
‘He is bigger than me.’
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
Heine (a: –) notes that there is an areal bias in the distribution
of source schemas, suggesting that language contact plays an important
role in the “choice” of schema, but the important fact is that the source
schemas seem to be in principle available to speakers of any language.
More generally, beyond the expression of grammatical categories,
Heine’s approach is based on the following five premises:² () The
basic function of language is to convey meaning; () form and struc-
ture of language are not arbitrary but motivated; () ‘external’ expla-
nations of language structure and change are superior to ‘internal’
motivations, since language use itself is externally motivated; ()
language is a product of history and can be explained with reference
to that history; and () language change proceeds unidirectionally
from the lexical to the grammatical. There are several ways to think
about the relationship between language and cognition; for example
language determines cognition, or language as an innate faculty, or
that language equals cognition. Heine assumes that language merely
reflects conceptualization, and this reflection also leaves its imprint on
grammaticalization.
At the core of grammaticalization is semantic change, which relies
on semantic extension based on unidirectional ‘conceptual transfer’
from more concrete to more abstract meanings. Grammatical concepts
are generally more abstract, derived, and relational, tend to have no
meaning in themselves, that is, only in conjunction with other words;
and they contribute to structure rather than to cognitive representation.
Thus, grammaticalization can be seen as a creative process that serves
to solve the problem of how to express the more abstract domains of
grammar by making use of more concrete concepts. Common sources
are from basic domains of human experience, such as body parts,
location, movement, or action, and they come to express meanings in
more abstract domains such as time and quality.
For example, cross-linguistically, numerals are often derived from
expressions for ‘one hand’ (five), ‘two hands’ (ten), and the expressions
for individual fingers for numbers in between. Adpositions for ‘on’ or
‘above’ may be derived from body part nouns like ‘head,’ adpositions
for ‘behind’ or ‘after’ may be derived from body part nouns like ‘back’
² The following paragraphs are based on Chapters , , and from Heine et al. ()
and Chapter from Heine (a).
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. COGNITION AND CONCEPTUALIZATION
or ‘butt,’ adpostions for ‘under’ from landmarks such as ‘earth,’ and
aspectual notions like inception, continuation, or result may be derived
from verbs of motion such as ‘go’ or ‘come.’
Not surprisingly, then, the relationship between the source and
target of concepts related through grammaticalization can often be
described in terms of metaphor, especially of so-called ‘experiential’
metaphor, that is, the kind of metaphor that conceptualizes abstract
concepts through concrete ones. Thus, ‘butt’ can be understood as a
metaphor for ‘behind’ or ‘head’ as a metaphor for ‘above.’ On the other
hand, the actual change from source to target meaning is assumed to be
gradual and to take place in small contiguous (metonymic) steps. Heine
et al. (: ) write that “[c]onceivably, metonymy is the more basic
component of this process in that metaphor is grounded in meto-
nymy.” However, metaphor may be “responsible for defining the direc-
tion of conceptual change” (Heine et al. : ). In this manner,
Heine et al. assume that metaphor and metonymy are, as a rule, both
involved in meaning change in grammaticalization.
As an example, in the development of since from temporal to causal
conjunction (cf. Section .), each step in the development from tem-
poral, (), to ambiguously temporal-causal, (), to unambiguously
causal, (), can be taken as a contiguous, metonymic extension of
meaning.
() I have done quite a bit of writing since we last met.
() Since Susan left him, John has been very miserable.
() Since you are not coming with me, I will have to go alone.
Thus, the metonymic extension of meaning can be said to be more
basic. However, when directly comparing the source meaning in ()
with the target meaning in (), a TIME-to-CAUSE metaphor can be
identified in which a sequence of events in time is used to metaphor-
ically refer to a sequence of events in a causal relation. While this
metaphor is not found in the actual change, which is metonymic, it
may nevertheless be a cognitively guiding principle, whose effects can
be found in many languages across the world.
In Heine’s model of grammaticalization, the second force that shapes
the process besides conceptualization is communication. This includes
the speaker–hearer relationship, but also cultural, religious, and socio-
political factors (cf. Heine a: –).
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
Further reading
The role of metaphor in particular has been contentious in gramaticalization.
Chapter of Sweetser () is a classic on semantic change in the English
modals that is based on the concept of metaphor. Stolz () is a short
monograph specifically on grammaticalization and metaphor. Among the
founding figures of grammaticalization studies, besides Heine, Bybee espe-
cially entertained the idea of metaphor as an important mechanism of
meaning change in grammaticalization (Bybee et al. : ch. ; Bybee and
Pagliuca ; and elsewhere). Hoefler and Smith (), based on a model
of ostensive-inferential communication, identify metaphor and reanalysis
domain-general mechanisms that are more basic and ancient than language.
They claim that these cognitive mechanisms are not only decisive for gram-
maticalization but can also provide a unified explanation for language change
and the origin and evolution of language in general. In contrast, Waltereit
(), for example, argues that only metonymy and not metaphor is rele-
vant for grammaticalization, and Juge () even presents a strong invective
against the idea of metaphor operating in grammaticalization.
Langacker (, , and elsewhere) dealt with grammaticalization
within his framework of ‘Cognitive Grammar,’ but mainly in terms of sub-
jectification (cf. Section .).
6.6 Frequency of use and language processing
Another classic approach to what drives grammaticalization is motiva-
tion in terms of frequency. This approach is mostly associated with the
research of Bybee and associates. The following paragraphs are there-
fore based on Bybee (a, , ) unless otherwise stated.
The basic idea is that grammaticalization is a process of ritualization
crucially involving repetition (cf. Haiman ), also referred to as
routinization (Detges and Waltereit ; Bybee a: , ).
Ritualization in general, that is, as a cultural phenomenon, implies
habituation, automatization, reduction of form, and emancipation, in
short, the development of symbolic function. In the case of language
change, and specifically grammaticalization, a certain pattern may start
out semantically and functionally transparent (analytic), but through
repetition it becomes ritualized and establishes itself as a conventional
(synthetic) pattern of its own. Frequent repetition brings about five
changes that are typically associated with grammaticalization:
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. FREQUENCY OF USE AND LANGUAGE PROCESSING
(i) semantic weakening through routinization (corresponding to
habituation);
(ii) phonetic reduction (corresponding to reduction of form);
(iii) a higher degree of autonomy as a grammatical structure, that is,
dissociation from the source (corresponding to automatization);
(iv) loss of semantic transparency (corresponding to habituation):
The internal structure of a grammaticalizing item becomes
intransparent. Semantically, bleaching occurs, while it may
simultaneously gain new pragmatic functions; and
(v) entrenchment in the language and preservation of structure
(corresponding to automatization and emancipation).
In quantitative linguistics, usually two types of frequency are distin-
guished, type frequency and token frequency. Type frequency in the
case of grammaticalization refers to the type and number of contexts in
which a linguistic form can be used, including for example the number
of verbs with which an aspect marker co-occurs. It corresponds to
‘context extension’ in the model of grammaticalization presented in
Chapter of this book, and instigates grammaticalization. Token fre-
quency refers to the frequency with which an individual form may be
used. It becomes relevant when grammaticalization has kicked in. For
one thing, high token frequency of one item may lead to its phonetic
reduction, since if the same word is used twice or more in the same
stretch of discourse in conversation, it becomes predictable and re-
quires less pronunciation effort. For another thing, high token fre-
quency also leads to loss in semantic substance (desemanticization; in
Bybee’s writings the preferred term is ‘bleaching’). It also leads to the
preservation of frequently used structural patterns, such as irregular
verb forms, as a result of storage in memory.
Another aspect of grammaticalization through high-frequency usage is
‘chunking’: The central noun or verb in a construction loses its verb or
noun properties. Linguistic forms that frequently co-occur become a unit,
and do not allow other forms to intervene. ‘Chunking,’ then, is the fusion
of linguistic forms that frequently co-occur. With processing efficiency
rising, the pronunciation of each form becomes shorter. In parallel, the
independent meaning of each linguistic form co-occuring becomes weaker
and a semantic-pragmatic interpretation gets attached to the whole unit.
Very generally, in this approach, grammar is seen as crucially influ-
enced by language processing in language activity. Those structures
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
that are easy to process for hearers and speakers will grammaticalize
more readily.
As an example of grammaticalization involving high usage frequency,
we give a short sketch of the development of the Old English verb cunnan
to the Modern English modal can as presented in Bybee (a).
Cunnan was originally a lexical verb with the meaning ‘know.’ It
could already have verbal complements in Old English, denoting a
‘mental ability’ (as Modern English can in I can understand what you
say). It then extended to general ‘ability’ in Early Middle English (as can in
I can lift pounds), and finally to general possibility (‘root possibility’)
(as can in Our service page can be accessed on any web browser). Each step
of semantic generalization means more contexts of use (higher type
frequency), involving more types of subjects and verbs in the complement
with which the auxiliary can co-occur. For example, while mental ability
and ability are basically restricted to human beings as subjects, root
possibility has no such restrictions. Likewise, while mental ability is
restricted to verbs of intellectual states and activities, skills, and commu-
nication, ability may include all kinds of activities and actions, while root
possibility is compatible with basically all kinds of verbs.
Examples () to () from Bybee (a) illustrate the gradual exten-
sion of uses. () is the lexical use in Old English, () the ability use with
a mental verb, () the general ability use, both in Middle English, and
() the general possibility use in Early Modern English.
() Old English (Ags. Gospel of Matthew xxii) (Bybee a: )
Ge dweliað and ne cunnon halige gewritu
‘You are led into error and do not know the holy writings.’
() Middle English (Chaucer: Monk’s Tale, ) (Bybee a: )
Ful wys is he that kan hymselven knowe!
‘Completely wise is one who knows himself!’
() Middle English (Chaucer: Nun’s Priest’s Tale, ) (Bybee
a: )
But I wol passe as lightly as I kan.
‘But I will pass by as lightly as I can.’
() Early Modern English (Hawes: The Passetyme of Pleasure xvi.
xlix) (Bybee a: )
No worldely thyng can be wythout stryfe.
‘No worldly thing can be without discord.’
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. FREQUENCY OF USE AND LANGUAGE PROCESSING
Bybee (a) notes a number of remarkable features of this develop-
ment. First, the verb classes with which cunnan/can was compatible
expanded gradually through similarity with those verbs that were
already used with cunnan/can. There was no sudden leap. The same
holds for subjects. Initially only compatible with human subjects, it
came to be used with inanimate subjects that metonymically stood for
humans, such as the body parts eyes and heart, and then finally
extended to inanimate subjects. () is a Middle English example of
wit standing metonymically for the person.
() Middle English(Chaucer: Troilus and Criseyde, Book IV, )
(Bybee a: )
As ferforth as my wit kan comprehend.
‘As far as my wit can comprehend.’
Secondly, cunnan/can also developed high frequency collocations
such as can tell or can see in Middle English. The emergence of such
high frequency collocations strengthened the entrenchment of can as
an auxiliary and led to further desemanticization. Thirdly, in the
general possibility use can was now also able to have stative predicates
(cf. ex. ()) and passives as complements, further extending its
contexts of use. Finally, there was also a loss of morphological and
phonetic substance, as cunnan lost its final inflectional syllable that was
still present in Old English and is further reduced to [kŋ] or [ŋ] in
high frequency contexts in Present-Day English, such as after the
subject pronoun I. Overall, the most salient feature of cunnan’s devel-
opment is continuous desemanticization or generalization through
habituation.
Bybee’s approach is embedded in so-called Usage-Based Theory, an
approach with the fundamental tenet that language structure emerges
from language use. Grammaticalization is a central part of this
approach. Within it, the mechanisms of frequency can explain the
development of grammatical structure and meaning. Furthermore,
since the mechanisms of frequency are based on universal properties
of language processing of humans, it is the mechanisms that are
universal. In short, it is the diachronic mechanisms that lead to the
structures which are universal, and not the structures themselves.
Although the frequency account of grammaticalization has been
immensely influential, it has also been subject to some criticism.
Heine et al. (: –) and Heine and Kuteva (: –) remark
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
that often very frequent words do not grammaticalize while less fre-
quent ones do. Furthermore, there are also so-called low-frequency
grammaticalizations such as the ‘threaten’ and ‘promise’ proximatives
in the European languages, and cases where the lexical word is more
frequent than the grammaticalized ones (note also the English progres-
sive at the first stage of its development as described in Section ..
and by Kranich (a)). Mair () analyses some of the low fre-
quency grammaticalizations, and observes that grammaticalization in
written registers is generally low-frequency. Traugott (a: )
notes that frequency is unlikely to be a motivation for grammaticaliza-
tion since () one would first have to explain the motivation for the rise
in frequency, and () in actual historical change, frequency often rises
after grammaticalization or does not rise at all.
Finally, using a quantitative approach based on English text corpora,
Saavedra (, chs , ) finds that there is a striking difference between
the analysis of synchronic and diachronic data with respect to fre-
quency. He finds substantial support for the claim that grammaticali-
zation and token frequency are positively correlated in synchrony, as
grammatical elements involve higher token frequencies than lexical
ones. In contrast, when looking at grammaticalization diachronically,
that is, as an ongoing process, there was no clear correlation, and what
is more, many grammatical elements studied by Saavedra were less
frequent than their lexical counterparts (Saavedra : –).
Further reading
Notable publications besides the work by Bybee and associates based on or
discussing the frequency approach to grammaticalization include Krug
(), Hoffmann (), Pustet (), Peng (), Szmrecsanyi (),
Schulte and Blas Arroyo (), and Saavedra (). Hawkins (, and
elsewhere) is remarkable for developing a unique theory about how language
processing determines language structure, whereby he also refers to gram-
maticalization. Ritualization is also referred to as ‘routinization’ (Detges and
Waltereit ; Bybee a: , ), by which a linguistic sign sequence
is made more frequent, thereby progressively ruling out alternatives and
choices (Detges and Waltereit ).
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. DISCOURSE
6.7 Discourse
Discourse has been proposed as a driving force of grammaticalization on
two different levels. For one thing, grammaticalized structures may
develop out of discourse structures, in particular, conversational structures
in spoken language. For another thing, ‘discourse’ can be taken in a very
broad meaning, including the language-dependent social-cultural interac-
tion at the level of a larger community of speakers. Section .. deals with
the forces of discourse operating at the level of conversational interaction
(or text construction), as has been the subject of research by several
scholars, while Section .. takes a broad view of ‘discourse,’ particularly
referring to the work of one scholar, namely Ariel ().
6.7.1 Grammaticalization from discourse structures
At the most concrete level, grammaticalizations may be motivated by,
and emerge directly out of conversational routines. Couper-Kuhlen
() presents three concrete examples for the process. They have in
common that grammaticalization is the integration or condensation of
an originally bipartite structure in turn taking to a single sequence.
Note that the examples are from contemporary spoken language and
are not diachronic, and thus simply illustrate how a change might
proceed diachronically based on synchronic observations.
‘Extraposition’ is a construction in which the nominal subject is filled
in English by a placeholder subject it, which is followed by an evaluative
adjective or noun and a complement clause presenting the proposition
that is the subject of the evaluation. An example would be It’s sad
having to spend my youth for people like this. Couper-Kuhlen (:
–) presents a possible conversational precursor to this construc-
tion in conversation (). The example has been abbreviated.
() Nan: They are so cute #yeah they really.
they were just (.) ve:ry .hhhhh very very sweet with me:
a:nd it was so funny
in fact one of the kids came up to me.
[ lines omitted]
.hhhhh en then afterwards Ra:lph
came up and he said (.)
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
I:’d like (.) Nancy?
he said I’d like to (.)
take you over to Shakey’s and buy you a #bee:r. (.)
uhhhh #huh [#huh.h] hhh
The highlighted clause it was so funny refers forward to the following
story about a particular student in the speaker’s psychology class that
ends with the speaker’s laughter in the last line quoted here. According
to Couper-Kuhlen (: ), the sequence represents a conversa-
tional routine consisting of two actions, namely (a) a prospective
evaluation, followed by (b) the delivery of what is being evaluated.
These two actions are not syntactically linked in (). However, they
can be integrated syntactically, especially when a single situation is
described. This is the case in (), where the two interlocutors discuss
TV reporting in the aftermath of Robert Kennedy’s assassination in Los
Angeles.
() Nan: well I had turned it on when I first got u:p just to see:
how thin:gs were: progressi:ng but the thing was so sad
and
all that horrible sad music they kept
[ lines omitted]
[.t.hhh we:ll ↑I think it’s sad
that they don’t uh: .h allo:w-you know the families
at least the decen[cy of hav]ing some privacy.
Emm: [eeyah]
This integrated form yields the grammatical structure of extraposition.
Couper-Kuhlen (: ) views it as the “coalescence and entrench-
ment of the more loosely organized ‘evaluative preface + telling’ con-
versational routine.” She provides similar examples for integration of
concessive conversational routines into concessive clauses and a ‘rec-
ognition search sequence’ routine into left dislocation. In a metaphor-
ical way of speaking these are ‘vertical-to-horizontal’ paths of
grammaticalization. That is, text that is stretched vertically across a
number of sequences gets integrated horizontally into just one
sequence. The author admits that it is unclear whether this scenario
of integration of bipartite structures is applicable to non-bipartite
structures as well, or even to all bipartite structures, but at least it is a
plausible avenue for grammaticalization of some structures.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. DISCOURSE
On a more abstract level than conversational routines, motivations for
grammaticalization can also be sought in discourse strategies or rhetoric
strategies. Detges (, ) and Waltereit (, ) propose
discourse strategies as the driver of grammatical change, including
grammaticalization. Their approach is inspired by Keller’s () ‘invis-
ible hand’ theory of language change (Section .), according to which
language change is the unintended consequence of intentional actions by
speakers. In this case, speakers choose specific discourse strategies in
order to speak successfully, and these strategies lead to change. As an
example, Waltereit (, ) presents the development of the French
adverb bien ‘well’ into a concessive conjunction as in (), a modal
particle as in (), and a discourse marker as in ().
() French (Waltereit : )
J’y ai participé, bien que je sois conscient du risque.
‘I took part, even though I am aware of the risk.’
() French (Waltereit : )
Vous avez bien reçu mon message?
‘You got my message, didn’t you?’
() French (Waltereit : )
GAS: non le problème oui c’est c’est la politique d’immigration
‘No the problem, yes, is the immigration policy’
CG: voilà oui d’accord bien non parce que votre formule elle
peut être ambiguë oui elle peut être interprétée hein y en a plein
qui disent le problème c’est l’immigration.
‘Right, I agree, well no, because what you say may be ambiguous: it
may be read, well, many people say the problem is immigration itself’
As a concessive conjunction, bien que has scope over a clause (‘I am
aware of the risk’), as a modal particle it has scope over the whole
sentence (‘You got my message’) with the interactional function of
“orientating the question towards a positive answer” (Waltereit :
). Finally, as a discourse marker it has a metatextual or interactional
meaning and free scope extending over a relevant stretch of discourse.
The development of a discourse marker is in the view of this book
and of Waltereit (: ) himself not a case of grammaticalization
(see Chapter ), so we focus here on the two other uses. The develop-
ment as a modal particle, according to Waltereit (: ), is due to a
strategy of scalar argumentation. A speaker claims that, contrary to
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
what the hearer expects, a state of affairs is the case to a high degree,
implying the weaker fact that the state of affairs is the case at all. () is
a possible example for the genesis of the modal particle use from the
thirteenth century. The speakers deny not having heard of Troy’s
destruction by claiming they have heard ‘a lot’ about it.
() Old French (Robert de Clari, La Conquête de Constantinople
) (Waltereit (: )
Et mesires Pierres respondi: ‘Ba!’, fist il, ‘de n’avés vous oï
comment Troies le grant fu destruite ne par quel tor ?—‘Ba
ouil !’, fisent li Blak et li Commain, ‘nous l’avons bien oï dire.’
‘And Mylord Pierre answered: “Ba,” he said, “haven’t you heard
about how Troy the great was destroyed and in which way this
happened?”—“Of course,” said Blak and Commain, “we heard
clearly/a lot about it”.’
As for the concessive use, bien was already used to concede a point in
argumentation in Old French without being part of a conjunction, as in
the example from a twelfth-century text in ().
() Old French (Chevalier de la Charrette, , TFA) (Waltereit
: )
Dahez ait qui vos oï onques, Ne vit onques mes, que je soie! Bien
puet estre, mes je pansoie, Que le gué me contredeïstes; Bien
sachiez que mar me feristes.
‘Cursed be whoever saw or heard you, even if it’s myself! It is well
possible that you forbade me the ford, but I was deep in thought.
You should know well that you did me wrong.’
Bien in () argumentatively strengthens the first proposition ‘It is pos-
sible that you forbade me [to cross] the ford’. However, it is overridden by
mes ‘but’ in the following clause. That is, no matter how strong the force
of the first proposition is, it cannot outweigh the force of the second
proposition. This is essentially already a concessive construction, and
repeated use of bien in initial position in the concessive clause will have
established the association with concessive meaning. In fact, according to
Waltereit (: ), bien was already established as a concessive con-
junction in Old French in collocation with se ‘if ’ as se bien.
Overall, then, for Waltereit (, ), grammar is diachronically
derived from discourse structures, and the specific outcome of
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. DISCOURSE
grammaticalization is determined by the discourse strategies for which
a lexical (or already grammaticalized) item is used.
Further reading
Other notable contributions to the study of grammaticalization as motivated
by discourse include Traugott (b), Haselow (), and Van Rompaey
().
6.7.2 Discourse as the arbiter
In the case of Detges and Waltereit, discourse is conceptualized as
driving grammatical change instead of other factors, whose validity is
questioned by them, especially cognition. Ariel (), in contrast,
acknowledges the role of cognition and inferencing, and instead locates
discourse at a level superordinate to those other factors. For her, other
factors ‘propose,’ and a certain type of discourse is the ‘arbiter’ between
practically limitless choices of linguistic expression. That is, for Ariel,
discourse is neither a ‘motivation’ nor a ‘mechanism’ but has a third
role, presiding over other factors that have been identified as motiva-
tions and mechanisms. Since this idea is also highly relevant for the
current chapter, we will discuss it here.
Ariel () comes up with a model for grammaticalization in which
all of the ‘driving’ factors from Sections . to ., that is the language
system, communicative goals, cognition and inferences, frequency and
discourse play a role. She views speakers’ communicative goals as “a
most important impetus for the formation of discoursal patterns”
(Ariel : ). In accordance with Keller’s invisible-hand theory of
linguistic change (Section .), speakers act with communicative goals
and local communicative intentions but not with the intention to create
grammatical conventions. Instead, new grammatical conventions are
the unintended side product of their actions. Cognition (Section .)
through processes such as metaphor and metonymy and principles
like iconicity, on the one hand, and inferences on the other hand
(Section .) are identified both as mechanisms ‘proposing’ innovations
(e.g. Ariel : ) and as the initial ‘filters’ in a ‘filtering chain’ (Ariel
: ). New uses enter discourse and form new discourse patterns
(Section .), but in order to actually get entrenched, they need to
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
acquire saliency. The most common form of saliency is frequency
(Section .). For Ariel (: ), the importance of repetition and
frequency of use is also plausible in terms of a neural theory of
grammar, since connections between memory units need to be
strengthened through frequent firing. However, it is clear that there is
no one-to-one relationship between frequency of words and grammat-
icalization (Ariel : ). Ariel (: –) also suggests that it
is not always the best motivated pattern in terms of the speaker’s
communicative goals that gets grammaticalized. Often, it is a relatively
arbitrary pattern that happens to gain salience, for example because an
influential speaker or an influential speech community or medium
created a precedence and other speakers followed out of conservatism,
and because of the mechanism of priming in discourse. Speakers may
even change language by conservatively trying to follow what they
perceive as the rules of a language (Ariel : ).
The extant grammar of a language (Section .) is considered as
another filter for possible new ways of expression. Yet other filters
include culture and society (e.g. taboos, male bias, politeness), leading
to different outcomes of grammaticalization in different languages. But
even when there are no stark contrasts in terms of grammar or culture
between two given languages the outcome may be different, just
because of the coincidence of which pattern happened to become
salient in the language in question. For example, the temporal conjunc-
tion while acquired concessive meaning in English while its German
cognate weil ‘because’ acquired causal meaning. Both readings are
cognitively equally plausible and corresponding implicatures are
equally available in temporal uses, but one inference became salient
in one language, and the other one in the other language. According to
Ariel, the decisive factor is which reading first acquires the status of the
‘salient discourse pattern’ in a speech community, and this may be due
to relatively arbitrary circumstances.
Perhaps due to the fact that it was part of a publication not obviously
related to language change or grammaticalization, Ariel’s () con-
cept of grammaticalization has not received the response from the
grammaticalization research community that it arguably deserves.
Accordingly, there is also no corresponding critical feedback that we
are aware of. Through her broad perspective Ariel () offers a
comprehensive and unique view of the flow of grammaticalization.
However, overall her model is not yet particularly well fleshed out.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
. CONCLUSION
A potential weakness with respect to the topic of this chapter is the
relative lack of explicitness about motivations, since for Ariel most of
the potential ‘motivations,’ including discourse as her main focus, are
in fact only ‘filters.’ Other concerns are the relative lack of concrete
examples from grammar (most of Ariel’s examples are from the lexi-
con), and a lack of concrete examples for the stage of explicate infer-
ences, which is crucial for grammaticalization in her model.
6.8 Conclusion
In this chapter we have discussed the factors that are most commonly
assumed to drive grammaticalization. They include so-called ‘mechan-
isms’ and ‘motivations.’ Reanalysis and analogy have been overwhelm-
ingly cited as ‘mechanisms.’ Cognitive processes such as metaphorization
and metonymization as well as inferences, frequency, and discourse have
been referred to both as mechanisms and as motivations. The commu-
nicative goals of the speaker have only been named as a motivations.
In fact, the communicative goals of the speaker, specifically em-
bedded in an ‘invisible hand’ theory of language change as the unin-
tended results of intentional actions à la Keller (), are the most
broadly accepted ‘motivation,’ perhaps because they do not obviously
compete with other proposed motivations. Other factors, especially the
language system, frequency, cognition, and conceptualization (in par-
ticular metaphor) and inferences have been subject to controversies
among their proponents. As we have seen in Section ., this is
especially true for reanalysis and analogy, whose validity has been
altogether denied by some scholars.
However, all of these potential mechanisms and motivations are not
necessarily mutually exclusive but may as well be complementing each
other. In fact, they are so different from each other that there is no
reason to assume that any of them logically excludes other ones. Each
may have a role in a chain of events that leads to grammaticalziation,
and moreover, the role and the weight of each factor may differ in each
concrete case of grammaticalization.
Here we present a rough outline of how these factors may collabo-
rate. We assume that the starting point is the speaker (or writer), since
there cannot be grammaticalization without speech production. We
thereby take for granted that the speaker also acts as a hearer (or
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
WHAT DRIVES GRAMMATICALIZATION? MECHANISMS AND MOTIVATIONS
reader) in communication and her speech output is influenced by
speech input. In some cases, a specific interpretation of the speaker’s
speech input may be crucial for an innovative output, but nevertheless
it can only be by her role as a speaker that she initiates change. The
speaker has a large number of communicative needs, including the
need to express some pre-linguistic concept verbally, at times the need
to communicate a concept in a way that attracts attention, at times the
need to express something only implicitly, at yet other times to be brief
in his or her verbal expression, and, in most instances, to cooperate in
conversation and conform to the norms of the speech community (cf.
Section .). As we have seen in the previous sections, each of these
potential needs has been claimed to potentially lead to change, and we
assume here that this is indeed the case.
The tools to respond to these needs are in the domain of cognition,
pragmatic inferencing, and the processing of speech. While the need to
express pre-linguistic concepts or express an extant linguistic concept
more effectively has been associated mainly with cognitive processes,
implicit expression and cooperation in conversation have been associ-
ated mainly with inferencing. Implicitness and shortening of speech
signals are also associated with language processing. Frequency comes
in when an innovation gains currency in a speech community. It is a
mechanism rather than a motivation for change, but it obviously plays
an important role in entrenching change in an individual speaker and
in a speech community, and can even ‘drive’ further change, especially
the kind of phonological and morphological changes that are typical for
high-frequency grammaticalizations.
We regard the extant grammar of the language in question and
analogy not as a motivation for change but as potential constraints
and models for the target of change. Clearly, most new patterns will
adhere to the grammatical structure of the language. Among those
patterns that are cognitively available to grammaticalize a category,
those that already have related means of expression, lexical or gram-
matical, and conform to the grammar of a language are naturally more
likely to prevail. Moreover, there is a strong tendency for speakers to
repeatedly grammaticalize categories that are already grammaticalized
in their language with different semantic nuances, taking extant struc-
tures as a model (Section .), or let grammaticalized categories
extend further, rather than to constantly create entirely novel cate-
gories. After all, each language only grammaticalizes a small part of
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
DISCUSSION POINTS
the cross-linguistically extant and cognitively available grammatical
categories. As we have argued, a possible reason for the tendency to
re-grammaticalize and extend extant categories rather than constantly
creating new ones is that extant grammatical (and lexical) categories,
unlike the non-existing ones, require the speaker’s attention when
verbalizing events (‘thinking-for-speaking’). They thus constantly gen-
erate focus on the expression of the category. Culture and society are
also likely to have the effect of narrowing down the potential sources
and targets of change. This still leaves many possibilities to express a
specific category, and the one that is actually chosen by speakers may
indeed often be due to historical coincidence, namely, as suggested by
Ariel (), the one that happens to be the first to become salient in
the discourse of a speech community. As the areal nature of grammat-
icalization also shows, there is a strong tendency for speakers to imitate
others and follow what they perceive as a norm. That is, once a pattern
has reached a certain level of saliency, through whatever circumstances,
it has a good chance to spread further.
In any case, despite the fact that the motivations and mechanisms of
grammaticalization are at the core of the phenomenon, the question of
what drives grammaticalization is not really resolved, and despite some
progress it is as controversial as ever. Due to the increasing availability
of corpora, especially spoken language data, and improvements in the
research methodology for spoken language, the trend is increasingly
towards seeking motivations in frequency and discourse, but this does
not necessarily mean that other factors are less valid. We have offered
here an outline of how various factors driving grammaticalization may
complement each other, but doubtlessly a lot of further research is
needed in this area.
Discussion points
() In your opinion, what assumption do you have to make about a “language
system” so that it functions as a trigger for language change?
() Make a list of communicative goals or needs of speakers that are plausible to
you. How may they lead to change in the way you speak?
() Can you think of other examples of dialogue that may end up as one turn
by a single speaker (Section ..)? What may or may not make them good
candidates for grammaticalization?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
7
Grammaticalization
in language contact
7.1 Grammaticalization vs. polysemy copying
7.2 A case study: Basque
7.3 Identifying contact-induced grammaticalization
7.4 Pidgins and creoles
7.5 Conclusions
Discussion points
As we saw in Chapter , ‘borrowing’ has traditionally been taken to be a
‘mechanism’ of change on a par with analogy and reanalysis. However,
the effects of language contact on language change in general and
grammaticalization in particular are far broader than just that, and
deserve to be discussed in a chapter of their own.
It has been widely assumed that there is a fundamental distinction
between internal and external linguistic change, where the latter is
influenced by language contact while the former is not. These two
kinds of change tend to be conceived as somehow mutually exclusive.
But as early as the nineteenth century, research on language contact has
suggested that this assumption is in need of reconsideration, and it is by
now fairly well established that multiple causation is a common phe-
nomenon: Linguistic change may simultaneously be an internal and an
external process.
We saw in Chapters and that there are tools for identifying
instances of grammaticalization even in languages for which no written
documents exist for reconstruction. As we will see in the present
chapter, reconstructions are also possible in cases where grammatical-
ization was induced by contact between different languages and
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION VS. POLYSEMY COPYING
dialects. However, the latter is faced with the following problem: How is
it possible to separate phenomena that are due to language-internal
development from those that are due to language contact? And perhaps
more crucially, is it possible in the first place to establish that language
contact was actually involved in a given case of linguistic change? There is
so far no conclusive answer to these questions, and contact-induced
grammaticalization is a notoriously controversial field, even in languages
which dispose of appropriate historical records (see Matras : –).
Nevertheless, there are some means of reconstruction that are of help in
answering the questions, as we will see in this chapter.
In the paragraphs to follow we will first look at one particular
example of processes of contact-induced grammaticalization and pro-
pose a four-stage scenario taking care of the gradual nature of such
processes (Section .). Section . then deals with one particular lan-
guage to show what effects such processes can have for speakers to
organize their discourses. Subsequently, a small set of criteria, or diag-
nostics, is proposed that may be of help in identifying instances of
processes of grammaticalization arising in situations of language contact.
Before presenting a number of cases of contact-induced grammati-
calization, a note on the terminology to be used seems in order.
Following Weinreich ([] : –; see also Heine and Kuteva
), the term (grammatical) replication is used when speakers create
new grammatical meanings or structures in one language, called the
replica language, on the model of another language, the model lan-
guage. Phenomena relating to replication have also been referred to
with terms such as ‘structural borrowing’ or ‘calquing.’ Replication thus
contrasts with borrowing, whereby phonological substance or form-
meaning units are transferred from one language to another.¹
7.1 Grammaticalization vs. polysemy copying
In a circum-Alpine area between between northern Italy and southern
Germany there are a number of languages which share the same kind
of passive construction, called the Alpine ‘come’-passive, which is
¹ There are many alternative terminologies; for example, Thomason and Kaufman
() and Thomason (: ) use borrowing, source language, and receiving language
for both kinds of transfer.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
illustrated in () (see also Wiemer : ). Two of the languages,
Italian and Ladin, are Romance languages while the other two, Pomat-
tertitsch and Bavarian, are dialects of German. What the four passive
constructions have in common is that they use essentially the same
construction, consisting of the verb ‘come’ as an auxiliary with the main
verb encoded as its complement in the past participle.
() The Alpine ‘come’-passive
a. Italian (Romance; Ramat : –)²
Qui viene costruita la scuola nuova.
here comes built the school new
‘Here the new school is being constructed.’
b. Ladin (Rhaeto-Romance; Ramat : –)
Cô vain fabricheda la scuola nuova.
here comes built the school new
‘Here the new school is being constructed.’
c. Pomattertitsch (Walser German; Giacalone Ramat and Sansò
: )
Der salam chun röwä gässä.
salami come:: raw eat:-
‘Salami is eaten raw.’
d. Bavarian (South German; Ramat : –)
Då kummt de nei(e) Schul gebaut.
here comes the new school built
‘Here the new school is being constructed.’
This situation raises the following questions:
(a) Why do verbs for ‘come’ express the function of passive
markers?
(b) Why are there four languages or varieties spoken in the same
general area employing the same construction?
A general answer to question (a) can be provided on the basis of the
discussions in the preceding chapters. Verbs commonly grammaticalize
into markers for grammatical functions such as tense, aspect, and
modality by being extended to new contexts where they take specific
² The glosses are ours, since Ramat () does not provide glosses.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION VS. POLYSEMY COPYING
verbal complements (context extension; see Section .). This is also a
way in which passive markers may arise, the English get-passive being a
case in point (Haspelmath ; Givón and Yang ). In specific
contexts, a verb may gradually turn into an auxiliary and its lexical
meaning will disappear (desemanticization; see Section .) in favor of
a grammatical function. As an auxiliary, it tends to lose salient mor-
phosyntactic features that it had as a lexical verb (decategorialization;
see Section .).
Passive markers can be historically derived from a number of differ-
ent verb types. The World Lexicon of Grammaticalization lists eight of
them (Kuteva et al. ). One of these eight verb types consists of
verbs meaning ‘come.’ But ‘come’-verbs are cross-linguistically not
particularly widespread as a source for passive markers. It may there-
fore be surprising that within a relatively small area of south-central
Europe there are four languages and dialects having ‘come’-passives.
At first sight, there is also a plausible answer to question (b). Among
the few languages that have been identified to have a grammaticalized
‘come’-passive, most are Romance, and they include Italian and
Rhaeto-Romance languages. As the detailed study by Giacalone
Ramat and Sansò (; see also Cennamo ) suggests, the rise of
the ‘come’-passive in Italian can eventually be traced back to fourth-
century Latin. The use of the Latin verb venire ‘come’ was extended to
specific types of verbal complements constructed in the past participle,
and it developed across northern Italian vernaculars. The rise of ‘come’-
passives in Rhaeto-Romance languages may also be a result of this
development.
However, the presence of a ‘come’-passive in German dialects such
as Pomattertitsch and Bavarian, as well as some other varieties of
German spoken in the same area, is in need of explanation. To our
knowledge, there is no history of such a passive in German nor is it
found in German dialects outside the circum-Alpine area. The most
reasonable hypothesis therefore is that the emergence of the ‘come’-
passive was the result of contact with Italian and/or Rhaeto-Romance
languages. Further evidence in support of this hypothesis is the follow-
ing: First, the two German dialects share essentially the same construc-
tion with their Romance neighbors. Second, this construction is also
found in other Alpine German-speaking communities in the western
region of the Canton of Fribourg, the whole German-speaking part of
Valais, the Walser settlements in Italy and Ticino, and parts of the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
Grisons (Giacalone Ramat and Sansò : ). Third, given the rare
occurrence of ‘come’-passives in the languages of the world it does not
seem likely that speakers of these German dialects would have devel-
oped the passives independently of their Romance-speaking neighbors.
And finally, the people of this circum-Alpine area between southern
Germany and northern Italy are known to share a history of close
contact over centuries (Ramat : –).
We can thus safely conclude, following the authors cited above, that
the grammaticalization of ‘come’-passives in these German dialects is
the result of language contact. This observation is of interest for the
following reasons: It suggests that grammaticalization is not inevitably
a language-internal process but can be induced by contact with other
languages or dialects.
In fact, the Alpine ‘come’-passive is not an isolated case. Such cases
of contact-induced grammaticalization have been documented in many
languages across the globe (e.g. Aikhenvald ; Johanson , ;
Heine and Kuteva , , ), and the present chapter is devoted
to this issue.
But our analysis of the Alpine ‘come’-passive raises the following
question: Did speakers of German dialects such as Pomattertitsch or
Bavarian really go through a process of grammaticalization to develop a
‘come’-passive or did they simply copy the use of the verb for ‘come’ as
a passive auxiliary, or of the construction at large, from Italian and/or
Rhaeto-Romance languages? In the latter case one would be dealing
with an example of what is described in the relevant literature as
calquing, structural borrowing, loan translation, or, more recently, as
polysemy copying (Heine and Kuteva : –).
Polysemy copying, whereby a combination of meanings of the model
language is copied analogically, that is, ‘translated’ into the replica
language, is in fact a ubiquitous phenomenon. It must have happened,
for example, when French and German speakers replicated the struc-
ture of the English word skyscraper by designing a close to literal
translation of the word, that is, French gratte-ciel (‘scratch-sky’) and
German Wolkenkratzer (‘cloud-scratcher’). In a similar fashion, one is
led to argue that the same must have happened when speakers of
Pomattertitsch and Bavarian replicated the Romance ‘come’-passive
by translating the construction word by word into their own dialects.
There are reasons to argue, however, that there is a fundamental
difference between the replication of lexical words or constructions and
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION VS. POLYSEMY COPYING
that of grammatical forms or constructions, in that the latter involves,
as a rule, grammaticalization rather than polysemy copying. These
reasons have been discussed in detail in Heine (); suffice it to
summarize the two main reasons here. First, unlike polysemy copying,
contact-induced grammaticalization is directional. For example, in
situations of language contact, lexical structures frequently develop
into grammatical structures, as has happened apparently in the case
of the Alpine ‘come’-passive. A development from passive auxiliary to a
lexical verb for ‘come,’ by contrast, has so far not been found—neither
in language contact nor elsewhere.
The second reason is the following: Unlike polysemy copying,
contact-induced grammaticalization is a gradual process which goes
through a number of stages before being completed. The Alpine
‘come’-passives surveyed above provide no evidence to substantiate
this point because there is little information on their diachronic devel-
opment. But there is a range of other cases which have been shown to
lend support to this hypothesis. These cases are discussed in detail in
Heine (). They primarily deal with the contact-induced rise of
tense-aspect markers and of definite and indefinite articles. In accord-
ance with the parameters proposed in Chapter , the grammaticaliza-
tion process involved can be sketched as in ().
() A scenario of contact-induced grammaticalization
I At the initial stage, speakers extend the use an existing expres-
sion to render the meaning of a grammatical category that
they are confronted with in their model language. This
expression, in most cases but not necessarily a lexical expres-
sion, is now ambiguous between its earlier meaning and the
one it acquires in its new context (cf. the bridging stage of the
context extension model in Section .).
II The expression is further extended to new contexts where the
old meaning is backgrounded, making little or no sense (cf.
the switch stage and the parameter of desemanticization).
III The expression spreads to most of the contexts where the
corresponding category of the model language is used, form-
ing essentially a translational equivalent of the latter. It is now
conventionalized, that is, it may no longer be conceived as
being the same as the earlier expression (the conventionaliza-
tion stage).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
IV In accordance with its new status as a functional category, the
expression loses morphosyntactic features that it once had
(decategorialization), and it may also lose part of its phonetic
substance (erosion).
To be sure, the possibility that a grammatical expression was transmit-
ted from one language to another via ‘loan translation’ cannot be ruled
out in some cases, but all cases that have been documented in some
detail are in support of the scenario in (), which we adopt as the
default hypothesis in the remainder of the chapter.
7.2 A case study: Basque
Contact-induced grammaticalization can have remarkable conse-
quences for the language concerned, as will be illustrated in this section
with examples from Basque.
The Basque language of northeastern Spain and southwestern France
is not only a genetic isolate but has also been claimed to be a language
that has remained largely unaffected by contact with other languages
(cf. Trask : ). Research carried out in the course of the last
decades shows, however, that this claim is in need of revision: The
language has been influenced by neighboring Romance languages to
the extent that its typological profile of discourse organization is no
longer what it used to be a few centuries ago, and many of the changes
involve contact-induced grammaticalization in some way or other
(Hurch ; Haase , ; Jendraschek ).
The main reason for looking at this language in more detail is that
the linguistic history of Basque was affected by two different contact
situations, with Spanish being the model language in Spain, while in
France the model languages were Gascon and French. Each of the two
situations had different effects on the language, and this fact is of help
in reconstructing grammatical change in Basque.
As the research cited suggests, one major driving force for speakers
to change their inherited use patterns is that they aim to establish
formulas of translational equivalence in relating the model languages
Spanish, Gascon, and French to their own language Basque. These
formulas tend to include features of the model languages, and once
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. A CASE STUDY: BASQUE
such formulas are used regularly and transmitted to following
generations of speakers they may lead to change.
Contact-induced grammaticalization in Basque affected all main
domains of grammar. In the remainder of this section we are restricted
to a few cases. For more examples see especially Hurch (), Haase
(, ), and Jendraschek ().
Change in the verbal system is manifested perhaps most of all in a
series of contact-induced grammaticalizations, some of which exhibit
a differential effect on the structure of the replica language Basque
depending on whether the model language was Spanish or Gascon and
French.
One of the features where such a difference may be found can be seen
in copular verbs. Spanish has two of them, namely estar ‘to be (localized
in space or time)’ and ser ‘to be (inherently)’, while French has only
one: être ‘to be’. In northern Basque of France, the Basque verb izan is
used for both meanings, exactly like French être. In modern southern
Basque of Spain, by contrast, the use of the verb izan corresponds to
that of Spanish ser. But for contexts where in Spanish estar would be
used, speakers appear to have grammaticalized the verb egon, histori-
cally meaning ‘wait’ or ‘remain,’ to a copular verb (Jendraschek ).
A similar distinction is found with the following verbs: Spanish
distinguishes between tener ‘to have (possess)’ (e.g. He has a red car)
and haber, the transitive perfect auxiliary (e.g. He has left), while French
uses avoir ‘to have’ for both. Jendraschek (: –) found that in
the Basque variety that is in contact with French, ukan is used for both
functions.³ In the Basque variety in contact with Spanish, by contrast,
the verb corresponding to Spanish haber is the copula izan (see above).
To form a possessive verb corresponding to Spanish tener ‘to have
(possess),’ Basque speakers appear to have grammaticalized their verb
eduki, originally meaning ‘hold (on to), contain.’ Note that, very much
earlier, a similar process had happened in Spanish, in that the possessive
verb tener can be traced back to the Latin verb tenere ‘to hold, keep.’
Table . summarizes the correspondences.
A third example suggesting that Basque speakers used grammatical-
ization to establish a structural equivalence relation between their own
language and the respective model languages is the following. Spanish
³ Ukan is one of the infinitive forms of izan, which we have encountered already in the
preceding example.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
Table 7.1 Verbal correspondences between two Romance model languages and varieties
of Basque (based on Jendraschek : )
Variety of Romance Spanish French
Romance tener haber avoir
Varieties of Basque eduki izan ukan
English translation ‘to have’ transitive auxiliary ‘to have’ + transitive auxiliary
has grammaticalized a construction containing the verb llevar ‘to carry’
to a kind of perfect progressive aspect (‘to have been doing’). The same
grammaticalization from ‘carry’ (e.g. Basque daramatza-) to a perfect
progressive is found in (southern) Basque. In contrast, French has not
undergone such a grammaticalization process, nor have the northern
varieties of Basque, which are in contact with French (Jendraschek
: ).
Progressive constructions are another area where language contact
appears to have had a remarkable impact on Basque, and here again
there appears to be a difference relating to the role of the model
languages involved. In northeastern varieties of Basque, both French
and the fellow Romance language Gascon provided a model. Gascon
has grammaticalized a progressive aspect based on a locative construc-
tion of the form ‘X is at doing Y,’ illustrated in (). Basque speakers
have done the same, apparently modeled on the pattern of Gascon.
They used a nominalizer (-tze) instead of the Gascon infinitive marker
and added the inessive () suffix -n on the main verb (subsequently
grammaticalizing the progressive further into an imperfective; Haase
: ), cf. ().
() Gascon (Romance; Haase : )
èste a her quaucòm
be at do: something
‘be in the process of doing something’
() Basque (Haase : )
kanta-tze-n dut.
sing-- :
‘I sing.’
A different progressive construction arose in Basque varieties of Spain.
Spanish has a progressive built on the grammaticalization of the verb
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. A CASE STUDY: BASQUE
for ‘go,’ as in (), and Jendraschek (: ) reports the same
grammaticalization for the Basque verb joan ‘go’ (here: doa), cf. ().
() Spanish (Jendraschek : )
El número de vascohablante-s va aumenta-ndo.
the number of Basque.speaker- go: increase-
‘The number of Basque speakers is gradually increasing.’
() Basque (Jendraschek : )
euskaldun-en kopuru-a gehi-tu-z doa.
Basque.speaker-: number- increase-- go:
‘The number of Basque speakers is gradually increasing.’
The following example relating to language contact in France is about a
periphrastic construction to express a causative function. It consists of a
verb for ‘do’ as an auxiliary followed by the infinitival main verb,
whereby the formula is [har ‘do’ + infinitive] in Gascon and [faire ‘do’ +
infinitive] in French. Basque speakers seem to have replicated this
construction to some extent by grammaticalizing their verb egin ‘do’ to
a causative marker with the main verb used in a participial form, cf. ().
() Basque (Haase : –)
[ . . . ] ber-e-k lei[h]o bat-eta-(r)at jautz-i [eg]in [ . . . ]
-- window one-- descend- do:
‘[ . . . ] they must take them (down) to a window [ . . . ]’
In a similar way to the verbal system, the noun phrase of Basque shows
a series of contact-induced grammaticalizations, leading in the direc-
tion of what Haase (: ) calls a ‘Romance system.’ A paradigm
case of grammaticalization is that from a numeral for ‘one’ to indefinite
article—one that is common in many languages across the globe (cf.
Section .; Kuteva et al. , > ), the English indef-
inite article a(n) being a case in point. This process usually involves the
following stages of evolution (see Section .):
() Stages of evolution of indefinite articles (Heine and Kuteva :
–)
I Numeral ‘one’
II Presentative marker (e.g. Once upon a time there was an old man.)
III Specific indefinite marker (This morning I saw a black cat.)
IV Non-specific indefinite marker (Draw a rabbit!)
V Generalized article
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
Basque speakers in France have developed such an article as a result of
contact with Gascon, and later on with French, resulting in both cases
in a Stage IV indefinite article (Haase : –). They did so via
the grammaticalization of the Basque numeral bat ‘one.’ There is
no information on when Stage II arose, but occasional uses of bat as
a Stage III article are attested in . The grammaticalization as a
Stage IV article is clearly a recent innovation, even if there were already
incipient uses as early as . Being less grammaticalized than its
equivalents in the model languages Gascon and French, the article bat
has not reached the final Stage V: It is is not used in some contexts
where in French or Spanish an indefinite article would be required. To
conclude, contact-induced grammaticalization appears to have pro-
ceeded roughly along the lines expected for language-internal pro-
cesses, even if it has not attained the same degree of development as
the model languages.
Another example involving noun phrase morphology is the rise of
comitative-instrumental case polysemy in Basque based on the model
of the three Romance languages. This example, relating to what has
been called ‘secondary grammaticalization’ (see Section .), involves a
well-known pathway leading from markers expressing a comitative
function (‘(together) with’) to also encode instrumental participants
(‘with, by means of’) (see Kuteva et al. , >
), the result being comitative-instrumental polysemy.
Basque is a language rich in case inflections, distinguishing at least
fourteen case suffixes, including both a comitative and an instrumental
suffix, thus distinguishing the two functions by means of their case
morphology. Like most other European languages (Heine and Kuteva
: –), the languages with which Basque is or has been in
contact, that is, Spanish, Gascon, and French, have a comitative-
instrumental polysemy, that is, Spanish con, Gascon dab, dambe, and
French avec ‘with’ are used for both comitative and instrumental
participants. Presumably based on the model of these languages, Bas-
que is in a process of giving up this distinction: Wherever in the model
languages a comitative-instrumental preposition is used, Basque speak-
ers tend to use the comitative case suffix -ekin (or -ekilan) also for
instrumentals. The instrumental case suffix -(e)z can still be found to
some extent to denote instruments or means, as in (a). However, in
most contexts it tends to be replaced by the comitative case, cf. (b),
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. A CASE STUDY: BASQUE
thereby replicating the polysemy of the comitative-instrumental
preposition con of the Spanish model language (Hurch ), cf. ().
() Basque (Hurch : )
a. kotxe-z
car-
‘with a car’
b. kotxe-a-r-ekin
car-the-r-
‘with the car’
() Spanish (Hurch : )
con el coche
with the car
‘with the car’
Thus, it appears that in an attempt to establish equivalence with the
model languages, the comitative-instrumental polysemy of the model
languages is replicated in Basque by grammaticalizing the comitative to
also refer to instrumental participants, with the earlier instrumental
case gradually being given up.
Another case of contact-induced grammaticalization in Basque in-
volves a common pathway whereby noun phrases having nouns as their
nucleus grammaticalize into adpositions, that is, prepositions or postposi-
tions (cf. English in front of, on account of, etc.). In Basque, case relations
are expressed primarily by an elaborate system of case suffixes. In the
three model languages Spanish, Gascon, and French, by contrast, case
relations are expressed overwhelmingly by prepositions. Apparently in an
attempt to replicate the structure of the model languages, Basque speakers
are gradually developing a set of postpositions via the grammaticalization
of relational nouns. Thus, by grammaticalizing relational nouns such as
buru ‘head’ or baita ‘interior,’ Basque is acquiring a set of complex
postpositions modeled after corresponding complex prepositions in the
Romance languages. For example, as demonstrated by Haase (: ),
the Basque postposition -ri/-ra(t) buru-z (/ head-) ‘in the
direction of’ is modeled after the Gascon preposition (de) cap a ((from)
head at) ‘in the direction of’ (Haase : –). For a whole set of
equivalences in such complex adpositions, see Haase (: ).
Grammaticalization induced by contact with Romance languages
has also affected the system of morphologically independent personal
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
pronouns: It has filled a gap in the deictic paradigm by producing a new
personal pronoun in Basque. While Gascon and French have third
person pronouns (Gascon eth ‘he,’ era ‘she’; French lui ‘he,’ elle ‘she’),
Basque has not. In order to develop an equivalent third person pro-
noun, thereby bringing their paradigm of personal pronouns in line
with that of the model languages, Basque speakers are grammaticalizing
their identity pronoun ber- ‘same, -self’ into a third person pronoun
(Haase : –).⁴
That this is a process of grammaticalization is suggested by the fact
that, first, this is a unidirectional development: Cross-linguistically
there are a number of languages where intensifiers (‘-self’) or identity
forms (‘the same’) have given rise to personal pronouns, while there is
no evidence for a development in the opposite direction. Second, this
development can be described in terms of desemanticization (see Sec-
tions . and .) whereby the specific intensifier or identity semantics
is bleached out—with the effect that third person reference is the only
semantic function that is left.
In addition to the noun phrase and the verb phrase, contact-induced
grammaticalization has also played a role in other domains of Basque
grammar. Our final example is clause subordination, which has also
been shaped to some extent by language contact (see Haase ;
Jendraschek ).
One of the linguistic features that has figured prominently in dis-
cussions on Europe as a linguistic area is the fact that forms used to
introduce relative clauses may be the same as question words
(Haspelmath b, ). For example, the English interrogative
where in (a) has the same form as the relative clause and complement
clause marker in (b).
() a. Where do you work?
b. I don’t know where you work.
The same phenomenon is found in most European languages and, as
has been shown in Heine and Kuteva (: –), it is largely due to
a general process of grammaticalization leading from question words
such as ‘who?,’ ‘what?,’ ‘where?,’ etc. to complement clause and relative
clause markers. These European languages also include the three
⁴ The ‘identity pronoun’ seems to largely corresponds to what König and Siemund
() label as ‘intensifier’.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. A CASE STUDY: BASQUE
Romance languages in contact with Basque, namely Spanish, Gascon, and
French. And more recently, Basque speakers seem to have replicated this
process (Hurch ; Haase ; Trask : ).
In Basque, a finite relative clause precedes its head. There is no
relative pronoun, but the verb takes a suffix marking it as subordinate.
However, for some generations now, a new relativization strategy has
been used in certain parts of the Basque-speaking area under the
influence of Spanish. What happened is that the question marker zein
‘which?’ (also‘who?’ in places) was used as an equivalent of the Spanish
interrogative pronouns and zein was grammaticalized to a relative
clause marker (Trask : ). Relative clauses introduced by zein
are documented in texts since the seventeenth century but then as a rule
only in translations (Hurch : ). The new relative clause marker
does not replace the earlier structure. Instead, zein is added to the
existing relative construction (see also Trask : ), with the result
that there is now double marking on post-nominal relative clauses as in
()—a case of what was described in Section .. as reinforcement.
() Basque (Jendraschek : )
neska zein-i lore-a-k eman
girl which- flower-- give
dizki-o-da-n [ . . . ].
::-:-:-
‘The girl to whom I gave the flowers [ . . . ].’
Basque as spoken in southwestern France apparently has gone one step
further in having developed a fairly productive type of relative clause on
the model of Romance languages. Thus, the construction in (), used
in the written language, corresponds in its major properties to that of
many other European languages: The relative clause is postnominal,
has inflecting relative pronouns introducing the relative clause,
and the pronoun is resumptive, signaling the head’s role within the
relative clause.
() Basque (Haase : )
Hiri bat ba-zen, zoin-tan ez
town one -: which-:
bait-zen eliza-rik.
-: church-
‘There was a town where there was no church.’
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
But it was not only relative clauses that were affected by the rise of the
interrogative-to-subordination grammaticalization and the ‘Romance
subordination type,’ as it tends to be referred to; Basque speakers in
France have used a range of question words, such as non ‘who?,’ zer
‘what?,’ nun ‘where?,’ nola ‘how?,’ and zoin ‘which?,’ to replicate com-
plement and adverbial clauses of Gascon and French (Haase : ).
The innovations discussed in this section constitute only a part of all
the contact-induced processes that Basque has undergone in the course
of the last centuries. For more examples, see especially Haase ()
and Jendraschek (). Jendraschek (: ) concludes that Bas-
que made few concessions on its typological characteristics but gained
much in intertranslatability in the contact with Romance languages. In
fact, there is clearly a movement away from discourse options and
linguistic structures traditionally available to Basque speakers towards
a Romance-type grammatical discourse organization, even if on the
surface the two kinds of languages still show remarkable typological
contrasts.
The examples provided above may have shown that the grammatical
development of the Basque language over the last centuries was shaped
to quite some extent by contact-induced grammaticalization. As the
available research findings suggest, speakers aimed at establishing for-
mulas of equivalence between Basque and its three Romance model
languages, changing inherited patterns of language use and norms in
the direction of the model languages.
These changes are most pronounced in the pragmatic domain of
context extension (Section .), that is, in the manipulation of context
in search for new discourse options that most appropriately match
grammatical functions expressed in the model languages. With the
rise of new functions emerging in new contexts, corresponding to
functions found in the model languages, old functions are back-
grounded and may eventually fall into disuse altogether (desemantici-
zation; Section .).
Most of the changes discussed here seem to be relatively new. But
they did not all arise in Basque at the same time. Rather, they spread
over quite a number of centuries. One major wave of replications had
already taken place prior to . They involve the development from
interrogative to relative pronouns as well as the rise of the indefinite
article, of complex postpositions, and of the periphrastic tense-aspect
constructions mentioned above. But most of the changes discussed are
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. IDENTIFYING CONTACT-INDUCED GRAMMATICALIZATION
no older than two hundred years (Haase : ). This means that
the changes have the appearance of weakly grammaticalized processes,
some of which have not proceeded much beyond the bridging stage of
context extension (see Section .). Little decategorialization has taken
place so far, and we have found no traces of erosion in the studies on
which the discussion in this section rests (Trask : ; also Hurch
; Haase , ; Jendraschek ).
As argued in those studies, and crucially for the purpose of this
chapter, the changes reported were contact-induced. Furthermore,
the changes were essentially no different from grammaticalization
processes in languages where no discernable effects of language contact
can be found. This observation is in accordance with findings made in
a number of other studies (e.g. Heine and Kuteva , , ),
namely, first, that grammaticalization can be triggered or accelerated by
language contact and, second, that contact-induced grammaticalization is
not fundamentally different from language-internal grammaticalization.
7.3 Identifying contact-induced grammaticalization
The observations made in the preceding section raise the question of
how to establish that replication has in fact taken place—we drew
attention to this problem in the introduction to this chapter. This
question is the main subject of the present section. More specifically,
we will be looking for answers to the questions listed in ().
() Questions
a. What evidence is there for contact-induced grammaticaliza-
tion to have taken place?
b. Which was the model language and which the replica
language?
c. Is it possible to prove that contact-induced grammaticaliza-
tion has taken place?
In the present section we will search for criteria to identify instances of
contact-induced grammaticalization by looking at a range of documen-
ted case studies. More specifically, we will look for diagnostics that are
suitable for reconstructing grammatical change in situations of lan-
guage contact. Note that the diagnostics are not all independent of one
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
another; rather, some of them are similar to others, with each high-
lighting a distinct analytic perspective.
Question (a) is the subject of Section .., while Section .. is
devoted to question (b). Question (c) must be answered in the
negative and will be ignored below, for the following reason: The
diagnostics are based on observations of established cases of language
contact, and are therefore only probabilistic. None of the diagnostics is
sufficient to ‘prove’ effects of language contact, but the more of them
apply, the stronger the hypothesis on contact-induced grammatical
change that can be formulated.
The data to be discussed are taken from attested cases of language
contact, and exemplification is kept at a minimum. For more details,
see Heine (, ) and Heine and Nomachi (). They involve a
set of two languages, namely a model language M and a replica
language R. Note that M and R are not necessarily full-fledged lan-
guages; they can also be dialects or varieties of a language.
7.3.1 Evidence for contact-induced grammaticalization
The present section proposes the following three diagnostics that can
be of help in proposing a hypothesis on whether a given process of
grammaticalization was induced by language contact:
(a) genetic patterning;
(b) rare grammatical category;
(c) paired grammaticalization.
These diagnostics are now looked at in turn.
D: Genetic patterning. If two neighboring languages, M and R, share
a grammaticalization process that is also found in languages closely
related to M but not in languages closely related to R, then this situation
justifies a hypothesis to the effect that there was contact-induced
grammaticalization leading from M to R.
This diagnostic is based on the observation that two languages in contact,
M and R, share the same kind of grammaticalization which is absent in
languages closely related to R but not participating in the contact situa-
tion. On the other hand, this very grammaticalization can be a general
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. IDENTIFYING CONTACT-INDUCED GRAMMATICALIZATION
feature of the language group to which M belongs genetically and, hence,
may also be found in other languages of that group.
For example, Eastern Yiddish, a language historically derived from
medieval German, has been spoken for centuries in a Slavic-speaking
environment. Russian has a focus construction illustrated in (). As
Prince () argues convincingly, Yiddish speakers have replicated
this construction (cf. ()) by grammaticalizing the linguistic resources
available in their own language. The similarities between the two
constructions, referred to respectively as the eto- and the dos-construc-
tions, are described by Prince (: ) as follows: “both are simple
sentences (i.e. without subordinate clauses), each has an expletive NP in
initial position, in both cases the expletive NP is the neuter demonstra-
tive pronoun, and both are roughly translatable by an English it-cleft”.
() Russian (Prince : )
Eto Leonid uvidel Eriku.
this: Leonard:: saw Erica::
‘It’s Leonard who saw Erica.’
() Yiddish (Prince : )
Dos hot Leyb gezen Erike-n.
this: has Leonard:: seen Erica-:
‘It’s Leonard who saw Erica.’
This construction has no obvious analog in Germanic languages other than
Yiddish, but there are analogs in other Slavic languages—hence, contact-
induced transfer from Slavic languages such as Russian, Ukrainian,
Belarusian, and/or Polish to Yiddish is the most plausible hypothesis.
The following example also relates to Germanic-Slavic language
contacts. The German reflexive pronoun sich () is restricted to
third person subject referents, cf. (a), while in the first and second
person, object pronouns agreeing with the correponding subject pro-
noun are used, as in (b).
() High German
a. Er wäscht sich.
he washes
‘He washes himself.’
b. Wir waschen uns.
we wash us
‘We wash ourselves.’
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
In the Slavic language Slovenian, by contrast, the same reflexive marker
is used for the whole paradigm of person markers. Under the influence
of Slovenian on German in Trieste, the Slovenian pattern of reflexive
marking was replicated to some extent by German speakers (Morfill
[] : ): The German reflexive pronoun sich was extended
from third person to second and first persons in the singular and plural,
that is, to the whole paradigm of person deixis, on the model of the
Slavic reflexive marker se, cf. ().
() German as spoken in Trieste, Slovenia (Morfill [] : )
Wir waschen sich.
we wash
‘We wash ourselves.’
A similar example is reported by Thomason and Kaufman (: ),
who observe that the Yiddish reflex six () of the German third
person reflexive sich was extended to all persons and numbers under
Slavic influence. This usage is uncommon in German but common in
Slavic languages, hence it is suggestive of transfer from Slavic languages
to German and Yiddish. Note that grammaticalization was restricted in
these examples to the parameter of context extension (Section .),
whereby the use of a given form was extended to new contexts—that is,
grammaticalization did not affect the semantic, morphosyntactic, or
phonological substance of the constructions concerned.
The Finnic language Estonian has grammaticalized the verb tulema ‘to
come’ to a modal auxiliary for the deontic necessity (‘must,’ ‘have to’),
presenting the agent in the stative-locative adessive case, an oblique case
form. Much the same is found in the Baltic language Latvian, which also
has grammaticalized the verb for ‘come’ in its reflexive form to a modal
auxiliary (nākties), also expressing deontic necessity, with the agent being
encoded by means of an oblique case marker, the dative (Stolz b:
–). The striking similarity between these two languages, as well as
the fact that grammaticalization of verbs for ‘come’ to deontic modals
appears to be cross-linguistically rare, can be interpreted meaningfully
only by assuming that the constructions are historically related. Estonian,
a Finnic language, and Latvian, a Baltic and hence, an Indo-European
language, are genetically unrelated. Therefore, genetic relationship can be
ruled out, and the only reasonable hypothesis is one in terms of contact.
The following fact may serve as evidence that it was a Finnic language
that provided the model: Finnish, a language closely related to Estonian,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. IDENTIFYING CONTACT-INDUCED GRAMMATICALIZATION
also uses the verb for ‘come’ as a modal auxiliary for deontic modality
and an oblique case, the genitive, for presenting the agent, while in Baltic,
such a grammaticalization is absent except for Latvian, that is, it is not
found in Lithuanian another Baltic language. In view of this genetic
patterning we follow Stolz in arguing that this as well is a case of
contact-induced grammaticalization from a Finnic to a Baltic language
rather than the other way round.⁵
D: Rare grammatical category. If two neighboring and genetically
unrelated or only remotely related languages share the same kind of
grammaticalization process which is cross-linguistically unusual, then
there is some probability that this commonality can be accounted for
with reference to language contact.
In varieties of the Indo-Aryan language Romani spoken on the Balkans,
Boretzky and Igla (: ) recorded a construction of the kind [‘It
does itself to X’], where the experiencer is encoded as a dative partic-
ipant [‘X’] and the verb phrase contains a third person subject referent
and a reflexive pronoun. This typologically somewhat unusual con-
struction has been grammaticalized to a volitive/desiderative construc-
tion meaning [‘X wants/desires to do’], cf. (). The same construction
is found in other languages of the Balkans such as Serbian, Bulgarian,
and Albanian, which are known to share a history of language contact
with Romani. () shows the Bulgarian counterpart.
() Southern Balkanic Romani (Boretzky and Igla : )
Na beš-el pes man.ge.
sit- to:me
‘I don’t want to sit.’ (Lit.: ‘It doesn’t sit itself to me’)
() Bulgarian (Boretzky and Igla : )
Ne mi se jade.
to:me eat:
‘I don’t want to eat.’ (Lit.: ‘It doesn’t eat itself to me’)
⁵ The most plausible hypothesis is that this transfer proceeded from a Finnic to a Baltic
language, that is, from Estonian to Latvian, rather than the other way round. But there is at
least one more possibility for a contact explanation of this particular case, not mentioned by
Stolz. According to this possibility, the model language was Russian, while Finnish,
Estonian, and Latvian are the replica languages. The arguments in support of such a
hypothesis would be: First, Russian has a pattern equivalent to the one that Latvian has
for expressing the deontic necessity, and second, Russian is geographically contiguous with
Finnish, Estonian, and Latvian, but not with Lithuanian (Östen Dahl, p.c.).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
A grammaticalization process of this kind is cross-linguistically not
particularly common and genetic relationship does not offer a convinc-
ing explanation for this similarity: All languages concerned are Indo-
European. We therefore follow Boretzky and Igla (: ) in hy-
pothesizing that the Romani construction is due to contact-induced
grammaticalization based on the model of other Balkan languages.
Another example of a rare category is found in the contact between
Italian and Maltese. Markers and constructions for deontic modality
can have a number of historical sources (see Kuteva et al. ), but the
following appears to be cross-linguistically rare, if not unique: In Italian
dialects including Sicilian, though not in Standard Italian, a construc-
tion [tocca a X] ‘it touches X’ has been grammaticalized to a functional
category of deontic modality, more precisely for ‘must,’ cf. (). This
construction appears to have been replicated in Maltese, which shares a
history of over years of contact with Sicily. Example () illustrates
the corresponding construction of Maltese. Considering that two
neighboring, genetically unrelated languages with a long history of
contact show a grammaticalization process that is rarely, if at all,
encountered elsewhere in the world, an explanation in terms of lan-
guage contact would seem to be plausible (Haase ).
() Spoken Italian (Haase )
Tocca a andare a vederlo.
touches to go: to see.him
‘You should go to see him a bit.’
() Maltese (Haase )
Imissek tmurlu ftit.
touch::: go::::to: a.bit
‘You should go to see him a bit.’
The impression that may have been conveyed is that rare phenomena
are particularly sensitive to contact-induced language change. This was
in no way the intention for proposing diagnostic D. To our knowl-
edge, there is no solid evidence in support of such a claim. What was
intended, rather, is that the presence of ‘rare grammatical categories’
has a higher diagnostic value than that of other categories.
D: Paired grammaticalization: If two neighboring but genetically
unrelated or only remotely related languages share not only one but
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. IDENTIFYING CONTACT-INDUCED GRAMMATICALIZATION
two grammaticalization processes for the same general grammatical
function, then this fact provides evidence for contact-induced transfer.
According to Boretzky and Igla (: ), speakers of Bulgarian
varieties of Romani have grammaticalized a future tense construction,
following the model of Bulgarian, by using the Romani verb kam ‘want’
plus the finite main verb, cf. (). The exact parallel in the Bulgarian
construction is illustrated in ().
() Romani varieties influenced by Bulgarian (Boretzky and Igla
: )
ka(m) ker-av.
want do-:
‘I will do (it).’
() Bulgarian (Boretzky and Igla : )
šte otida.
want go::
‘I will go.’
This parallelism on its own does not provide strong evidence in favor
of a contact hypothesis since de-volitive futures, that is, future tense
markers grammaticalized from verbs for ‘want’ (Dahl ), are cross-
linguistically widespread, especially on the Balkans. But Boretzky and
Igla demonstrate that there is additional evidence to establish that
language contact must have been involved in this grammaticalization.
For the negative future, Bulgarian speakers use a different construction,
namely a possessive construction based on an auxiliary verb meaning
‘have,’ cf. (), and this again is exactly what is found in Romani
varieties of Bulgaria, as illustrated in ().
() Bulgarian (Boretzky and Igla : )
njama da otida.
not:have to go::
‘I will not go.’
() Romani varieties influenced by Bulgarian (Boretzky and Igla
: )
naj/nane te ker-av.
have:not that do-:
‘I will not do (it).’
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
Thus, there are two parallel grammaticalization processes leading to the
rise of future tense forms in two neighboring languages, involving
exactly the same conceptual schemas and leading to the same affirm-
ative/negative split situation of future tense marking. This situation is
presumably unique in the languages of the world. Since Romani and
Bulgarian are known to have had a history of intense contact, and since
neither genetic relationship nor chance can serve as evidence to
account for the similarity, it is strongly suggestive of contact-induced
change. This hypothesis is further supported by the following observa-
tion. The grammaticalization of possessive constructions to future
tenses is widespread in European languages but fairly uncommon
elsewhere (see, e.g., Kuteva et al. ). It therefore seems plausible
that speakers of Romani, a language of Indian origin, acquired it from
the European language Bulgarian.
7.3.2 Diagnostics for determining the direction of transfer
The diagnostics listed above can be of help to determine whether or not
contact-induced grammaticalization has taken place (see question (a)
above). Once this question is answered then the next question concerns
the direction of the contact-induced transfer. This question was already
addressed in some form or other in the preceding sections but in the
present section we propose the following more specific set of diagnostics
(see question (b)).
(a) degree of grammaticalization;
(b) frequency of use;
(c) demographic variables.
D: Degree of grammaticalization: If two neighboring languages have
undergone the same process of grammaticalization as a result of lan-
guage contact but one of the languages exhibits a high and the other a
low degree of grammaticalization, then the former is likely to be the
model language.
A category that has been transferred from one language to another is as
a rule—though not always—more grammaticalized in the model lan-
guage than the corresponding category in the replica language. This can
be illustrated with the example from northern Brazil given in Table ..
The North Arawak language Tariana of northwestern Amazonia has an
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. IDENTIFYING CONTACT-INDUCED GRAMMATICALIZATION
Table 7.2 Portuguese expressions grammaticalized by Tariana speakers on the model of
evidential categories in Tariana (Aikhenvald : –, –; Heine and Kuteva ,
Table .)
Tariana evidential category Corresponding Tariana Portuguese expressions
Visual eu vi ‘I saw’ eu tenho prova ‘I have proof ’ eu tenha experiencia
‘I have experience’
Nonvisual eu escutei ‘I heard’ eu senti ‘I felt’
Inferred parece ‘it appears, it seems’
Reported diz que ‘it is said that’
obligatory paradigm of four clitics for tense and evidentiality, distin-
guishing between visual, non-visual, inferred, and reported evidence.
Tariana speakers use Portuguese, the official language of Brazil, as an
important lingua franca, and in their use of Portuguese, they tend to
replicate their evidentiality system by drawing on lexical expressions of
Portuguese, using them more frequently and grammaticalizing them
into what appear to be incipient categories of evidentiality for which
there is no equivalent in Standard Portuguese (Aikhenvald :
–, –; cf. Aikhenvald ).
The fact that Tariana Portuguese is presumably the only variety of
Portuguese to have such a paradigm of four evidential structures can be
accounted for meaningfully only with reference to language contact.
Compared to the Tariana model categories, which are fully grammati-
calized, the replicated categories in Tariana Portuguese are only
weakly grammaticalized. In terms of the context extension model of
Section ., expressions like eu vi ‘I saw’ do not appear to have
proceeded much beyond the bridging stage II of extension, that is,
they can still be interpreted with reference to their literal lexical mean-
ing. Thus, grammatical replication must have proceeded from Tariana
to Tariana Portuguese rather than the other way round.
Rayfield (: ) provides another example from Yiddish speakers
in Venice along the coast next to Los Angeles, most of them strongly
bilingual in English. They created a future tense on the model of the
English be going to future, using the Yiddish verb ge- ‘go’ as an auxiliary
and the main verb in the infinitive. As is to be expected, the replicated
future category is less strongly grammaticalized than its English model.
In particular, it is not generally acceptable when the andative deixis (‘to
move away from the speaker’) of Yiddish ge ‘go’ is violated, for
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
example, when serving as an auxiliary for kumen ‘to come’ (i.e. ‘to move
towards the speaker’), as in (). Thus, it does not seem to have
proceeded beyond the bridging stage II of the context extension
model in Section ..
() Yiddish of Venice, California (Rayfield : )
All right, ge ikh kumen bald.
all right go come: soon.
‘All right, I’m going to come in a minute.’
Thus, while in English it is possible to say I’m going to come in a minute,
a Yiddish speaker at the meeting of a Yiddish cultural group in Venice
was reproved for saying something like () because of the conflict
of spatial deixis expressed by ge- ‘go’ and kumen ‘come.’ Thus, with
regard to the parameter of desemanticization (Section .), the
Yiddish future construction is less desemanticized—and hence, less
grammaticalized—than the corresponding English be going to future
since it has not lost its association with the andative meaning of ‘go.’
Relative degree of grammaticalization usually, though not necessar-
ily, correlates with the relative age of the categories in question.
Contact-induced transfer implies that a model category is older than
the corresponding replica category in that it already existed prior to
language contact, while the replica category is, at least to some extent, a
product of language contact. Now, older categories may have properties
that bear witness to their age, such as showing a higher degree of
erosion (see Section .). Accordingly, in language contact between
Mayan and Mixe-Zoquean languages in the Mexican state of Chiapas,
Zavala () noted that since there is more phonological attrition, that
is erosion, in the relevant material of the Mayan languages, there is
reason to argue in favor of a transfer from Mayan to Mixe-Zoquean
languages rather than the other way round (Gast and van der Auwera
: ).
D: Frequency of use. The category of the model language tends to be
used more frequently than a corresponding category of the replica
language.
According to this diagnostic, differences in the relative frequency of use
may provide evidence to determine the direction of contact-induced
transfer, based on the following observation: A grammaticalization
process in the model language is generally older than a corresponding
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. IDENTIFYING CONTACT-INDUCED GRAMMATICALIZATION
process in the replica language. Therefore, the former is likely to
be more strongly grammaticalized and to be used more frequently
than the latter.
For example, English has fully grammaticalized, obligatory plural
marking on nouns. English is a secondary lingua franca for speakers of
the Austronesian language Tigak of Papua New Guinea, which does not
have obligatory plural marking on nouns. But under the influence of
English, plural marking on nouns appears to have risen in Tigak.
Jenkins (: ) found in her transcribed Tigak material that
young educated Tigak increase the use of their overt plural morpheme,
using it with percent of all Tigak plural nouns whereas traditional,
conservative Tigak speakers use the plural marker with only percent
of plural nouns.⁶ This change appears to be restricted to the grammat-
icalization parameter of context extension (Section .) but does not
have any further effect on the structure of Tigak.
In a similar fashion, Turks in Germany have been found to use their
conjunction ve ‘and’ and their plural marker much more frequently
than Turks in Turkey, and Johanson (: –) takes this to reflect
German influence, drawing attention to the fact that Turkic languages
that have been influenced by Indo-European languages generally show
an increased frequency of use of conjunctions.
Thus, frequency of use may also offer a clue for reconstructing
directionality in contact-induced grammaticalization. Obviously, this
diagnostic may not apply to a situation where the model language and
the replica language share a long history of intense contact. In such a
situation, the replica language can have grammaticalized the category
in question to the extent that there no longer is a difference in the
frequency of use.
D: Demographic variables. If in a given situation of language contact
some process of grammaticalization is restricted to or less pronounced
in one social group of speakers of one language while the other lan-
guage does not show such a restriction, then the former is likely to be
the replica language and the latter the model language.
Language contact in general and contact-induced grammaticalization
in particular affect certain social or demographic groups more than
others. Accordingly, social groups that are more strongly affected by
⁶ There are unfortunately no details on what these figures exactly refer to.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
contact are likely to show more pronounced effects of contact-induced
change. Linguistic effects of language contact can be restricted to, or be
especially intense in a social group having had some formal education,
people of some particular profession, such as traders, or people of the
upper class rather than of lower classes, etc. Or contact may be a recent
phenomenon that has affected only younger speakers, or younger
people earlier than older speakers. Accordingly, contact-induced gram-
maticalization can be restricted to or be more pronounced in certain
social groups that can be shown to have been exposed to contact more
than other groups,⁷ and appropriate sociolinguistic information may
assist in formulating hypotheses on directionality in contact.
Aikhenvald () in particular has demonstrated that sociolinguis-
tic factors, especially demographic variables such as age, sex, profes-
sion, social status, provide clues for identifying instances of
grammatical replication (see also Heine and Kuteva : –). One
of the examples volunteered by Aikhenvald (: ), is the follow-
ing: The North Arawak language Tariana of northwestern Brazil is in
close contact with Portuguese, the official language of Brazil, and has
been influenced by the latter in various ways. In Portuguese, but not in
Tariana, interrogative pronouns are also used as relative clause
markers—we had a similar example from Basque in Section .. Tari-
ana speakers have also grammaticalized their own interrogative pro-
nouns to markers of clause subordination on the model language
Portuguese, retaining their own relative construction and simply add-
ing their interrogative pronoun (e.g. kwana ‘who?’). As Aikhenvald
points out, it is young speakers of Tariana that have innovatively
created this new structure, and the structure seems to be confined to
this demographic group. Since Portuguese does not have such a soci-
olinguistic restriction on the use of its interrogative-derived relative
clause markers, the direction of replication must have gone from
Portuguese to Tariana.
As we saw above (under D), the situation on the island of New
Ireland in Papua New Guinea also illustrates this factor. The Austro-
nesian language Tigak does not have obligatory plural marking on
nouns, but under the influence of English, which has obligatory plural
marking and is a secondary lingua franca for Tigak speakers, educated
young Tigak increase the use of their overt plural morpheme (Jenkins
⁷ We are grateful to Brian Joseph (p.c.) for reminding us of this point.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. PIDGINS AND CREOLES
: ). Thus, once again, it is one particular segment of the
population that provides clues for identifying directionality in ongoing
processes of contact-induced change.
The catalog of diagnostics presented in the preceding sections is
meant to provide some guidelines on how to identify instances of
contact-induced grammatical replication. As noted above, these guide-
lines are not based on exceptionless generalizations, they merely rest on
probabilities derived from a larger number of studies on contact-
induced grammatical change (see especially Aikhenvald ; Heine
and Kuteva ). The greater the number of diagnostics that apply,
the stronger the hypothesis on language contact that can be proposed.
As Kuteva () has demonstrated, it is possible to use a combination
of these diagnostics in cases where a single criterion might not be very
revealing.
Further reading
Diagnostics of contact-induced grammaticalization have been proposed in
Heine (, ) and Heine and Nomachi (). The role played by
grammaticalization in areal linguistics, and in particular in areal diffusion
is discussed, e.g., in Stolz (b), Kuteva (b, ), and Heine and
Kuteva (: –; ).
7.4 Pidgins and creoles
Pidgins are languages that have a reduced lexicon and grammatical
structure, resulting typically from extended contact between groups of
people with no language in common. Creoles typically evolve from
pidgins by being learned by children as first languages, thereby acquir-
ing an extended lexicon and grammar. As a result, creoles are in a
number of ways indistinguishable from other languages. This general
characterization is not meant to do justice to the wide social and
linguistic space covered by the terms ‘pidgin’ and ‘creole,’ which are
commonly subclassified into jargons (‘unstable’ or ‘rudimentary pid-
gins’), stable pidgins, expanded pidgins, pidgin/creoles, etc., nor to the
rich literature that exists on these languages (see Further reading).
Pidgins and creoles arise in complex socio-cultural situations. They
are commonly believed to be the pronounced or ‘extreme’ result of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
language contact, and they have been the subject of a range of
discussions on issues such as first and second language acquisition,
language simplification and language mixture, and language genesis.
The linguistic factors leading to the rise of these languages are also
complex, but grammaticalization appears to have been a major force in
the evolution of both pidgins and creoles; see Bruyn (a, b,
, ), Heine (, ), Romaine (), Heine and Kuteva
(: –, : –), as well as the contributions in Baker and
Syea () for details and generalizations.
The present section is restricted to demonstrating that grammatical-
ization processes can also be commonly observed in pidgins and
creoles. Overall, these processes are in accordance with processes that
are described elsewhere in this book. At the same time, it has been
argued that, due to the special sociolinguistic conditions characterizing
the development of pidgins and creoles, these processes are not exactly
the same as those found elsewhere in languages (see, e.g., Keesing :
; Bruyn ). There are in fact some features of grammatical
change that have so far not been encountered in non-pidgins and
non-creoles.
Arguably the most spectacular observation on the modern pidgins
and creoles is that they are young languages: They arose, as far as we
know, in the course of the last four centuries, and some are less than
years old. This fact can be held responsible for a number of special
hypotheses, especially the ones listed in ().
() Some hypotheses on grammaticalization in pidgins and creoles
a. It is shaped by language-external factors.
b. Its processes must have taken place more rapidly than
elsewhere.
c. It is more advanced in creoles than in pidgins.
d. Its processes are on the whole less advanced than elsewhere.
We will now review these hypotheses in turn. As we saw in the
preceding sections of this chapter, (a) is not really a characteristic
of pidgins or creoles; rather, language-external factors can be identified
more generally in situations of intense language contact.
There is in fact empirical support for (b), as the case of Kenya
Pidgin Swahili may show. This pidgin presumably emerged around the
end of the nineteenth century in up-country Kenya as a result of the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. PIDGINS AND CREOLES
British occupation of the country.⁸ The pidgin fell into disuse towards
the end of the twentieth century. When the pidgin was documented in
the early s (Heine , ; Heine and Kuteva : –), it
had already developed a range of new grammatical categories by
extending the linguistic resources that it had inherited from coastal
Swahili to new contexts suggesting new grammatical meanings (see
Section .). For example, the use of the verb toka ‘come from,’ illus-
trated in (a), was extended from nominal to verbal complements and
developed into an immediate past tense auxiliary, as in (b).
() Kenya Pidgin Swahili (Heine and Kuteva : )
a. Mimi na-toka taun.
I -come.from town
‘I am coming from town.’
b. Mimi na-toka andika barua.
I -come.from write letter
‘I just wrote a letter.’
Grammaticalization in Kenya Pidgin Swahili affected all major structural
domains, and it was generally in accordance with major pathways of
grammatical change. For example, in a number of languages across
the world, including English (that), demonstrative pronouns were
grammaticalized to relative clause markers (see Kuteva et al. ,
> ). This development also took place in
Kenya Pidgin Swahili, where the distal demonstrative ile ‘that,’ illustrated
in (a), was grammaticalized to a relative clause marker, as in (b).
() Kenya Pidgin Swahili (Heine : –)
a. wewe hapana ona daraja ile?
you see bridge that
‘Don’t you see that bridge?’
b. paka wewe na-fika katika mahali ile nyang’ao
until you -arrive in place game
ya kila kabila na-patikana.
of each tribe -be.available
‘until you arrive at a place where there are wild animals of any
kind’
⁸ Kenya Pidgin Swahili, historically derived from Swahili varieties spoken natively along
the East African coast, appears to be extinct by now.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
Thus, in accordance with (b), Kenya Pidgin Swahili underwent a
number of grammaticalizations within a period of roughly eighty years
since it emerged towards the end of the nineteenth century. Such speed
in grammatical change is presumably hard to find in languages other
than pidgins and creoles, where such changes normally take centuries
to materialize.
A similar development took place in the English-based creole Sranan
of Suriname (Bruyn a, b, ). Sranan arose in the late
seventeenth century on the plantations of Suriname, which was made
a British Colony in , and in , Dutch became the colonial
language. Probably around the end of the seventeenth century, the
demonstrative disi (or diesi) ‘this’ (< English this) developed into a
relative clause marker within the short period of half a century. An
example from the eighteenth century is provided in ().
() Sranan (English-based creole; Bruyn a: )
Hoe fa mi zel fom wan zomma diesi
manner beat . person
no doe ogeri.
do harm
‘How would I beat someone who didn’t do any harm.’
But in accordance with the hypothesis in (c), the process in Sranan is
clearly more advanced than in Kenya Pidgin Swahili. First, disi under-
went a further grammaticalization, namely one from relative clause
marker to adverbial clause subordinator. In the second half of the
eighteenth century, if not earlier, its use was extended to introduce
temporal, causal, and concessive clauses, as can be seen in (), where it
serves as a temporal conjunction.
() Sranan (English-based creole; Bryun : )
ary fadom trange disi mi de na gron.
rain fall.down strong be at ground
‘Rain fell heavily when I was in the fields.’ (Source from )
And second, the clause subordinator disi underwent erosion (see
Section .), being reduced to di in the course of the nineteenth
century, even if the reduced form is already attested earlier, as example
() shows.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. PIDGINS AND CREOLES
() Sranan (English-based creole; Bryun : )
di ju brokko mi nefi, ju musse gi mi wan
break knife must give a
so srefiwan.
so same.one
‘Since you’ve broken my knife, you must give me a similar one.’
(Source from )
Considering their short history, (d) is to be expected in pidgins and
creoles: Grammaticalization processes are on the whole less advanced
than in other languages. Grammatical functions are predominantly
expressed by lexical items, particles, and clitics and fairly rarely by
affixes—that is, there is a relatively low degree of decategorialization
(see Section .) in pidgins and creoles. For example, in coastal Swahili,
from which Kenya Pidgin Swahili is historically derived, verbal nega-
tion is expressed by the circumfix h- . . . -i, as shown in (). Speakers of
Kenya Pidgin Swahili have replaced the circumfix by extending the use
of the interjective particle hapana ‘No!’ to that of a general marker of
verbal negation, as the corresponding utterance in () shows.
() Coastal Swahili (Bantu, Niger-Congo; Heine and Kuteva :
)
h-a-sem-i kitu.
--say- thing
‘He doesn’t say anything.’
() Kenya Pidgin Swahili (Heine and Kuteva : )
Yeye hapana sema kitu.
s/he say thing
‘He doesn’t say anything.’
In the introduction to this section we noted that pidgins and creoles are
commonly assumed to be ‘extreme’ results of language contact. How-
ever, once they evolved as languages of their own, language contact may
no longer play any significant role in their further development. In fact,
many of the grammaticalizations that have been identified—including
processes presented above—were possibly not affected by contact.
Rather, these processes could as well be the result of internal develop-
ments whereby speakers of pidgins and creoles exploited the linguistic
resources available to them to create new forms of expression via
grammaticalization. Unfortunately there is little detailed information
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
for most of the languages in question about how exactly grammatical
change proceeded.
Nevertheless, there are also documented cases where grammaticali-
zation in pidgins and creoles apears to have been contact-induced. In
concluding, one example may suffice to demonstrate this. Like other
Eastern Oceanic languages, the Kwaio language of Solomon Islands has
a deictic distinction between andative (kau ‘thither’) and venitive (mai
‘hither’) directional forms, illustrated in ().
() Kwaio (Eastern Oceanic, Austronesian; Keesing : )
ngari-a kau ngari-a mai
take-it thither take-it hither
‘take it (away)’ ‘bring it’
As is argued convincingly by Keesing (), this distinction was
replicated in Solomons Pijin, the English-based pidgin used by Kwaio
speakers, by grammaticalizing (‘calquing’) the Solomons Pijin verbs go
and kam, respectively, cf. ().
() Solomons Pijin (Keesing : )
tek-em go tek-em kam
take- thither take- hither
‘take it (away)’ ‘bring it’
Further reading
Many studies on grammaticalization in pidgins and creoles have already been
mentioned in the preceding sections; they include in particular Boretzky
(), Mühlhäusler (), Romaine (), Holm (, ), Adone
and Plag (), Arends et al. (eds.) (), and Mufwene (). More
detailed information on grammaticalization in pidgins and creoles is found
in Adone and Plag (), Bruyn (a, b, ), Romaine (),
Heine (), and Heine and Kuteva (: –).
7.5 Conclusions
The discussion in the preceding paragraphs takes issue with assumptions
surfacing in some of the linguistic literature. In particular, there is the
assumption made implicitly or explicitly in some linguistic studies that
grammatical structure, or syntax, cannot be ‘borrowed,’ that is,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONCLUSIONS
transferred from one language to another. For example, Sankoff ()
concludes that “[w]hether or not ‘grammar’ or ‘syntax’ can be borrowed
at all is still very much in question. . . . many students of language contact
are convinced that grammatical or syntactic borrowing is impossible or
close to it” (Sankoff : ; see also Silva-Corvalán ). We consider
this to no longer be an issue considering that there is by now abundant
evidence to demonstrate that parts of both grammar and syntax can be
‘borrowed’ or, as we will say here, replicated, and the present chapter
provides further evidence in support of these observations.
More specifically, it has been argued or implied that if a certain
grammatical change has been shown to have taken place in one lan-
guage without external motivation, then this makes it unlikely that this
very change could have been externally motivated in another language
(e.g. Lass ). It would seem that this assumption is in need of
reconsideration. There are many examples to show that one and the
same kind of change can have happened internally in one language
while in another language, contact must have been involved (Thomason
and Kaufman : –). For example, there are reasons to assume that
the English de-volitive future tense, using will as an auxiliary (e.g. Jane will
be here soon), evolved largely language-internally. At the same time, such
a future tense has also been shown to arise as a result of language contact.
Thus, Balkanists have demonstrated that the development of the de-
volitive futures characterizing the languages of the Balkan Sprachbund,
such as in Greek, Tosk Albanian, Romanian, Macedonian, Bulgarian,
Serbian, and Croatian, was—at least to some extent—the result of lan-
guage contact (see Heine and Kuteva : –).
As we argued in the present chapter, grammaticalization does not
behave intrinsically differently in situations of language contact from
grammaticalization not affected by contact. But contact can have a
reinforcing, or even a triggering effect on changes to happen, as the
Basque example in Section . may have shown.
The case of Basque is also well-suited to showing another effect that
contact-induced grammatical change can have. As a result of contact-
induced grammaticalization, speakers adapt their discourse patterns
and grammatical constructions to those of another language with
which they are in contact. They create formulas of equivalence
(Heine ), thereby achieving a higher degree of intertranslatability
and structural similarity. Thus, there is reason to assume that the
languages involved become typologically increasingly alike.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION IN LANGUAGE CONTACT
On the other hand, the following example from Basque suggests that
the situation can be more complex (see Haase : for more
details). At some earlier stage, Basque speakers were using hi (‘you,
singular’) as a second person singular pronoun and zu (‘you, plural’) for
second person plural. The neighboring Romance languages Gascon and
French have a second person singular distinction between an informal/
familiar form and a formal/polite form, e.g. French tu (‘you,’ singular)
vs. vous (‘you,’ plural). In an attempt to replicate the polite form they
recognized in their Romance model languages, Basque speakers did
what the speakers of the model languages had apparently done centu-
ries earlier, namely extend the meaning of their plural pronoun zu to
uses as a polite second person singular pronoun. Unlike their model
languages, however, Basque speakers developed a new plural pronoun,
namely zu-ek, consisting of the old second person plural pronoun zu
and the plural marker -ek added to it. At the same time, the meaning
of the Basque pronouns changed: hi became restricted to highly famil-
iar second person reference while zu turned into the general marker of
second person address, and zu-ek is the form for second person plural.
The result is that there is no longer structural isomorphism between
Basque and the Romance languages: Neither does the meaning of hi
clearly correspond to the modern reflexes of Romance tu, nor does zu
correspond to either tu or vous. Instead, zu can be used for both in
many contexts.
What this example suggests is that speakers aiming to establish
intertranslatability and structural isomorphism with their model lan-
guage or languages may actually achieve the contrary: Instead of creat-
ing a structural one-to-one correspondence between the model and the
replica language, they can end up with a situation where the grammat-
ical categories in the languages in contact are less equivalent than they
used to be prior to language contact. More generally, this example is
meant to show that more research is needed to reconstruct all the
motivations that language users have when they change their discourse
options in the direction of another language (see Chapter ).
In spite of such observations, all the evidence that is available on
contact-induced grammatical change, such as the one presented in the
present chapter, suggests, first, that grammaticalization is a ubiquitous
phenomenon in situations of language contact and, second, that it does
not seem to behave intrinsically differently from grammaticalization
not involving contact.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCUSSION POINTS
Further reading
On the distinction between internal and external processes of grammatical
change, see, for example, Thomason and Kaufman (), Thomason (:
), Johanson (, ). Case studies of contact-induced grammatical
change are found in a number of collected volumes on language contact,
especially in Aikhenvald and Dixon (eds.) (, ). The relation between
language contact and grammaticalization is discussed in particular in
Aikhenvald (, , ); Heine (), Heine and Kuteva (,
: –, : –), and Heine ().
Discussion points
() Why do you think would borrowing or replicating grammatical patterns
be more difficult than borrowing from the lexicon of another language?
() What do you think might be the advantage of replicating a grammatical
pattern from a neighboring language instead of simply borrowing it?
() What might be the conditions for a clear case of borrowing and replication
in a pidgin and a creole language?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
8
Grammaticalization and
language typology
8.1 Typological features guiding grammaticalization
8.2 Grammaticalization as a possible explanation for typological
features of languages
8.3 Conclusion
Discussion points
Language typology can be defined as “the study of structural differences
and similarities between languages. . . . . [It] is the study and interpre-
tation of linguistic or language types” (Velupillai : ). The rela-
tionship between grammaticalization and language typology has
already been referred to at several points in this book. Recall, for
example, the typological patterns emerging as a result of grammatical-
ization (Section .). In this chapter we will provide a more systematic
view of the issue, exploring how typological features may influence
grammaticalization, on the one hand (Section .), and how grammat-
icalization may lead to the formation of typological features, on the
other (Section .). The issue of reconstruction is primarily related to
the latter, that is, the case when principles of grammaticalization can
inform reconstruction of the typological features of a language. On the
other hand, reconstruction making use of grammaticalization may also
refer to typological features of language that guide grammaticalization.
8.1 Typological features guiding grammaticalization
In Chapter , we addressed the question what may motivate or ‘drive’
grammaticalization. Typological features of language would fall into
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. TYPOLOGICAL FEATURES GUIDING GRAMMATICALIZATION
the basket of potential structural motivations (Section .). We con-
cluded then that while structural motivations are controversial, extant
language structures surely guide the course of grammaticalization.
Many of these structural properties can be typologized and are there-
fore potential material for the influence that typological features of a
language may exert on grammaticalization.
Park () compared the grammaticalization of verbs in three
Tibeto-Burman languages with typologically divergent features, namely
highly inflecting Hayu, isolating Burmese, and in-between Tibetan,
concluding that both similarities and differences between the languages
are due to their respective typological features. That is, for Park (:
) “grammatical structure is a better predictor of grammaticalization
patterns than is genetic affiliation.” The reason is simply that “struc-
tures that are available in a language are utilized in the tailoring of new
features” (Park (: –).
How does this come about? First, as seen in Chapter , motivations
for grammaticalization itself must be sought in cognition and syn-
chronic language activity. This is a point readily acknowledged by
scholars who engaged in the study of the relationship between syn-
chrony and diachrony. For example, Greenberg (: ) suggested
that “ . . . there is a gradual process by which innovations originating
in discourse ultimately become established as part of the system of
language.” Or, Hawkins (: ) considers grammars as the “con-
ventionalizations of the same processing mechanisms that psycholo-
gists find evidence for in experimental and corpus data.” When it
comes to the shape that new grammaticalizations take, or the paths
through which they grammaticalize, it is only reasonable to assume that
extant language structures, including typologically relevant structures,
play a role as speakers can utilize existent channels of expression more
efficiently than creating entirely new ones. It is difficult, if not impos-
sible, though, to demonstrate a clear cause–effect relationship between
older structures and the form of emerging structures, since one-to-one
replication of an older structure would be a rarity.
The extent to which features of grammaticalization are analyzed as
being influenced by typological features will also depend on the defi-
nition of grammaticalization adopted. Especially, the more grammati-
calization is reduced to a universal essence, the less room there is for
typological variation. Himmelmann (: ), for example, claims
that “a grammaticisation process can be defined as a process of context
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
expansion” on the three levels of (a) host class formation/expansion,
(b) syntactic context expansion, and (c) semantic-pragmatic context
expansion. This definition intentionally excludes typological factors
from grammaticalization, which are viewed as epiphenomenal. Likewise,
in generative approaches to grammaticalization as well (Section .),
grammaticalization tends to be reduced to an essence that is universal
and therefore less amenable to typological influences. As we saw,
Roberts and Roussou (: ) regard grammaticalization “as the
diachronic development of lexical heads into functional heads,” and
van Gelderen () conceptualizes grammaticalization as change from
lower head to higher head (‘Late Merge’) or from Spec to Head. There
might be typological variation in the heads that are the sources and the
targets of change, but the essence will remain unaffected. However,
while acknowledging that the essence of grammaticalization is on the
functional side, we have also kept the formal side of the process within
the scope of this book. Recall that the process of grammaticalization
can be divided into the four basic aspects discussed in length in
Chapter , and listed here again in ().
() a. extension (or context generalization)—use in new contexts;
b. desemanticization—loss (or generalization) in meaning content;
c. decategorialization—loss in morphosyntactic properties char-
acteristic of lexical or other less grammaticalized forms; and
d. erosion (or “phonetic reduction”)—loss in phonetic substance.
While the pragmatic-semantic processes (a) and (b) are essential for
grammaticalization, the structural processes given in (c) and (d) may
follow. If there is typological variation in grammaticalization, it may in
principle affect all aspects of grammaticalization. There is no reason to
exclude some aspect a priori. However, if functional change is the essence
of grammaticalization and other aspects follow in the order given above,
we may assume that the magnitude in which typological variation in
grammaticalization occurs is reversely (d) > (c) > (b) > (a). That is, the
formal reductive changes should be most susceptible to variation.
So, we start with examples of how phonological and morphological
aspects of grammaticalization may be influenced or guided by the
corresponding typological features of a language or a group of languages.
A well-known example of that is the claim that grammaticalization
in tonal, isolating languages, like the Sinitic languages does not lead
to reduction of syllables. Ansaldo and Lim (: ) suggest that
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. TYPOLOGICAL FEATURES GUIDING GRAMMATICALIZATION
“[s]trongly isolating languages typically do not allow yesterday’s syntax
to become today’s morphology . . . syllable boundaries are discrete and
phonotactic constraints rule out reduced syllables of the kind observed
elsewhere, the material available for reduction is not easily found at the
morphological level.” However, even in these languages grammaticaliz-
ing words tend to undergo subtle changes in pronunciation. The
authors demonstrate that in two morphologically strongly isolating
and tonally complex Sinitic languages spoken in Singapore, namely
Cantonese and Hokkien, function words show vowel/syllable reduction
and erosion typical of grammaticalization. It is just the extent of
reduction that is much smaller. Bisang (, , ) claims a
number of constraints on grammaticalization in Sinitic languages,
which stand representatively for the East and mainland Southeast
Asian languages. Besides the small extent of changes in phonology
and morphology, these also include the lack of obligatoriness of
grammatical categories; that is, the lack of grammatical paradigms in
general. Likewise, Ansaldo et al. () claim that “elaboration of
morphological structure only happens in a certain type of languages.”
That is, “the formal aspects of canonical grammaticalization do not
happen in [East and mainland Southeast Asian] languages.” Bisang
() goes one step further and argues that different processes are
at work in grammaticalization: In European languages like English
or German “explicitness-based maturation” usually takes place.
“Explicitness-based” is a term coined by Dahl () and stands for
the accumulation of grammatical material that was absent at an earlier
stage in the grammar of a language. In Sinitic and many other East and
mainland Southeast Asian languages, by contrast, “economy-based
maturation” dominates in processes of grammaticalization. In these
languages, little grammatical material is accumulated, leading to a
reduced coevolution of meaning and form, and high relevance of
discourse and pragmatic inference. Grammaticalization in turn influ-
ences extant structures, thus potentially leading to a cycle in which
structures of a specific type are recreated.
In contrast to the “isolating” languages of East and mainland South-
east Asia, Narrog et al. (; also Narrog and Ohori ) point out
that morphological parameters of grammaticalization seem to apply
particularly well to Northeast Asian languages like Korean and
Japanese, and potentially to the so-called Transeurasian languages in
general, which tend to be agglutinating. Similarly Mithun ()
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
emphasizes the relatively high degree of morphological complexity as a
product of extensive grammaticalization in North American languages.
On the other hand, the fact that morphological complexity is the result
of grammaticalization also points to the fact that there is a two-way
relationship between grammaticalization and typological features, that is,
typological features do not unilaterally determine features of grammati-
calization. In Indo-European languages, the inflectional endings on verbs
and nouns that modern European languages have still preserved (but to a
good part already lost) are likely to be a remnant of grammaticalization in
head-final proto-Indo-European (cf. Dahl : ).
Besides phonological and morphological typological features, there
are also synchronic semantic and syntactic features of languages that
may guide grammaticalization. First, as discussed in more detail in
Section ., in head-final languages grammaticalization is more likely
to lead to bound morphemes than in head-initial languages because of
the tendency for postposed rather than preposed morphemes to
become bound. Second, Bisang (, , ), also counts the
following among the constraints on grammaticalization in Sinitic lan-
guages: (i) lack of clearly determined semantic domains; (ii) existence
of rigid word-order patterns within which lexical items grammaticalize,
and pervasiveness of inference, which enables language users to encode
and decode the function of a specific item in a specific semantic context;
and (iii) lack of grammaticalization chains (i.e. continuous grammati-
calization from one category to the next). Similarly, Xing (, )
proposes that while in European languages, chain-like developments
along the schema A > AB > B may be common (cf. Heine : :
xAb > aBc > bC),¹ in Chinese languages, older meanings and functions
are typically retained, leading to “clines of semantic accretion” of the
type A > AB > ABC.² According to Ansaldo et al. (: ), “wide-
spread polyfunctionality undermines the semantic dimension of
canonical grammaticalization in [East and mainland Southeast Asian]
languages.” Therefore, none of Lehmann’s () commonly cited six
parameters of grammaticalization applies except syntagmatic variability
(see Section .). But even this parameter, according to Ansaldo et al.
(: ) may be dubious: “An aspect of grammaticalization in this area
¹ The letters “x, a, b, c, d” stand for specific meanings and functions. Uppercase letters
signify focal senses and lowercase letters non-focal senses.
² See also the concept of ‘persistence’ in Section ..
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. TYPOLOGICAL FEATURES GUIDING GRAMMATICALIZATION
may be the loss of autonomy, or constructionalization, but even this is
undermined by polyfunctionality and lack of obligatory marking.”
Morphologically highly complex languages again seem to be at the
other end of the spectrum. For example, according to Esseesy (: )
“there appears to be no typological limit found on the evolution of
meaning and form in Semitic of the type described in Bisang’s ()
study.” Something similar can be said of North Caucasian languages
(Arkadiev and Maisak ), Oceanic languages (Moyse-Faurie ),
Northeast Asian languages (Narrog et al. ), Iranian languages
(Haig ), and most North and South American languages (Mithun
; Aikhenvald ).
Another way in which extant typological structures guide new gram-
maticalizations is the renewal and structural elaboration of extant
grammatical functions. Ideally, every language would have just one
exponent for every cross-linguistically available grammatical category.
In reality, in any given language, expressions for specific categories may
be grammaticalized copiously, while other categories are not gramma-
ticalized at all. These categories may then be expressed indirectly
through other categories, lexically, or not at all. As a result, a certain
set of categories is grammaticalized over and over in waves or cycles
(cf. Section .). Those dominant categories may further impact the
way other categories are grammaticalized.
In the area of tense-aspect-mood (TAM), Bhat () suggests that
languages tend to be either tense-, aspect-, or mood prominent, and the
not overtly expressed categories are often indirectly expressed through
the well-grammaticalized ones. In terms of grammaticalization, we can
surmise that languages tend to maintain this typological profile by
repeatedly grammaticalizing their preferred category rather than
grammaticalizing the “neglected” ones. Bhat (: ) even suggests
different paths of grammaticalization for each type of language:
“[L]anguages that give greater prominence to aspect than to tense
develop a perfective form from an earlier perfect construction and an
imperfective form from an earlier progressive construction, whereas
languages that give greater prominence to tense than to aspect develop
past and present forms directly from their perfect and progressive
constructions respectively.” Furthermore, “ . . . aspect-prominent and
mood-prominent languages show distinct tendencies of change when
they develop temporal distinctions. In the case of aspect-prominent
languages, we generally find a two-way past/non-past distinction or a
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
three-way past-present-future distinction developing from an earlier
perfective-imperfective distinction. . . . In the case of mood-prominent
languages, on the other hand, the general tendency is to develop
primarily a future/non-future distinction” (Bhat : –).
A related phenomenon is known under the term ‘florescence.’ Chafe
(: ) defines it as follows: “[l]ike forests, languages may develop
toward a climax stage where particular combinations of features, like
plant communities, may flourish to define a particular language type.
I think it is useful to think in terms of the florescence of linguistic
features in this sense—the flowering of features that come to dominate
the form a language takes.” Some features of Iroquoian languages serve
as examples. First, Northern Iroquoian languages especially can have
an elaborate inventory of up to ~ pronominal prefixes. Besides
singular, plural, and dual number distinctions in the first and second
persons of both agents and patients, Cherokee has even an inclusive–
exclusive distinction and gender distinctions with third persons
(cf. Chafe : ). This elaborate pronominal prefix system is de-
scribed (reconstructed) by Chafe as the result of not one but successive
waves of grammaticalizations. Secondly, Iroquoian languages are char-
acterized by noun incorporation of a wide range of categories, for
example animals, foods, and body parts. Even whole events can be
incorporated into nouns as verb roots. In this case as well, inter-
language comparison suggests that these incorporations developed
not at once but successively, newer incorporations following older
ones. They also range from productive to idiomatic.
In languages such as English or German, the relatively large variety
of modal and semi-modal verbs may stand for the same phenomenon,
although surely not with the same abundance. Historically speaking,
there was a wave of emergence of the modals from Old to Middle
English, and another wave of emergence of the semi-modals from
Middle to Modern English. The prior emergence of one or two auxil-
iary verbs that pioneered this move seems to have drawn others along
the same path, and the modals as extant structures probably induced
the later development of the semi-modals. Krug () explained the
development of the semi-modals in terms of a ‘gravitational model,’
which operates on the principle that “larger masses (in our case
highly frequent emerging auxiliaries) attract smaller masses (in our
case less common constructions)” (Krug : ). Based on fre-
quency and similarity, Krug (: ) calculated to what extent
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. TYPOLOGICAL FEATURES GUIDING GRAMMATICALIZATION
specific semi-modals are influencing others. More generally then,
grammatical items and constructions with low frequency change in
analogy to high frequency items and constructions. Type frequency
determines the influence of a group of items on other items outside that
group. This was already noted by Paul (: ), who writes that
“ . . . all that part of language which lacks the support of an environing
group, or which enjoys it only in a limited measure, proves, unless
impressed by repeated usage intensely upon the memory, not strong
enough to withstand the power of the larger groups.”³ Bybee and
Thompson (: ) likewise suggest that, “high type frequency
ensures that a construction will be used frequently, which will
strengthen its representational schema, making it more accessible for
further use, possibly with new items.” In contrast to type frequency,
token frequency may determine the item or construction with the
biggest influence within a specific group.
In this manner, the dominant structural and functional types of a
language offer well-trodden paths for new grammaticalizations. Dom-
inant structural and functional types that grammaticalize repeatedly are
highly diverse across different languages. They include the extensive
range of aspectual and actional categories that is grammaticalized across
Turkic languages (Johanson and Csató ), hundreds of instrumental
affixes in many North American languages such as Kutenai (Mithun
), applicative prefixes in Papuan languages (Klamer ), body-
part nouns in South American (Zariquiey ) and Caucasian lan-
guages (Arkadiev and Maisak ), or serial verbs in West African,
Sino-Tibetan, and in Oceanic languages (Lord , ; Park ;
Hwang ; Moyse-Faurie ).
On the other hand, forces influencing grammaticalizations are not
restricted to extant language- or language-area-specific structures.
DeLancey (: ) suggests that “a function which is important
enough, cross-linguistically, that in [a] language which does not for-
mally express it with dedicated grammatical machinery, any construc-
tion or lexical means which expresses a related function is a likely
candidate for grammaticalization.” He labels this kind of cross-
³ In the German original Paul (: ), this reads as ,“Alles dasjenige aber, was die Stütze
durch eine Gruppe entbehrt oder nur in geringem Masse geniesst, ist, wenn es nicht durch
häufige Wiederholung besonders intensiv dem Gedächtnisse eingeprägt wird, nicht wide-
rstandsfähig genug gegen die Macht der grösseren Gruppen.”
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
linguistically salient function as a ‘functional sink.’ For example, the
formation of adjectives is based on the function of noun modification,
which is universal, even if adjectives as a part of speech are not. Certain
nouns or verbs may be drawn into this functional sink, eventually
leading to the development of a new category adjective in a given
language. In the domain of grammar, Thornes () claims that such a
functional sink is at work in the grammaticalization of causative construc-
tions in Northern Paiute. In his view, causative is a grammatical function
with a high communicative need that is usually available in some form in
any language, even in the absence of grammatical means. Because of the
frequent need of expression, it attracts lexical materials to grammaticalize.
In summary, structural typological features seem to perpetuate
themselves horizontally, so to speak, across languages and vertically
across the history of individual languages. Very generally, we may
assume that, unless there is some disruption, the typological morpho-
logical tendencies of a language result in the extant morphological types
as target structures; that is, for example in languages with agglutinative
morphology, grammaticalization is more likely to lead to affixation
than in isolating languages.
Further reading
The claim about the singularity of features of grammaticalization in South-
east Asian languages has not gone undisputed. Ansaldo and Lim ()
showed that some phonetic reduction does take place. Arcodia (), taking
examples from aspect markers in North Chinese dialects, argues against all
the alleged distinctive features except reliance on pragmatic inference. Dahl
() suggests that under careful consideration of the data, the differences in
grammaticalization between languages in Southeast Asia and, for example,
Europe may not be as large as they appear to be. Robbeets () argues in
favor of genealogically motivated grammaticalization.
8.2 Grammaticalization as a possible explanation
for typological features of languages
Grammaticalization can also serve at least as a partial explanation for
certain typological features of languages. At least four cases of this kind
have been put forward. First, grammaticalization has been proposed as
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION AS A POSSIBLE EXPLANATION FOR TYPOLOGY
a partial explanation for some implicational universals. Secondly, gram-
maticalization has been proffered as a (partial) explanation for the order
of affixed material. Thirdly, grammaticalization has been given as an
explanation for cross-linguistic types of the expression of certain gram-
matical categories, and lastly, grammaticalization has been seen as instru-
mental in creating so-called analytic (vs. synthetic) language structures.
8.2.1 Grammaticalization and word order
With respect to the first case, Lehmann (: ) suggested that the
process of grammaticalization can be taken as a causal factor for some
of Greenberg’s implicational universals of word order (the numbers
refer to the numbering of the forty-five universals proposed in the
seminal paper of Greenberg (b); SOV = subject–object–verb,
VSO = verb–subject–object. The numbering refers to that of the listing
in Greenberg (b)).
. In languages with prepositions, the genitive almost always fol-
lows the governing noun, while in languages with postpositions
it almost always precedes.
. Languages with dominant VSO order are always prepositional.
. With overwhelmingly greater than chance frequency, languages
with normal SOV order are postpositional.
. In languages with dominant order VSO, an inflected auxiliary
always precedes the main verb. In languages with dominant
order SOV, an inflected auxiliary always follows the main verb.
The above generalizations can all be explained by the fact that when
lexical heads of complex constructions grammaticalize, they usually
remain in their original position. For example, relational nouns gram-
maticalize to adpositions (, , ), and verbs grammaticalize to auxili-
aries () and remain in their original position.
However, Lehmann (: ) does not consider grammaticalization
as the ultimate cause or motivation for the order and relationship
between elements of the sentence. Instead, it is a channel of change
from the lexical to the grammatical category, because “the specific
function common to the syntactic relations of a relational noun to its
complement and of an adposition to its complement, and the specific
function common to the syntactic relations of a verb to its object and of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
an adposition to its complement, are concepts not inherent in gram-
maticalization, but are, instead, a presupposition for these construc-
tions to be connected by grammaticalization.” Thus, Lehmann argues
that grammaticalization does not serve here as a full explanation for the
change.
Relatedly, Greenberg himself () referred to grammaticalization
within a four-part approach to diachronic typology, which consists of
the following elements (AN = adjective–noun, NA = noun–adjective,
OV = object–verb, VO = verb–object).
(i) The dynamicized state-process model: Synchronically, there are
certain states within a specific typology, within a specific lan-
guage; e.g. OV (vs. VO); AN (vs. NA) etc. The typology de-
scribes the process by which the language changes from one
state to another. In order for this model to lead to interesting
generalizations, a ‘strong connection’ must hold, i.e. it must be
possible to move from one state to another within that
typology.
(ii) The dynamicized state-process model of an elaborate (or: sub-)
typology: This model looks at intermediate states between the
main states in a typology; and tries to find generalizations about
the diachronic relationship between these states.
(iii) Intragenetic comparison: comparison within one genetic lan-
guage group.
(iv) Intergenetic comparison: comparison between two or more
genetic language groups.
Within this approach, Greenberg (: , –) primarily saw a
significant role for grammaticalization in intragenetic comparison,
where grammaticalization theory provides knowledge about the direc-
tionality of change, but secondarily also in intergenetic comparison,
where grammaticalization theory does the same on a larger scale.
Greenberg was mainly interested in how grammaticalization interacts
with global constituent order and word order changes to come up with
detailed variations in word order within one language, and counter-
examples to implicational universals. In the reconstructed stages of
word order development in Ethiopian Semitic languages in (), as an
example of intragenetic comparison, the highlighted part of the chain
constitutes a violation of implicational universal above.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION AS A POSSIBLE EXPLANATION FOR TYPOLOGY
() Pr/NG/NA ! Pr/NG/AN ! Pr/GN/AN ! Pp/GN/AN (Greenberg
: ; A = adjective, G = genitive; N = noun, Pr = preposition,
Pp = postposition)
This violation can be explained through the interplay of principles of
grammaticalization, on the one hand, and global constituent order and
word order change in this language, on the other. Noun–adjective order
is the type of word order that is the least stable and the most susceptible
to change. It will change first, followed by noun–genitive order, while
grammaticalization from relational noun to adposition will take the
most time, and therefore adpositions will lag behind, leading to the
apparent violation of universals.
Lehmann (: –) also discussed the role of grammaticaliza-
tion in this type of historically conditioned intra-language (and intra-
genetic) variation. However, he argues that not constituent order and
word order (within the clause), but the area on the borderline between
syntax and morphology should be at the center of the study of gram-
maticalization and linguistic typology. According to him, high-level
syntax like constituent order is often pragmatically and not grammat-
ically determined. Pure meaning (semantics) and expression (phonol-
ogy) are also excluded because it is the combination of meaning and
form that is typical for language. The lexicon (lexical typology) is not
regular enough. Indeed, as predicted by Lehmann (), the study of
the role of grammaticalization in the development of constituent order
and word order (within the clause) has remained at the fringe of
typologically oriented research in grammaticalization.
8.2.2 Grammaticalization, the suffixing preference,
and morpheme order
A second case in which grammaticalization is at least partially respon-
sible for typological structures is the order of affixed material. There are
at least two related topics in linguistic typology, namely the suffixing
preference, and morpheme order in complex words. To start out with
the former, there is an overall tendency in languages to prefer suffixes
over prefixes, therefore called suffixing preference, as has been well-
known for a long time. As Bybee et al. () demonstrated, this
tendency even holds in the majority of head-initial, especially SVO,
languages. Table . shows the overall preference for suffixation,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
Table 8.1 Prefixing versus suffixing in inflectional morphology (Dryer : )
Number of languages
Little or no inflectional morphology
Predominantly suffixing
Moderate preference for suffixing
Approximately equal amounts of suffixing and prefixing
Moderate preference for prefixing
Predominantly prefixing
Total
irrespective of basic word order, according to cross-linguistic data by
Dryer ().
Although the discussion on the suffixing preference has continued,
Bybee et al. () is probably still the most detailed study on the extent of
the suffixation preference and its causes. The authors give an overview of
the suffixation preference, overall and by word order type, investigate a
number of psycholinguistic (processing) and phonological factors possi-
bly leading to the suffixing preference, and find that none of these factors
can account for it. They conclude that the “fossilized syntax hypothesis”
explains the suffixing preference best (Bybee et al. : ).
The “fossilized syntax hypothesis” goes back to the idea by Givón
() that “yesterday’s syntax is today’s morphology.” It says that “the
position of an affix is the same as the position of the non-bound lexical
or grammatical material from which the affix developed” (Bybee et al.
: ). It is this idea of fossilized syntax that led to the revival of
interest in grammaticalization since the s.
A remarkable fact about fossilized syntax is that it does not neces-
sarily reflect normal word order. The prevalence of suffixed person
endings even in SOV languages, where subjects precede verbs, is
the most salient example for that. As a response to this apparent
divergence between word order and morpheme order, it has been
hypothesized that unstressed pronominal subject pronouns postposed
to the verb, rather than pronouns in their normal position, get
grammaticalized (cf. Bybee et al. : ). Further research has shown
that within person paradigms, there is a prefixing preference both for
very small and very large paradigms, while medium-size paradigms
are predominantly suffixing (Cysouw ). Cysouw concludes that
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION AS A POSSIBLE EXPLANATION FOR TYPOLOGY
“[t]he big riddle of the suffixation preference thus actually consists of
various smaller-scale riddles concerning different kinds of affixation
asymmetry” (Cysouw : ). Mithun () likewise suggests that
the ultimate answer to the suffixing preference must be sought in the
history (i.e. in the grammaticalization) of the individual morphemes that
as an aggregate make up the suffixing preference.
In general, the position of a bound morpheme with respect to the
lexical stem should reflect the order of the erstwhile independent word
vis-à-vis its host, unless one can make the well-founded assumption
that a morpheme changed its position after grammaticalization. While
Harris and Campbell (: –) refer rather abstractly to cases
where reanalysis in grammaticalization-like changes led to a change
in position, these cases should be considered as exceptional, since
bound morphemes are far less mobile than independent ones. How-
ever, a change of position of a bound morpheme becomes more likely if
the morpheme goes through a clitic stage. Comrie (: –, :
–) suggested that the position of clitics is freer than that of other
bound morphemes, and that for prosodic reasons, clitics may follow
rules differing from those for independent words. Thus, if a clitic stage
is involved in the grammaticalization of a morpheme, the likelihood
that morpheme order does not reflect erstwhile word order increases.
In any case, it does seem that grammaticalization is the most important
mechanism behind the suffixing preference. Of course, grammaticaliza-
tion as such usually cannot explain why an element is in a position before
or after a lexical stem at the time when it grammaticalizes.
Beyond the phenomenon of the suffixing preference, morpheme
order among affixes in morphologically complex languages is also an
intriguing problem that is not fully resolved. But grammaticalization
appears to be at least one important motivation, as has been shown by
Mithun () in her study on the Navajo verb. We have provided the
details in Section . (Figure .).
Recall that one important generalization in morphology that was
violated by the Navajo morpheme order was that inflection should
occur ‘outside’ derivation in a given word. This generalization is graph-
ically represented in Figure ..
As a result of consecutive grammaticalization, it can happen that a
newly evolved derivational suffix is separated from the root of that word
by an inflectional suffix which was already part of the word, thereby
resulting in a violation of the typological principle captured in Figure ..
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
Inflectional Derivational ROOT Derivational Inflectional
prefixes prefixes suffixes suffixes
Figure 8.1 Inflection outside of derivation (Mithun : )
In such a case, thus, grammaticalization has created the structure
[ROOT–Inflection–Derivation] which is at variance with the typological
generalization of Figure .. Hence it is grammaticalization that explains
morpheme order best. Also, in contrast to the case of the suffixing
preference, in the case of morpheme order, grammaticalization appears
to be the immediate cause. According to Mithun (: ), the same
generalization holds even more in general. She concludes that “to a
significant extent, the order of affixes within words reflects their historical
order of attachment. Those affixes closest to the root were generally
grammaticalized earlier than those on the periphery of the word.”⁴
On the other hand, conflicts between general rules of morpheme
order and order of grammaticalization are sometimes subsequently
resolved by reordering mechanisms such as externalization of inflection
(cf. Haspelmath b).
Mithun (: –) provides an example from the Eskimoan
language Yup’ik. In this language, derivational suffixes are placed
next to the verbal or nominal root and they are followed by inflectional
suffixes in accordance with the typological generalization of Figure ..
Now, some derivational nominalizers, such as Yup’ik -ller- ‘the former’
(noun), evolved into inflectional mood suffixes. For example, the
nominalizer turned into the past contemporative marker -ller- ‘when
(in the past).’ The contradiction with canonical morpheme order was
resolved when the erstwhile nominalizer moved from close to the root
to the right periphery of the word, thereby restoring the order [ROOT–
Derivation–Inflection] in accordance with Figure ..
Mithun (: –) also found a similar development in the
Iroquoian language Cherokee. In this language,
[e]arlier derivational instrumental suffixes, whose ultimate source can be traced to
verb roots meaning ‘use’, evolved into inflectional infinitive markers, and in the
⁴ In a similar fashion, Bybee et al. (: ) conclude that, “[w]e would expect grams
that are older—ie., that have undergone more development—to be closer to the stem, more
fused and shorter or more reduced in segmental material than younger grams of equal
relevance.”
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION AS A POSSIBLE EXPLANATION FOR TYPOLOGY
process shifted their position outward, across a number of other derivational suffixes,
to take their place within the inflectional aspectual paradigm. (Mithun : )
According to Mithun (: ), this shift “did not involve a simple
one-stage hop across morpheme slots, but rather a reinterpretation of
the boundary between derivation and inflection.”
In Bari, an Eastern Nilotic language of Southern Sudan, a similar
development took place on the level of clausal morphosyntax. In this
language, the basic word order is subject–verb–object (SVO) with tense
markers placed between subject and verb. The basic position of adverbs
is clause-initial, cf. (). But when the adverb dé ‘then, afterwards’ was
desemanticized and turned into a future tense marker (FUT) it was
moved from the clause-initial position to the tense slot between subject
and verb, cf. (), thereby restoring the canonical order of clauses.
() Bari (Eastern Nilotic, Nilotic; Heine and Reh : )
a. dé nan kɔn . . .
then I do
‘I shall do . . . then’
b. nan dé kɔn . . .
I do
‘I shall do . . . ’
Examples as the above suggest that typological constraints, like the
arrangement of affixal morphology or clausal constituent order, can be
violated by grammaticalization processes, but at least in some lan-
guages, or in specific structures, speakers seem to make an effort
towards undoing the violation and restoring the canonical order of
meaningful elements. Clarifying the conditions under which this hap-
pens in certain languages and structures and not in others would
require a larger scale investigation.
Overall, though, in the case of morpheme order, grammaticalization
appears to be an immediate cause and can therefore serve as a valid
explanation.
8.2.3 Grammaticalization and expression types
of grammatical categories
The third area in which grammaticalization has been found to motivate
typological patterns is grammaticalization as the source, and hence also
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
explanation for cross-linguistically recurring types of expression of
certain grammatical categories. Many linguistic categories are cross-
linguistically expressed by a limited number of structural types. These
structural types in turn are the product of grammaticalization. Further-
more, the source and the degree of grammaticalization of these struc-
tures can explain at least some of their morphosyntactic and semantic
features. Especially prominent research linking typological patterns
with grammaticalization is associated with Heine (b; Heine and
Kuteva ) and Bybee (; Bybee et al. ). In the following, we
present a number of examples.
Indefinite articles: According to Heine (; Heine and Kuteva ),
about percent of all indefinite articles cross-linguistically are derived
from the numeral ‘one.’ This explains some positional tendencies of
indefinite articles, the fact that they are often confined to singulars, and
the following implicational hierarchy for their application: mass noun >
plural noun > singular noun (see Section . for more details).
Possessive Constructions: Heine (a) identified the eight cross-
linguistic source schemas for possessive constructions see Section .).
The source schemas in Section ., () can account for some character-
istics of specific possessive constructions, such as why they often have
non-verbal, or copular-like, predicates, or how the possessor is encoded
in a specific language, that is, as a comitative, locative, etc., or why they
frequently have locative morphology, etc. (cf. Heine a: –).
Furthermore, these constructions often undergo a development at the
end of which,
() the possessor precedes the possessee;
() the possessor has properties of a subject, and the possessee has
properties of a clausal object;
() the possessor is definite and the possessee is indefinite.
Therefore, possessive constructions often display ‘hybrid’ properties
between source and target structures depending on the stage of
development. In this manner, grammaticalization as a process can
account for the properties of these constructions.
Future: In her analysis of the cross-linguistic polysemy of future
morphemes, Bybee () found that the polysemy can be explained
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION AS A POSSIBLE EXPLANATION FOR TYPOLOGY
by reference to their diachronic evolution, which takes place in the
form of paths, such as from movement (‘go’ or ‘come’) to intention,
then to prediction, the core future meaning, and further to other
meanings such as supposition or imperative, as in Figure . (see
Section . for concrete examples).
movement
characteristic
intentionm behavior
desire intentiond prediction supposition
intentiono
imperative
obligation
Figure 8.2 Paths for the development of future morphemes (Bybee : )
According to Bybee (: –), these paths of development are
explanatory because they explain () why it is difficult to find a single
abstract meaning for a polysemous morpheme (like many future mor-
phemes), () the cross-linguistic similarities of grammatical meanings
by similar paths of development and principles of historical change, and
() differences between morphemes in different languages with refer-
ence to different lexical sources and different extent of change along the
universal paths of change. Furthermore, () they make it possible to
predict possible combinations of meanings, and () they allow the
reconstruction of the lexical sources of grammatical morphemes, and
() they suggest processes in synchrony that are probably instrumental
in explaining the changes and the nature of meaning.
Passives: According to Givón (, ), six common types of
passive constructions can be identified. () the adjectival-stative passive
(e.g. the English common passive); () the reflexive passive (e.g. the
English get-passive); () the serial-verb adverse passive (e.g. in
Chinese); () the VP-nominalization passive (e.g. in the Amerind
language Ute); () the left-dislocation-cum-impersonal-passive (e.g.
in Kimbundu); () the zero-anaphora passive (e.g. in Sherpa). ()–()
are so-called promotional passives (i.e. the erstwhile object is promoted
to subject), while ()–() are non-promotional. Note that the concept
of passives applied here is fairly broad, and not every study of passives
would include the same range of constructions. In any case, the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
concrete structural properties of these passive structures in individual
languages can be explained by the degree to which they have gramma-
ticalized to more prototypical passives. For example, non-promotional
passives can become eventually promotional as subject properties
gradually shift to the object-patient. Similarly, oblique agents may
eventually be added. In this way, types of passives and their morpho-
syntactic features can be explained with reference to their source
construction and their degree of grammaticalization.
Overall, then, grammaticalization cannot only account for the shape
that grammatical categories take in specific languages, but in some
cases even for their specific grammatical features.
8.2.4 Grammaticalization and the cycle of synthetic vs.
analytic expression of categories
Lastly, grammaticalization plays a decisive role in the change from
synthetic to analytic structures, and back to synthetic structures over
longer periods of time. Analytic structures are structures in which each
grammatical category is expressed by a separate word, while in syn-
thetic structures, grammatical formants are affixed to or even fused
with lexical stems. For example, the simple future of chanter ‘to sing’ in
French is chanterai (sing..) ‘I will sing,’ which is considered as
‘synthetic,’ while the newer and more colloquial expression je vais
chanter (‘I am going to sing’) is considered as ‘analytic.’ Inflections
are typical synthetic structures, and their loss in many European
languages in favor of analytic grammatical expressions compared to
older Indo-European languages alerted linguists from the nineteenth
century on to this issue (cf. Ledgeway ).
Thus, changes between analytic and synthetic are a traditional and
long-standing topic in language typology (cf. Kastovsky ; Ramat
). They have already been alluded to in the section on concrete
grammaticalization cycles (Section .). However, in accordance with
the tradition in linguistic typology, a much wider perspective, namely
in terms of changes in the typological profile of a language or language
groups as a whole can also be taken. And grammaticalization is instru-
mental both in the emergence of grammatical categories from the
lexicon, usually starting out as ‘analytic’ structures, and in the aggluti-
nation or even fusion that may or may not follow (cf. Section .), and
that leads to ‘synthetic’ structures. For example, with respect to Semitic
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GRAMMATICALIZATION AS A POSSIBLE EXPLANATION FOR TYPOLOGY
languages, Esseesy () notes that “[g]rammaticalization has been
shown to facilitate the change from the direction of synthetic to analytic
in several Semitic languages . . . and facilitates the transition from one
state to another and in some cases perhaps back in a cyclical fashion.”
Or, referring to Iranian languages, Haig () also emphasizes the role
of grammaticalization in cyclical typological change, suggesting that
“the history of grammaticalization can to some extent be seen as the
gradual re-acquisition of lost morphological categories.” Haspelmath
() coins the term ‘anasynthetic spiral’ for this large-scale cyclical
change. He defines anasynthetic change as,
a change whereby a new analytic construction arises that competes with an earlier
synthetic pattern and grammaticalizes, eventually becoming the primary expression of
its meaning, and thus ‘synthetic again.’ (Haspelmath : )
Going back to the French expressions of future as an example, Latin
had a ‘synthetic’ future of the form cantabō ‘I will sing’ based on the
verb stem canta- sing. The newer synthetic future of French, chanterai,
developed out of the ‘analytic’ Latin construction cantare habeō (to.sing
have.I) ‘I have to sing.’ This is replaced again in Modern French by the
new analytic construction je vais chanter (I go to.sing), that at some
point may fuse and become synthetic as well. The image is that of a
‘spiral’ (von der Gabelentz []; Meillet []) rather
than a cycle, because none of the newer constructions is really identical
to its predecessor. Again, the driving force behind this is grammatical-
ization, while grammaticalization itself is, of course, driven by some-
thing else, that is, not the diachrony, but synchronic motivations in
communication and cognition (Chapter ).
The application of labels such as ‘analytic’ or ‘synthetic’ to languages
as a whole is controversial. Note that even in languages that have
comparatively little agglutinative or inflecting morphology such as
English or Mandarin Chinese, one can still find such structures (e.g.
the English past tense went for go, or plural mice for mouse). Haspelmath
(: ) suggests that the best historical cases that can be made for the
change, namely from Latin to Romance and from Ancient Egyptian to
Coptic, are under conditions of intense language contact (see Chapter ).
On the other hand, the renewal of synthetically expressed categories
through analytically expressed ones is nothing specific to these lan-
guage groups (cf. Ledgeway : ), and the occurrence of ‘anasyn-
thetic’ change in specific categories in a specific language should be
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
much easier to demonstrate. However, its observation in actual histor-
ical data is actually often fraught with difficulties. The development of
inflections in particular usually seems to take a longer period of time
than is available through historical records even in well-documented
languages. Cross-linguistically the origins of flectional/fusional patterns
are mostly unknown (cf. Haspelmath : ). Clear examples like
that of the French future above are not easy to come upon. Based on
historical data from Iranian languages, which are relatively well docu-
mented, Haig (: ) concludes that “inflectionalization is evidently
a process that requires millennia, not centuries, to achieve.” That is,
historically observing an entire process of inflectionalization from
lexical item to inflectional ending, and then even its repetition starting
out from an analytic structure would require a time-depth of data that
extant historical records of languages do not afford us. Thus, Haig
writes that “the assumed final stage of grammaticalization, namely
into full-fledged inflection, is an exceedingly slow process indeed,
taking millennia before all traces of the lexical, or at least non-
inflectional, origins of grammatical formatives are lost” (Haig :
). This may be one reason why even in Indo-European languages,
which have been the cornerstone of grammaticalization research, only
few inflectionalizations have been observed historically.
It goes without saying that the situation becomes even more difficult
when historical records are not available. For example, with respect to
Australian languages, Mushin () points out that, “[f]ew grammatical
categories are regularly marked by forms whose lexical source is still
available as a free form. . . . It is therefore challenging to find clear com-
parative evidence of contemporary bound affixal forms that in some
languages may retain features of their lexical origins.” Often, the best that
comparative evidence can provide is to identify different stages in the
process of grammaticalization of the same categories in related lan-
guages, that is, reconstruction through intragenetic comparison.
The slow pace of the genesis of inflections contrasts with a poten-
tially rapid pace of their decay. With respect to the Iranian languages
mentioned above, Haig () suggests that, “[i]nflectionalization is
evidently a process that requires millennia, not centuries, to achieve,
though paradoxically, its loss can be quite rapid, even catastrophic.”
Profound changes in the typological profile of a language are more
likely to be brought about by intense language contact than by the
primarily language-internal changes that are at the core of the idea of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONCLUSION
an anasynthetic spiral. But under conditions of language contact as
well, grammaticalization plays an important role, as seen in Chapter .
So, overall, the traditional analytic–synthetic cycle or spiral in the
development of languages, or at least of specific language categories,
is due to loss on the one hand, which can be very rapid, and grammat-
icalization on the other hand, which can take a long time, millennia in
the case of inflections. Intense language contact, as for example in
pidgin and creole languages, can accelerate grammaticalization, and
potentially also the formation of ‘synthetic’ structures (cf. McWhorter
).
8.3 Conclusion
This chapter has dealt with cases where typological features of language
apparently influence grammaticalization paths on the one hand, and
cases in which grammaticalization plays a crucial role in creating
typological features on the other. We have seen that variation in
grammaticalization due to the grammatical features of a language is
most obvious with respect to phonological and morphological aspects
of grammaticalization, but can also extend to syntax and semantics.
The most well-known case in which typological features of a language
constrain morphological and phonological grammaticalization is the
tendency of isolating languages not to develop affixal material and
grammatical paradigms (Bisang , , ; Ansaldo et al.
). However, the extent to which grammaticalization in these lan-
guages differs from that found in other languages may not be as great as
is sometimes thought (cf. Dahl ). On the semantic and syntactic
side, it seems that there is generally a tendency in languages to follow
established grammaticalization paths and renew or elaborate on already
established grammatical categories rather than to create entirely new
structures and categories. A salient departure from this tendency towards
conservatism is most likely to take place under intense language contact,
as has been demonstrated clearly in Aikhenvald’s () analysis of
Amazonian languages.
As for the case of grammaticalization creating language structures of
typological relevance, it has been found that the order of grammatical
morphemes can be explained by their position at the time of their
grammaticalization. This seems to hold for morphologically bound as
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
GRAMMATICALIZATION AND LANGUAGE TYPOLOGY
well as non-bound morphemes. Second, commonalities and diver-
gences in the coding of grammatical categories across languages seem
to be motivated to a large degree by grammaticalization. Thus, as Bybee
(: ) puts it, “grammaticalization has great potential for explain-
ing the similarities as well as the differences among languages.” A third
area where grammaticalization strongly contributes to typological fea-
tures of languages is the cycle—or spiral—between analytic and syn-
thetic language structures as it clearly takes place in the life of specific
categories in specific languages, but perhaps even at the level of overall
structure of a specific language. This cycle/spiral feeds on the mechan-
isms of grammaticalization. A complicating factor in documenting full
cycles or spirals is that while the decay or disappearance of extant
morphological marking can be very quick, its development may take
a very long time, especially when it comes to full-fledged morphologi-
zation such as inflectionalization. Lastly, grammaticalization is an
important player when languages develop new categories or a new
typological profile under intense language contact. There is a strong
tendency for languages in contact to adopt features from each other
and thus develop a similar typological profile.
Further reading
For modern introductions into language typology featuring both grammat-
icalization and morphological structures, see Whaley (), Song (),
Veluppilai (), and Moravcsik ().
For thoughts on the relationship between reconstruction and typology, see
Comrie (, ). Features of grammaticalization in specific areal and
genetically defined language groups are discussed in Part V of Narrog and
Heine () and in Narrog and Heine () overall. For an example of a
whole morphosyntactic cycle leading from synthetic via analytic to polysyn-
thetic structures in the development from Early Egyptian to Coptic, see
Hodge ().
Discussion points
() What do you think are the typological features of Modern English or
another language that you know well that may ‘guide’ future
grammaticalizations?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCUSSION POINTS
() What categories would you consider ‘florescent’ in Modern English or
another language that you know well?
() What do you think are the potential advantages and disadvantages in
speaking of a ‘cycle’ versus a ‘spiral’ in grammaticalization? What could be
a problem when applied to languages as a whole?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
9
Theory-specific approaches
9.1 Generative grammar
9.2 Functional Discourse Grammar
9.3 Variationist Theory (sociolinguistics)
9.4 Cognitive Grammar
9.5 Construction Grammar
Discussion points
The majority of research in grammaticalization is not strongly tied to a
specific theoretical framework. Nevertheless some significant influences
can be identified, as we saw in previous chapters. These are, especially,
influences from functional linguistics, cognitive linguistics, descriptive
linguistics, pragmatics, usage-based grammar, study of language proces-
sing, and linguistic typology. Individual researchers usually approached
grammaticalization based on one or more of these frameworks rather
than the others, leading to sometimes considerable differences in their
view of grammaticalization. Still, these generally-speaking cognitive-
functional backgrounds were typically compatible with each other, leav-
ing more common ground than fundamental differences. This common
ground for example includes the notion that it is often useful to explain
synchronic language states with reference to diachrony, or that diachrony
is even implicit in synchrony, and that semantic-pragmatic rather than
formal factors are pivotal in grammaticalization.
In this chapter, we will briefly discuss approaches to grammaticali-
zation in which the framework within which grammaticalization is
studied leaves a stronger imprint on the view of grammaticalization
than is commonly the case. In some approaches, there is still plenty of
common ground with the “main stream” of grammaticalization studies,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GENERATIVE GRAMMAR
but in other approaches the contrast is more pronounced and reaching
a common ground requires more negotiation.
We start by discussing grammaticalization in generative grammar
(Section .), proceed to grammaticalization in Functional-Discourse
Grammar (Section .), grammaticalization in variationist studies
(Section .), grammaticalization and Cognitive Grammar (Section .),
and finally to the relationship of grammaticalization to Construction
Grammar and constructionalization (Section .).
9.1 Generative grammar
Generative grammar is based on a linguistic theory that is outwardly in
stark contrast to the cognitive-functional background of most grammat-
icalization studies. First, the ultimate object of study is the internal
language system of the individual and its acquisition. This results in a
strict separation of synchrony and diachrony, and a focus on synchronic
states of language. Diachronic aspects have generally been relegated to the
periphery of the study of language, and language change was often
understood as basically random and without directionality (cf. van
Gelderen c: ). Moreover, syntax is the core object of study, and
semantics-pragmatics are regarded as secondary. Furthermore, generative
grammar has no reason to be interested in cognitive, functional, or
discourse factors as an explanation for change since the syntactic struc-
ture in a specific model of syntax itself is already taken as an explanation.
Nevertheless, there are some striking and obvious overlaps between
empirical observations in grammaticalization studies and structural de-
scriptions in generative grammar, which eventually led to a rise in interest
in grammaticalization, enhanced by the rise of minimalism. Among them
is the idea of change from lexical to grammatical, and the possibility to
generalize across languages, especially in terms of unidirectionality. As
described by van Gelderen (c), pioneering work in the early s
(e.g. Abraham , ; Roberts ; van Gelderen ) was fol-
lowed by a notable surge in research activities in the s.
The distinctive feature of generative grammar vs. some other
syntactic frameworks is that it centrally posits hierarchical, that is,
non-relational, constituent structures in the clause. That is, sentence
structure is analyzed as hierarchical tree structure of lexical and func-
tional phrases. The higher nodes in the tree are occupied by functional
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
phrases built around functional heads, while the lower nodes are
occupied by lexical phrases built around lexical heads. The general
consensus description of grammaticalization in generative grammar,
then, is essentially a ‘climbing up’ the tree by a category that was
initially merged lower in the tree structure to a higher position. That
is, a lexical category can become the realization of a functional (gram-
matical) category, or a functional category located lower in the tree can
become the realization of a functional category located higher in the
tree. In the former case, one can speak of the “creation of new func-
tional material” (Roberts and Roussou : ). Note that when a word
or morpheme climbs up a tree it also means it acquires a wider scope.
This is at first glance a contrast to Lehmann’s ( []) idea of
reduction of ‘structural scope’ as a parameter of grammaticalization
(cf. Section .). However, as we saw in Sections . and ., ‘structural
scope’ refers to an entirely different concept of morphosyntactic struc-
ture, and would be better labeled ‘structural size.’
In order to eventually ‘climb,’ a linguistic item merged lower in the
tree must have a feature that was checked at a position higher in the
tree. That is, this item originally instantiated features at two or more
positions in the tree structure, and underwent Move in order to check a
feature at a higher position. If reanalysis (or grammaticalization) takes
place, the same feature is merged directly in a higher position. This
observation has been formulated as ‘structural simplification’ by
Roberts and Roussou (), and as ‘economy’ by van Gelderen
(). Van Gelderen () distinguished two Economy Principles,
namely () the Spec to Head Principle (Head Preference Principle), and
(), the Late Merge Principle. These are motivated by the theory-internal
assumptions that () checking between two heads is more economical
than checking between a specifier and a head, and () merging is more
economical than moving, since merging is required anyway, but mov-
ing comes at an additional cost. With respect to ‘structural simplifica-
tion,’ the structure resulting from grammaticalization is always simpler
than the erstwhile structure if the erstwhile structure implied feature
syncretism (i.e. realization of two or more features in one item).
The (secondary) grammaticalization of the numeral classifier ge as an
indefinite determiner in colloquial Mandarin Chinese (Wu ;
Chapter ) may serve as an illustration. Example () shows ge as a
numeral classifier, attached to a numeral. Example () represents the
outcome of this change, namely ge as an unambiguous indefinite
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GENERATIVE GRAMMAR
determiner. In this position it cannot be quantified by a numeral. Also
note the co-ocurrence with a different classifier ping that is occupying the
kind of position that originally ge had.
() Mandarin Chinese (Wu : )
Wo you liang-ge pinguguo.
I have two- apple
‘I have two apples.’
() Mandarin Chinese (Wu : )
Ta he-le ge san-ping jiu.
he drink- three- wine
‘He did a drinking of three bottles of wine.’
According to Wu (: –), ge as in () occupies the position in the
DP as indicated in the structural tree in (). That is, originally, ge is
merged as the head of a classifier phrase in ⁰.¹ Ge as an indefinite
determiner in () occupies the position as indicated in (). The realization
of D in D⁰ cannot be the result of Move since ⁰ is realized as ping
‘bottle’. () represents an intermediate stage of change at which ge was still
base-generated in the P and had to move to D⁰ position via Num⁰.
() DP
D′
D0 NumP
Ø Num′
Num0 ClP
(yi)/er/san Cl′
‘(one)/two/three’
Cl0 NP
ge N0
¹ Note that this analysis presupposes that Mandarin Chinese already has a DP structure,
which is not uncontroversial.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
() DP
D′
D0 NumP
Ø Num′
Num0 ClP
(yi ‘one’) Cl′
Cl0 NP
ge N0
() DP
D′
D0 NumP
ge Num′
Num0 ClP
Cl′
Cl0 NP
N0
While the description of the basic mechanism of grammaticalization as
upward reanalysis in a hierarchical clause structure is uncontroversial
within the framework, the way it is theoretically framed and motivated
differs to some extent from researcher to researcher. Roberts (Roberts
and Rousseau ; Roberts , and elsewhere) emphasizes that
grammaticalization is parameter change, that is, change in the features
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. GENERATIVE GRAMMAR
associated with a specific lexical or functional item in the lexicon. Van
Gelderen () emphasizes that grammaticalization is associated with
language acquisition. Innate principles of economy cause the language
learner to reanalyse items in the lexicon.
More recently, Van Gelderen (c: –) has described grammat-
icalization in terms of feature economy: Since Chomsky (), fea-
tures in generative grammar are distinguished into interpretable and
uninterpretable ones. Interpretable features remain visible after check-
ing and cannot be deleted, while uninterpretable features can only be
checked once (cf. Van Gelderen : –). Now, it is more eco-
nomical for an item to have uninterpretable features, so items develop
towards having uninterpretable features.
For example, the French negation particle pas was initially a noun
with the meaning ‘step’ and generated (merged) as a complement in the
VP. When used for negation together with the negation particle ne it
moved to the Spec of NegP and was reanalyzed as being generated
(merged) there. But it still had an interpretable feature. Finally it was
reanalyzed as being generated (merged) in the Head of NegP. At this
point, negation became an uninterpretable feature.
Further reading
Van Gelderen (c) contains a list of prominent publications on gram-
maticalization by generativists up to the point of publication. Additionally, in
an earlier attempt to explain language change in terms of generative gram-
mar, Lightfoot () applied the framework of principles and parameters to
it, claiming that language learners initiate change by selecting specific struc-
tures from a genetically encoded finite set of language structures. Change
in this framework is viewed as abrupt and catastrophic. See Hopper and
Traugott (: –) for a critique. Haspelmath () was an early attempt
to point out commonalities between grammaticalization theory and genera-
tive grammar. Fischer and Rosenbach () offer a discussion of formal vs.
functional approaches to grammaticalization. Osawa has argued in a number
of publications (e.g. Osawa , ) that grammaticalization is an instan-
tiation of category maturation, that is, an emergence of functional categories
heading their own projections. Van Gelderen’s more recent research on
grammaticalization has been focused on cyclicity in grammaticalization
(e.g. Van Gelderen a, ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
9.2 Functional Discourse Grammar
Hierarchical clause structures are not an exclusive domain of formal
grammars. Functional Discourse Grammar (FDG; formerly ‘Functional
Grammar’, FG) distinguishes four levels of grammar, the phonological
level, the morphosyntactic level, the representational level, and the
interpersonal level, each of which is in turn organized into layers.
Each layer maximally has the following structure:
() π v: [head (v)Φ]: [σ(v)Φ])Φ
V is the variable of the relevant layer, which is restricted by a head and
(optionally) further by a modifier σ. Head and modifier both take the
variable as their argument. Finally, π (also optional) is an operator
specifying the layer and Φ a function relating the layer to other units
(Hengeveld and Mackenzie : ). Central to grammaticalization
is the Representational Level, which consists of the hierarchical arrange-
ment and the realization of semantic categories. This is a purely semantic
level which corresponds to syntactic realizations via encoding rules. The
layered structure of the Representational Level is given in () below in
linear order (as an alternative to a tree structure representation):
() The Representational Level (Hengeveld and Mackenzie : )
(p1: [(ep1: [(e1: [(f1: [(f2)n (x)Φ . . . (x+n)Φ] (f)) . . . (f+n)(e)Φ])
. . . (e+n)[Φ]] (ep)) . . . (ep+n) [Φ]] (p))
The outermost layer is the proposition (p), which recursively contains
episodes (ep), which in turn recursively contain States-of-Affairs (e). These
contain Configurational Properties (f) that can further contain descrip-
tions of Individuals (x) and Lexical Properties (f) (Hengeveld and Mack-
enzie : , , ). Also, as mentioned in (), every layer can have one
or more operators, that is, exponents of grammatical categories specifying the
layer. In contradistinction to recent models of generative grammar, there is
no hierarchical relationship between elements of the same layer.
Now, within this model, Hengeveld claimed as early as that
“diachronic developments in the field of operators tend to follow the
direction [from lower layer to higher layer]”² (Hengeveld : ).
² According to the model of Functional Grammar of that time, this was represented
as “π > π > π > π,” whereby π = predicate operator; π = predication operator;
π = proposition operator; π = illocution operator.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. FUNCTIONAL DISCOURSE GRAMMAR
To our knowledge, this is the first time that a mechanism of grammat-
icalization was formulated within a specific theory of syntax. Note that
the direction of change posited, and empirically observed, is the same as
in generative grammar (Section .), namely towards a higher layer and
towards wider scope. As formulated in FDG, grammaticalization in-
volves an increase in unidirectional scope at both the representational
and the interpersonal level (cf. Hengeveld : –). Hengeveld
(: –) furthermore suggests that grammaticalization, and thus
category climbing across layers, may also occur on the interpersonal
level, and from the representational to the interpersonal level. Accord-
ingly, there is an interesting third kind of scope increase in this
model, namely when a marker jumps from the representational to the
interpersonal level in a process that does not require gradual grammat-
icalization (Hengeveld : ). In the terminology of the present
book, this process is treated as ‘cooptation’ (Chapter ).
Since FDG is also mainly concerned with synchronic issues, research
following up on the observation in Hengeveld () has not been
copious but includes Boland (), Hengeveld (), and Hengeveld
et al. (), all primarily dealing with change in the area of tense,
aspect, and modality (TAM). These are categories for which FDG has
traditionally made fine distinctions (cf. Dik a). Different types of
tense, aspect, and modality are located at different layers of the clause.
For example, phasal aspect is located at the layer of Configurational
Properties (f; also labeled as ‘Situational Concept’ layer), event quan-
tification and relative tense at the State-of-Affairs layer (e), and abso-
lute tense at the Episode layer (ep). The predicted direction of change in
the tense/aspect domain is then from phasal aspect to event quantifi-
cation and relative tense to absolute tense, that is from narrower to
wider scope. This is what has been actually found historically in a
number of languages. Hengeveld (: –) provides the example
of the Spanish haber ‘have’ + past participle construction. This
construction underwent a development in three stages that can be
schematically represented as in ().
()
Stage I Stage II Stage III
category resultative perfect past
TA distinction phasal aspect relative tense absolute tense
layer f e ep
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
More generally, Hengeveld (: ) posits the pathways of grammat-
icalization represented in (). The units of the representational level
have already been discussed, except that “f” has been renamed as “f c,”
that is, ‘configurational property’ in . The abbreviations at the
Interpersonal Level are M = move, A = discourse act, C = commu-
nicated content, R = referential act, and T = ascriptive act. The bold
letters indicate the path of haber in Pensinsular Spanish from f¹ (fc) to
ep as sketched out in ().
()
Lexicon Lex Lex Lex Lex Lex
Interpersonal level M ← A ← C ← R ← T
Representational level p ← ep ← e ← fc ← f
Lexicon
Lex Lex Lex Lex Lex
Hengeveld (: ) succinctly describes the meaning of the repre-
sentation in () as follows:
Lexical items may enter the system at any point, but once this point has been selected
they cannot move down to a lower point on the interpersonal or representational scale.
Items can move up from the representational level to the interpersonal level at any
point, but once they have entered the interpersonal level they cannot move down the
interpersonal scale. (Hengeveld : )
With respect to the categorical change of items in grammaticalization,
Hengeveld (), based on Keizer (), proposes a change from
lexeme (modifiable, can be focused) to lexical operator (not modifi-
able, but can be focused) to operator (not modifiable, cannot be
focused). Contentive change does not have to be paralleled by cate-
gorical change.
Further reading
Hengeveld et al. () contains a number of case studies and theoretical
discussions of grammaticalization within this framework.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. VARIATIONIST THEORY (SOCIOLINGUISTICS)
9.3 Variationist Theory (sociolinguistics)
The topic of researching grammaticalization in sociolinguistics differs
from the other topics in this chapter in that the grammatical analysis of
the phenomena in question is not affected, as it is affected when
choosing a generative framework or an FDG framework, for example.
Sociolinguistics also shares with the main stream of grammaticalization
studies an emphasis on the study of language use. However, the soci-
olinguistic study of grammaticalization highlights different factors of
change compared to grammatically motivated research.
When speaking of the “sociolinguistic study of grammaticalization”
we do not intend to conceal the fact that this approach is not yet very
well established. This may primarily be due to the fact that historically
oriented sociolinguistics has been overwhelmingly concerned with
phonetic and phonological change, but also to the fact that important
proponents of grammaticalization theory did not have sociolinguistic
aspects of change on their agenda. Nevertheless, sociolinguistics offers a
complementary perspective on grammaticalization that deserves more
attention than it has so far received.
According to Nevalainen and Palander-Collin (), from the
perspective of a sociolinguistic approach to language change, grammat-
icalization is no different from other types of change. A primary focus
for historical sociolinguistics is the diffusion of change in a specific
community of speakers. Diffusion usually takes the shape of an
S-Curve, that is, starting slowly, then accelerating, and finally plateau-
ing off when the new linguistic convention is established. A change is
considered as completed when reaching a frequency of percent of
the competing variant forms. Some changes may take several genera-
tions, and the progress can be studied on a contemporary community
of speakers by slicing the community into generations. In other cases, a
whole community of speakers may adopt the change across genera-
tions. Factors that are of particular interest include the age of the
speaker, their gender, and their social rank and professions in correla-
tion to their adoption of the new linguistic convention. For example,
Nevalainen () in a study of the spread of the adverbial suffix -ly on
intensifying adjectives in adverbial function (e.g. extreme/extremely,
exceeding/exceedingly) in fifteenth- to eighteenth-century English,
showed that -ly adverbialization accelerated in literate social ranks,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
and the eventual social stigmatization of the bare forms through social
evaluation led to broad adoption of the marked adverbialized forms.
‘Variants,’ such as extremely vs. extreme in the example above, are
central elements within this concept of change. The variant is also at the
center of variation theory, an approach within sociolinguistics that has
particular affinity to historical change in general and grammaticaliza-
tion in particular. According to Poplack (: ), key concepts of
variation theory are:
(i) the linguistic variable, that is, different ways of saying the same
thing;
(ii) the variable context, that is, the specific domain in which
variants alternate without change in representational meaning;
(iii) the principle of accountable reporting, which means to cover
not only cases in which the forms of interest materialized but
also cases where they did not although they might have.
For variation theory, characteristic for any language change, including
grammaticalization, is that a specific variant extends its use in a certain
context while another variant retreats. In traditional grammaticaliza-
tion theory, this kind of variation has been known under the label of
‘layering’ (cf. Section .). Nevertheless, it is rarely studied systemati-
cally, because the emphasis tends to be just on the element (or the
‘variant’) that successfully grammaticalized at the expense of the others.
In studies based on variation theory, layering takes center stage. Fur-
thermore, detailed quantitative analysis provides evidence with respect
to which factor significantly contributes to the choice of one variant
over others. Constraints on variant choice are ranked, and this ranking
is understood as the ‘grammar’ determining surface realization of the
variable (Poplack : ).
As an example, be going to has grammaticalized as a future variant in
English, but in Present Day English it is still competing with will, with
which it now shares the variable context of expressing future. While the
choice between be going to and will has been traditionally attributed to
semantic nuances such as temporal proximity, certainty and willing-
ness, a multivariate analysis of the factors contributing to variant choice
reveals a slightly different picture. Namely, the strongest effects on
variant choice can be attributed to specific collocations (cf. Torres
Cacoullos and Walker , ). For example, will has a very low
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. VARIATIONIST THEORY (SOCIOLINGUISTICS)
likelihood of occurring vs. be going to in questions, and with second
person subjects. Both factors are combined in example ().
() a. What are you gonna/going to do? (likely)
b. What will you do? (less likely)
On the other hand, will has carved out a niche in collocation with never,
with which it is much more likely to co-occur than be going to, and in
fixed discourse formulas as in ():
() I’ll tell . . . / I’ll pay . . . / I’ll ask . . . / I’ll call . . .
In contrast, be going to is much more likely to occur in the complement
of cognition verbs with a first person subject, such as I think . . . Overall,
then, it is clear that both will and be going to are very much entrenched in
specific contexts, which explains the equilibrium between the two var-
iants, instead of the relatively newer be going to wiping out the older will.
Sticking with the concept of future, Poplack and Malvar () and
Poplack () analyze the rise of the Portuguese periphrastic future
based on the verb ir ‘go’ at the expense of the synthetic future and the
haver + infinitive construction. () shows each form:
() Portuguese future forms
a. cantarei ‘I will sing’ (synthetic future)
b. hei de cantar ‘I will sing’ (haver-periphrasis)
c. vou cantar ‘I will sing’ (ir-periphrasis)
According to the authors, unlike English be going to and French and
Spanish go-futures, the Portuguese ir-periphrasis practically obliterated
the competing variants (note, though, that the present tense can still be
used with future reference). It did so by entering through proximate
future contexts. Furthermore, a multivariate analysis of the factors
contributing to the choice of future variant shows that in each language,
specific variants are preferred in different contexts, thus contributing to
the spread or non-spread of the new future form. For example, while
the Portuguese ir-periphrasis became entrenched in proximal future
contexts at an early stage, this was not the case with the French aller-
periphrasis. The upshot of this is, while in grammaticalization studies it
is often emphasized that expressions for certain categories cross-
linguistically have the same etymological sources, the actual process
of change and the end state in relation to other variants of the category
can be quite different. Specific functions and meanings of the lexical
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
item may remain as constraints in specific languages to varying degrees,
and specific forms may remain, as they are strongly associated with
specific contexts. Variation theory offers the tools to bring different
processes of grammaticalization and factors involved in them to light.
Further reading
Other relevant papers by the authors already cited above include Torres
Cacoullos (), Poplack and Torres Cacoullos (), and Poplack et al.
(). Poplack () contains a list of prominent contributions to the study
of grammaticalization in a variationist framework up to the point of publi-
cation. There is a strong connection between Variationist Theory and corpus
linguistics (see Section ..; Mair ), since variationist studies of
grammaticalization rely on corpora, on the one hand, and variationist meth-
odology is an obvious way to make use of historical corpora, on the other.
Studying grammaticalization with the help of historical corpora in the con-
text of register variation (e.g. Biber ; Biber and Gray ) is another
useful application of sociolinguistic methods to grammaticalization and
language change in general. Variation of grammaticalization in varieties of
English (e.g. Kortmann and Schneider ) (or other languages) is yet
another type of variation that can be studied with corpora.
9.4 Cognitive Grammar
Cognitive Grammar may not be a major grammatical framework
currently, and the relationship between grammaticalization and cogni-
tion has already been reflected throughout this book without referring
to this specific framework. Nevertheless, the Cognitive Grammar
framework offers a unique view of the relationship between cognition
and grammar, and the analysis of grammaticalization in this frame-
work warrants some attention because it is unique but nevertheless ties
in well with the overall directionality of semantic change in grammat-
icalization that we have identified in Chapter .
For Langacker (, , , ), grammaticalization is a
semantically-driven process and the result of schematicization and
especially subjectification, partially corresponding to the concepts of
abstraction and part of discourse orientation in Sections . and . of
this book, respectively. In Cognitive Grammar, linguistic expressions
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. COGNITIVE GRAMMAR
are viewed in terms of construals involving a conceptualizer (speaker)
and an object of conceptualization (an event). A construal is conceived
of as subjective if the conceptualizer’s perspective is reflected but not
explicitly put ‘on-stage’ in a linguistic expression. Subjectification, then,
has been characterized as “the realignment of some relationship from
the objective to the subjective axis” (Langacker : ), but later
re-defined as “a gradual process of progressive attenuation,” in which
“an objective relationship fades away, leaving behind a subjective
relationship that was originally immanent in it” (Langacker : f ).
By means of illustration, Figure . represents the gradual subjecti-
fication, and grammaticalization of the lexical verb go to the future
marker be going to/gonna in English.
(a) (b) (c)
objective scene space (source) objective scene
mental mental
tr
tr tr
objective scene
space
t
tr tr
mental (target) T
t C
t
T T
C C
(d) (e) (f )
objective scene objective scene objective scene
mental
tr tr tr
t t t
T T T
C C C
Figure 9.1 Grammaticalization of be going to in Cognitive Grammar (Langacker : )
Diagram (a) in Figure . represents the starting point of the devel-
opment and (f) the end point. “C” is the conceptualizer, that is, the
locus of conceptualization and language production. The object of
conception is circled by the dashed line. The part of the event that is
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
profiled by the linguistic expression is highlighted with bold lines.
The circle marked as “tr” is the ‘trajector,’ the focused particiant,
“t” is the conceived time, that is, the time of situation as it is described
while “T” refers to the ‘processing time,’ that is, time as the medium of
conception. At the starting point (a), which may be a sentence like Sean
is going to New Hampshire, the trajectory moves in physical motion
across space from an unspecified current location to the profiled
destination (New Hampshire). This spatial movement is mentally
tracked by the conceptualizer. Beyond the physical motion, the con-
struction also refers to a ‘mental’ relationship of intention between the
subject and the event, which is indicated by the double-dashed arrow in
the upper square of the diagram.
Starting from the source construction, through (b) to (f), step-wise
the more concrete and physical aspects of the event fade from the
construal. In (b) and (c), the spatial movement is fading, and what
remains is the mental aspect, that is, the intention of the trajector,
the subject to perform the event, as in the sentence Sean is going
to graduate. The event could entail physical spatial movement
(diagram b) or not (diagram c). In a next step, represented in diagram
(d), the intention can be removed from the trajectory and attributed to
some external force, that is the conceptualizer herself, or a third party,
as in the sentence, Sean is going to graduate—his mother is determined.
In diagram (e), even the mental aspect of the event, that is, some force
of intention is removed, and the projection is purely temporal.
A sentence such as Sean is going to graduate on Sunday, I hear would
be a possible example. The last stage of development depicted in
diagram (f) shifts the profile to the projected event per se. According
to Langacker (: ), this is a stage not yet reached by English be
going to/gonna.
In this way, the grammaticalization of be going to is characterized by
conceptual ‘attenuation.’ Through this process, the meaning shifts from
a more concrete one to a ‘schematic’ one devoid of details. Therefore,
‘schematicization’ takes place. Langacker has also labeled the concep-
tual attenuation as a kind of semantic ‘bleaching’ (cf. Section .;
Langacker : , : ). The lexical sources of grammatical
categories, such as go in this example are often frequent, basic, and
general. Langacker () calls them ‘conceptual archetypes.’ Finally,
the grammaticalized item is ‘transparent.’ While the lexical verb go
requires an animate, volitional subject, the future marker be going to/
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONSTRUCTION GRAMMAR
gonna is compatible with any kind of subject, including expletives such
as it or there as in It’s gonna be great. Overall, for Langacker, gram-
maticalization is a reductive process leading to reduction and erosion at
several levels of grammar. Conceptually, it is characterized as a pro-
gression from specific content to mental operations applicable to any
content (Langacker : , ).
Further reading
Langacker’s approach to grammaticalization is not only unusual in being
largely identified with subjectification but also in that his own writing is
entirely from a synchronic perspective. Athanasiadou et al. () is a col-
lection of papers that assembles the most prominent authors following the
Langackerian perspective on subjectification and grammaticalization, namely
Frank Brisard, Tanja Mortelmans, Péter Pelyvás, and Satoshi Uehara. Narrog
(b) points out a number of problems occurring when the Langackerian
concepts are applied to actual historical change, and in Narrog (b) the
subjectivity and subjectification concepts of Langacker are compared to those
of other proponents of subjectification.
9.5 Construction Grammar
While the relationship between grammaticalization theory and gener-
ative grammar (Section .) is problematic from the outset because of
the starkly different theoretical backgrounds but actually reveals sur-
prising commonalities, just the opposite is the case with the relation-
ship between grammaticalization theory and Construction Grammar.
The theoretical background in cognitive-functionalist grammar is
essentially shared, but some of the basic tenets when it comes to
actually analyzing phenomena of grammaticalization seem almost
irreconcilable. Furthermore, constructionalization theory has not only
been suggested as a framework within which to study grammaticaliza-
tion but also as an alternative framing of the same processes that have
been traditionally known as grammaticalization. That is, processes
traditionally labeled or conceived of as grammaticalization, would
now be conceived of in terms of ‘constructionalization.’ In that case,
the relationship of grammaticalization to Construction Grammar
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
would strictly speaking not be a topic for the current chapter, but
instead for Chapter , where we discuss processes adjacent to gram-
maticalization, such as lexicalization and degrammaticalization.
Construction Grammar is a framework that analyzes language as the
inventory of symbolic combinations of form and meaning, that is, ‘con-
structions.’ This inventory consists of symbolic units of all sizes, from
clauses down to the equivalents of traditional morphemes. That is, both
lexical and grammatical units of language are essentially the same kinds
of ‘constructions.’ As Trousdale (a: ) states, constructions in this
framework are “form-meaning pairings at different levels of abstractness
and atomicity that are organized in a taxonomy; thus, there is no
substantial difference between the lexicon and the syntax.”
Furthermore, while constituent structure and hierarchical clause
structure are in principle compatible with Construction Grammar, in
contrast to most modern structuralist frameworks of grammar, hier-
archies have been used only for describing relationships within the
inventory of constructions, and not for describing clause structure. In
contrast to traditional descriptive grammars, the emphasis in Con-
struction Grammar has been on identifying and describing individual
constructions, especially of the kind that are difficult to capture in
traditional phrase structures, that is, idiomatic structures, and not on
systematically describing the grammar of any language. The ‘grammar’
of any language would in fact be a very large catalogue of constructions,
including what has traditionally been treated as the ‘lexicon.’
This has a number of ramifications for Construction Grammar’s
relationship to grammaticalization. First of all, while for grammatical-
ization, the distinction between lexical and grammatical, and less vs.
more grammatical, categories is definitional, the denial of such a
distinction is programmatic at least to some versions of Construction
Grammar. Thus, Noël (: ) submits that “grammaticalization, as
a change from lexical to grammatical, is not an issue in Construction
Grammar: construction grammatical units can by definition not
become more grammatical.”
Secondly, when Construction Grammar is applied to diachrony as
the study of ‘constructional change’ (Hilpert a) or ‘constructiona-
lization’ (Traugott and Trousdale ),³ its scope “goes beyond
³ According to Traugott (), constructionalization involves changes in both form and
meaning whereas constructional change involves changes in form or meaning but not in
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONSTRUCTION GRAMMAR
grammaticalization,” as Hilpert (a: ) puts it. Some phenomena
that are not typically in the scope of grammaticalization include:
(a) lexical change;
(b) any kind of word order change, while grammaticalization is
restricted to word order change involving substantive material;
(c) all the processes adjacent to grammaticalization to be discussed
in Chapters and .
It will depend on the specific concept of Construction Grammar and
constructionalization if grammaticalization and any of the changes in
(a) to (c) are classified as specific sub-categories of constructionaliza-
tion, or if simply all changes are lumped together as individual in-
stances of constructionalization.
For the purposes of this book, however, it is more interesting to
consider the question whether phenomena traditionally associated with
‘grammaticalization’ should at least partially remain so, or should
better be reconceptualized as ‘constructionalizations.’ Hüning and
Booij () believe that the latter is the case, and provide a few
examples of change that seem to be particularly suitable to illustrate
the claim. These include the Dutch and German nouns hoofd and
Haupt ‘head,’ respectively, in compounds, the German suffix -fähig
derived from the lexical adjective fähig ‘able,’ and the Geman prefix
stock-, derived from the noun Stock ‘stick.’ These lexemes and mor-
phemes are all involved in word formation, a tricky area for the study of
grammaticalization and lexicalization since it is variously associated
with one or the other.
To take stock- as an example (Hüning and Booij : –), it was
originally involved in noun + adjective compounds, where the com-
pound expressed the meaning ‘as [adjective] as a stick’; for example
stocksteif ‘as stiff as a stick.’ In the sixteenth century, formations with
stock- became suddenly very productive in German and extended
beyond the literal meaning ‘stick’ to a general intensifying meaning;
for example stockalt ‘very old’ or stockblind ‘very blind.’ Furthermore, a
both—a terminological distinction not made by Hilpert (a). Since all symbolic combina-
tions of form and meaning are constructions, one could argue that ʽconstructionalizationʼ
should more appropriately be called ʽre-constructionalizationʼ as it involves the transition from
one kind of construction to another rather than from non-construction to construction.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
subclass of compounds developed in which stock- was not only inten-
sifying but also pejorative in combination with mental attitudes, for
example stockkonservativ ‘conservative to the core.’ With respect to the
central element Stock/stock-, this development exhibits some important
characteristics of grammaticalization: (a) context extension of the erst-
while noun to a number of adjectives with which it could not associate
in this position; (b) desementicization of Stock ‘stick’ to an intensifier;
and (c) decategorialization of the noun to a prefix, as stock- in the
compounds does not exhibit any noun characteristics. However, later
the productivity abated, and in Modern German, only very occasionally
new compounds are formed with stock-. Synchronically in Modern
German, one would therefore be inclined to consider formations with
stock- as lexicalized.
For Hüning and Booij (: –) both the terms grammaticali-
zation and lexicalization are misleading. Instead, stock- has undergone
constructionalization across four constructional schemas:
() (a) General schema for endocentric compounds
< [[a]Xi + [b]Yj]Yk $ [kind of semj related to semi]k >
(b) Schema for comparative (elative) adjectives
< [[a]Ni + [b]Aj]Ak $ [as semj as semi / very semj]k >
e.g. schneeweiss ‘as white as snow,’ stocksteif ‘as stiff/rigid as
a stick’
(c) Schema for elative compounds with stock-
< [[stock] + [b]Aj]Ak $ [very semj]k >
e.g. stockalt ‘very old’
(d) Schema for adjectives with stock- denoting a human trait or
behavior
< [[stock] + [b]Aj]Ak $ [very/extremely/too semj // semj to
the core]k >
e.g. stockkonservativ ‘conservative to the core’
The existence of the schemas (a) and (b) preceded the formation of
compounds with stock-. That is, German already had productive pat-
terns for endocentric compounds in general and for compounds with
comparative adjectives in particular. First, the compounds with com-
parative meaning evolved in analogy to extant compounds, then the
compound schema with stock- developed a life of its own in (c), when
the pattern developed a general intensifying meaning. In a further step,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONSTRUCTION GRAMMAR
the narrower sub-schema in (d) developed that remained the only
productive formation pattern for some time.
Hüning and Booij (: –) suggest that terms like grammati-
calization and lexicalization would not appropriately describe the
development since they presuppose that an independent element
stock with intensifying meaning enters the compound. However,
there is no such independent element, and instead the intensifying
meaning is only obtained in certain contexts, namely with certain
adjectives. Scholars of grammaticalization and lexicalization might
immediately counter that context is an essential element for both
processes (cf. Chapter of this book, with respect to grammaticaliza-
tion). Also note that the concept of a construction as a unit for
grammaticalization was in no way alien to the study of grammaticali-
zation (cf. Noël : –). However, while the notion of ‘construc-
tion’ was flexibly used by many authors, it was not the fulcrum of the
analysis. In any case, the decisive claim for scholars of constructiona-
lization necessarily has to be that the concept of constructions captures
essential parts of change that are not captured by reference to context,
as applied in grammaticalization or lexicalization studies.
In contrast, scholars of grammaticalization have pointed out features of
grammaticalization that are not covered by a constructionalist approach
and that would suggest the framework of grammaticalization theory is in
fact more appropriate to study phenomena traditionally associated with
grammaticalization. These include (cf. Heine et al. ):
(i) The distinction between lexical and grammatical categories. Even
if this distinction may be gradual, there is plenty of evidence that
it is real and has manifold effects in grammar. Furthermore,
even morphologically dependent affixes can be shown to have a
meaning/function on their own, and not only in constructions.
(ii) It can be shown that semantic change accompanying shifts
from the lexical to the grammatical cannot be entirely reduced
to effects internal to constructions.
(iii) Grammaticalizations can be generalized typologically and
reveal universal processes in language change.
(iv) While constructionalization is explained with reference to
frequency of use, increasing schematicity and productivity of
constructions, and analogy, it can be questioned if any of these
factors is really a motivation for speakers to initiate change the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
way that the communicative goals or cognitive and pragmatic
factors adduced in grammaticalization studies are.
(v) Most of the criteria and parameters of grammaticalizaton, such
as the ones discussed in Chapter , are not immediately appli-
cable to constructionalization, and therefore constructionaliza-
tion might be less suitable for reconstruction.
While grammaticalization theory and Construction Grammar and its
concept of ‘constructionalization’ seem to be wide apart, there have also
been attempts for intermediate approaches, for example, treating change
in constructions within grammaticalization theory, or trying to retain
certain features of grammaticalization within Construction Grammar.
As for the first approach, Noël (: –) suggests that in order to
treat constructional change within a theory of grammaticalization, (a) the
construction must be grammatical, and (b) the construction must undergo
the type of change that is typical for grammaticalization. Himmelmann
(: ) attempts a definition of grammaticalization that includes both
the terms ‘construction’ and ‘grammaticizing element.’
Van Bogaert () may be a good example for this type of approach.
The author discusses the developing parenthetical use of complement-
taking mental predicates such as I think as “constructional grammatical-
ization.” According to her, decategorialization takes place, the whole
class of verbs undergoes the change, token frequency has risen signifi-
cantly, and thus the change displays the hallmarks of grammaticalization.
Major aspects of the grammaticalization of a construction include that
() the construction becomes more general and schematic/abstract,
() the construction becomes more productive, and () the elements of
the construction fuse and become intransparent (Van Bogaert (:
–). Similarly, Narrog (d) provides an analysis of the grammat-
icalization of the Japanese future marker -(a)m-/-(y)oo, which is com-
plemented by a description in terms of constructionalization. This
description sheds additional light on the change in the constructional
contexts in which the marker has been used.
As for the second approach, that is, analyzing grammaticalization
within Construction Grammar while to some extent retaining the
concept of grammaticalization, Gisborne and Patten (: ) sug-
gest that, in that case, Construction Grammar needs to allow for a
counterpart to the distinction between lexical and grammatical, or even
to the concept of less vs. more grammatical. This is possible if a shift
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONSTRUCTION GRAMMAR
from lexical to grammatical and to more grammatical is re-cast as a
shift from more substantive to more schematic constructions. Trous-
dale (a) is an attempt to provide a framework for this kind of
approach. For him, the general features of constructions relevant to
language change are (a) generality, that is, the schematicity of the
semantics of a construction, (b) productivity, that is, the ability to
sanction less schematic constructions, and (c) compositionality, that
is, the extent to which the meaning of a construction is derivable from
its components. The features of the change traditionally identified as
grammaticalization, then, are (a) an increase in generality, (b) an
increase in productivity, and (c) a decrease in compositionality of a
construction.
In conclusion, Construction Grammar is a rapidly evolving frame-
work of grammar that shares the same cognitive-functional theoretical
background with mainstream grammaticalization theory but has come
up with assumptions about grammar that are basically incompatible
with the notion of grammaticalization. There have been attempts both
to accommodate constructions within grammaticalization theory and
to integrate and somehow retain grammaticalization as a concept
within Construction Grammar, and it remains to be seen what further
development Construction Grammar will take. For the time being,
though, it seems appropriate to keep these two frameworks apart.
Further reading
Hilpert (a) and Traugott and Trousdale () are seminal mono-
graphic studies applying Construction Grammar to diachronic issues.
These three authors have also published a number of other important papers
on the topic (e.g. Traugott a, , b; Hilpert , b;
Trousdale , ; Traugott and Trousdale ). Other important con-
tributions include Rostila (), Bergs and Diewald (), Diewald (),
Bisang (), Fried (), de Smet (), Gisborne (), Patten (),
and Barðdal et al. ().
Börjars et al. () and Heine (a) are critical reviews of Traugott and
Trousdale (). Trousdale () and Heine et al. () are discussions of
the relationship between grammaticalization and constructionalization, the
former from the perspective of Construction Grammar and the latter from
the perspective of grammaticalization theory.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
THEORY-SPECIFIC APPROACHES
Discussion points
() Do you think that any of the frameworks presented here is (or are) difficult
to reconcile with the fundamental ideas about grammaticalization as
presented in first five chapters of this book?
() Which of the frameworks, in your opinion, will add the most value to the
approach to the study of grammaticalization presented there?
() Both the generative and the Functional Discourse Grammar framework
associate grammaticalization with expansion of scope, and development
towards ‘higher’ levels of grammars. What do you think might be the
motivation for this development from the perspective of the speaker?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
10
Beyond grammaticalization
10.1 Lexicalization
10.2 Degrammaticalization (anti-grammaticalization)
10.3 Exaptation/regrammaticalization
10.4 Other -izations
10.5 Conclusion
Discussion points
There are a number of diachronic processes besides grammaticalization
that bear some semblance to or have some relationship with grammat-
icalization. Many of them have been discussed quite controversially and
have sparked a fair amount of research literature. We will introduce
each of the most important of them in a section of their own: Lexica-
lization (Section .), anti- and de-grammaticalization (Section .),
and exaptation or regrammaticalization (Section .). The remaining
concepts receive a brief description in a section dedicated to other
“-izations” (Section .). “-Ization” is a deliberate new word forma-
tion, tongue in cheek, alluding to the fact that not every label necessar-
ily means a different process, but sometimes is just a matter of excessive
terminology. At the end of Section . the reader should know by
which process a new word like “-ization” is derived.
10.1 Lexicalization
To start with, lexicalization can be simply defined as a process by which
linguistic units enter the lexicon that previously had a different status,
for example, as phrases (cf. Wischer ; Lightfoot ). So, we may
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
in the first place think of syntactic combinations of two or more words
that become one lexical unit, such as Modern English tomorrow from
the Middle English preposition to ‘at, on’ and the noun morrow ‘morn-
ing,’ or similarly Modern German heute ‘today’ from Old High German
hiu tagu ‘this day,’ and secondarily of morphological combinations
between lexemes or lexemes and affixes that fuse into one word that
eventually becomes indistinguishable; for example infamous from
in- (negative prefix) and famous, which is still fairly transparent, or
Modern English near from the comparative form (adding the suffix -er)
of Old English nēah ‘nigh, near’ (cf. Brinton : ), which has
become entirely intransparent for contemporary speakers.
Dong (: –) presents a number of easy to understand ex-
amples of lexicalizations from the combination of words in Chinese
language history. Hòuhuǐ in Modern Chinese is a verb meaning ‘regret’.
However, originally it consisted of two words, the adverb hòu ‘later’,
and the verb huǐ ‘regret’, as in ():
() Ancient Chinese (Dong : )
Yíng suō zhuǎn huà, hòu jiāng huǐ zhī
advance retreat turn change later will regret it
‘Times change, and later you will regret it.’ (fifth century , Guoyu)
According to Dong (), the first lexicalized uses of hòuhuǐ can be
found in the seventh century, when the position of negation (bù) before
hòu and huǐ instead of in between them, as in (), indicates that they
have become a unit.
() Middle Chinese (Dong : )
Fū jī shì bù xiān, xiǎn bù hòuhǔi
important thing not first few not regret
‘If the important things are not prioritized, few will not regret it
(later)’. (seventh century , Jinshu)
Presumably the frequent collocation of ‘later’ and ‘regret’, as in (), led
to the lexicalization.
() Ancient Chinese (Dong : )
Zù ér hòu huī, yì wú jí yě
do and later regret also not reach
‘If you do [something] and regret [it] later, you will also not come
up.’ (> ‘It is too late to regret an act already done.’) (fifth century
, Zuozhuan)
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. LEXICALIZATION
The new lexical item was even borrowed into Japanese in the twelfth
century as the totally inseparable word kōkai, illustrated in ().
() Modern Japanese (NKD , : )
Kōkai~si-na.i zisei=no tukuri-kata
regret~do-. life= create-method
‘How to create a life without regret.’
In this example from Chinese, grammaticalization is clearly not
involved, and lexicalization looks like something totally unrelated.
However, there are other changes that seem to have properties of
both lexicalization and grammaticalization. They include the formation
of complex adpositions, and of complex predicates, as well as nominal
affixes in word formation.
Brinton (a), investigating the properties of composite predicates
in English, argues that some of them might be lexicalizations and others
grammaticalizations. Consider lose sight of as in () and take a look at
as in ().
() Don’t lose sight of who you are.
() Take a look at Vancouver’s changing skyline.
Are both expressions lexicalizations, or only one of them, or none of
them? In order to enable well-informed decisions about classifying
a change as ‘lexicalization’ vs. ‘grammaticalization,’ Brinton and
Traugott () proposed a more detailed definition of lexicalization,
as in (), and a list of features of lexicalization vs. grammaticalization as
in Table . (next page).
() Lexicalization: “the change whereby in certain linguistic contexts
speakers use a syntactic construction or word formation as a new
contentful form with formal and semantic properties that are not
completely derivable or predictable from the constituents of the
construction or the word formation pattern. Over time there may
be further loss of internal constituency and the item may become
more lexical.” (Brinton and Traugott : )
There is no space to explain and discuss each of the features, not all of
which are equally important. However, as can be seen, about half of the
features listed here are shared between the two processes. Among those
distinguishing them, Brinton and Traugott (: –) consider ‘pro-
ductivity’ to be the most important.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
Table 10.1 Features of lexicalization vs. grammaticalization (Brinton and Traugott : )
Lexicalization Grammaticalization
Gradualness + +
Unidirectionality + +
Fusion + +
Coalescence + +
Demotivation + +
Metaphorization/metonymization + +
Decategorialization +
Bleaching +
Subjectification +
Productivity +
Frequency +
Typological generality +
Now, Brinton (a) argues that lose sight of is a lexicalization,
because it is not part of a productive pattern. Only two other composite
predicates are formed on the same pattern: lose count (of), and lose
track (of). Also the meaning is non-compositional and idiomaticized
(interestingly, these concepts are not listed among the distinguishing
criteria by Brinton and Traugott ), since it usually means ‘forget
about’ or ‘not keep fresh in one’s mind,’ as in (). Furthermore, the
possibilities of collocation are rather limited. Lose has become fixed to
sight of and vice versa. In contrast, Brinton (a) considers take a
look at as a grammaticalization, since the verb take is a highly produc-
tive so-called light verb (a verb with reduced, generalized lexical mean-
ing) that can be used with many other nouns in the same pattern, and it
also seems to have an aspectual function, signaling a bounded activity.
Not everybody might agree with Brinton’s (a) analysis. Presup-
posing that productivity is the decisive criterion, the take a [noun] at
pattern is indeed more productive than the lose [noun] of pattern.
An exact number is not provided but let us assume that there are a
hundred instances of take a [noun] at pattern, that is, there are
a hundred different nouns with which take a [noun] at is used, versus
only three of the lose [noun] of pattern. This is indeed a much higher
frequency, but it is dwarfed by the productivity of markers or
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. LEXICALIZATION
constructions in the verb phrase that are generally considered as part of
grammar, for example the modals or semi-modals, which can be
applied to practically any verb in the lexicon. Likewise, the light verb
take has a very generalized meaning but it still has every property of a
lexical verb. Restrictions to the form of the verb and the noun in this
construction (e.g. ?A look has been taken at . . . , *He takes looks at . . . )
can be ascribed to a process of idiomatization that can indicate lexica-
lization as well. In short, one can also make a good case that take a
[noun] at is simply a frequent lexical pattern.
Another prominent take on lexicalization vs. grammaticalization is
by Lehmann (, ), for whom two criteria are distinctive:¹
() grammatical forms are ‘regular’ while lexical forms are ‘idiosyn-
cratic,’ and () grammatical forms are ‘analytic’ while lexical forms are
‘holistic.’ ‘Regular’ is close to Brinton and Traugott’s () ‘produc-
tive,’ but it is actually more complex, since for a form to be regular, we
would expect it to be not only formed productively, but also be pre-
dictable in its meaning and form, that is, the opposite of ‘idiosyncratic.’
‘Analytic’ refers to the fact that something is formed compositionally
and transparently. Lehmann’s criteria reveal a different perspective on
lexicalization. Brinton and Traugott’s () is primarily diachronic,
focusing on grammaticalization vs. lexicalization as a historical change,
while Lehmann’s starts from synchrony, characterizing a construction
as the result of either of the two processes.
Different perspectives can in fact have a large influence on how a
change or process is classified. Word formation, an in-between area
between the grammar and the lexicon, is also ambiguous when it comes
to its dynamic and diachronic aspect. English noun suffixes like -hood,
-ship, and -dom (cf. Trips ; Haselow ), or German counter-
parts like -tum, -heit, -e, and -t (Lehmann ; Lightfoot , )
are a case in point. Lightfoot () and Habermann () investi-
gated the German suffix -heit, cognate of, and used similar to English
-hood, derived from a West Germanic noun *haid denoting meanings
such as ‘way, appearance, characterstic, rank’. Lightfoot () con-
cluded that it displayed both features of grammaticalization and
lexicalization, and Habermann (: ) suggests that “[e]ven though
¹ For Lehmann, lexicalization and grammaticalization have in common that they are
reductive changes, leading to a loss of the autonomy of a linguistic sign and lower
complexity (cf. Chapter on criteria of grammaticalization).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
new word-formation elements enrich a language’s lexicon, the process of
lexicalization is inextricably connected with grammaticalization”.
We will look into the case of -heit in a little more detail. As a suffix,
-heit roughly means ‘characteristic of ’ or ‘quality of,’ although the
concrete meaning may vary from word to word. It became especially
productive in Middle High German. () shows a few nouns in Modern
German that are formed on -heit.
() Modern German
Kindheit ‘childhood’ (< ‘child’ + -heit), Gesundheit ‘health’
(< ‘healthy’ + -heit), Krankheit ‘illness’ (< ‘ill’ + -heit), Vergan-
genheit ‘past’ (< ‘passed’ + -heit), Feinheit ‘detail’ (< ‘fine’ + -heit),
Menschheit ‘humankind’ (< ‘human’ + -heit)
Taking Lehmann’s perspective with an emphasis on the resulting
construction, we can make a strong case for lexicalization. Essentially,
-heit contributes to the formation of lexical items, and does not realize a
grammatical category like tense or definiteness. It is not a regular
process, since nouns that are formed on -heit are synchronically already
determined lexically in Modern German. -heit can be applied to nouns,
adjectives, and verb participles but it is probably applied to less than
percent of the entire noun, adjective, and verb inventory. Spontaneous
formations are possible but in many cases are blocked by already
existent other formations (hart ‘hard’ > Härte ‘hardness’/*Hartheit)
or are semantically incompatible (verdammt ‘damned’ > ?Ver-
dammtheit ‘damnedness’). The formations are also only partially ana-
lytical, since some nouns have developed an idiosyncratic meaning that
cannot be derived solely from its parts. Arguably this is the case with
Feinheit, Kindheit, and Menschheit in (). Feinheit is simply ‘detail’
rather than the ‘property of being fine,’ Kindheit is not ‘the quality/
property of being a child’ but ‘the time during which a human is a child’
and Menschheit ‘humankind’ rather than ‘the quality/property of being
a human.’ Krankheit is not an exact counterpart of Gesundheit since the
latter has more abstract meanings. In fact, every instance of -heit above
is rendered differently in English, pointing to the fact that even in two
closely related languages like English and German the means of word
formation can vary widely and be largely idiosyncratic. Furthermore,
-heit’s regularity or analyticity have not increased historically, except
for the medieval times in which -heit was particularly productive.
However, -heit is still more regular and analytical than most other
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. LEXICALIZATION
nominal affixes (such as -e in Härte ‘hardness’ or -t in Flucht ‘flight/
escape’), making it more ‘grammatical’ than those.
From Brinton and Traugott’s () diachronic perspective, how-
ever, one can make a much better case in favor of grammaticalization.
Decisively, -heit has decategorialized from original noun to affix, has
adopted an abstract subjectified meaning of ‘quality/property of ’ and is
fairly productive and frequent, in comparison with many other nom-
inal suffixes. Note, though, that this productivity still bears no com-
parison with regular grammatical categories—articles as markers of
definiteness for example, can be applied to the vast majority of nouns
as can temporal, aspectual, or modal categories to verbs. Lastly, -heit
does not realize a specific grammatical category and thus lacks ‘typo-
logical generality.’ One could argue, though, that -heit is simply a
nominalizer and by that virtue realizes a grammatical category. Indeed,
Brinton and Traugott (: ) suggest that those nominal affixes that
do not change category are lexical while those that do are grammatical.
In the case of -heit this would lead to an odd split between -heit being a
lexical suffix when suffixed to nouns, as in Kindheit, and a grammatical
suffix when suffixed to adjectives as in Feinheit.
In any case, processes of word formation do not render themselves to
simple classifications in terms of lexicon vs. grammar. Different perspec-
tives will lead to different conclusions. Wischer (: ) poignantly says
that affixes as the above, although they “are not grammaticalized from a
synchronic point of view, diachronically . . . as long as they have their origin
in independent lexemes, have run through a process of grammaticalizaton.”
From yet another viewpoint, recall that we also cited fusions of
lexemes and affixes as lexicalizations at the beginning of this section,
for example infamous and near. The classification of lexicalization or
something else may differ when we view a change either from the
perspective of the newly formed word as a whole, or from the perspec-
tive of its components. Modern English near as a word is a lexicaliza-
tion of Old English nēah ‘nigh, near’ and the suffix -er. However, if we
focus on the comparative suffix -er, we are dealing with a ‘fossilization’
of the suffix rather than a ‘lexicalization.’² Similarly, if we focus on the
² The term ‘degrammaticalization’ has also been used for fossilizations in lexical items
(cf. Brinton : , citing scholars who take this stance), but this appears to be a
misnomer. Rather, loss of grammatical status and function is a fairly regular last stage on
traditional grammaticalization clines.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
new nouns formed on -heit, these are clearly lexicalizations. However,
focusing on the affix -heit itself, as seen above, we are able to make a
better case in favor of grammaticalization.
Finally, there are some phenomena in the lexicon that have been
treated under the heading ‘lexicalization’ but would not fall under the
concept as defined here. These include () the coinage of new words for
new concepts (e.g. for refrigerator or for socialism at the time these things
or concepts newly arose in society), () the conversion of one part of
speech to another, including of more to less grammatical parts of speech
(e.g. up from adverb to noun as in the ups and downs), and () the
extraction of parts of words to create a hyperonym as in ism extracted
socialism and stoicism, etc., a process known as hypostasis (cf. Wischer
: ). These are all abrupt and conscious processes of word forma-
tion, for which the label ‘lexicalization’ would only be appropriate if
‘lexicalization’ is defined and used in a broader sense than here. The
concept of ‘lexicalization’ has also been extended to account for the
development of discourse markers (see especially Wischer and
Fischer a, b: –). We will return to this issue in Section ...
Further reading
Brinton and Traugott (), which was cited above, is still the standard
resource for lexicalization in relation to grammaticalization. Himmelmann
() is a relatively frequently cited paper on the distinction of grammati-
calization and lexicalization. For Himmelmann, both are processes of con-
ventionalization, but in lexicalization a string of items is conventionalized
and in grammaticalization an expression pattern with one or more fixed
items. Davidse and De Wolf () likewise try to distinguish grammatical-
ization and lexicalization, and from a Hallidayan perspective, see ‘system-
icness’ as the decisive distinguishing factor. For them, it is not the process as
such, but the outcome of the process that should distinguish between gram-
maticalization and lexicalization. Sauer () presents a very broad picture
of possible meanings of lexicalization, especially with respect to other con-
cepts except grammaticalization, such as idiomatization, institutionalization,
etc. Beijering () discusses lexicalization vis-à-vis grammaticalization and
pragmaticalization, and Trousdale (b) tries to distinguish grammatical-
ization, lexicalization, and constructionalization from a constructional per-
spective. Lastly, Habermann () is a rich source for the development of
suffixes in word formation, providing food for thought on the distinction
between lexicalization and grammaticalization in this area.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DEGRAMMATICALIZATION (ANTI-GRAMMATICALIZATION)
10.2 Degrammaticalization (anti-grammaticalization)
Unidirectionality has been a central claim in grammaticalization theory
(cf. Section .). The centrality of this claim has served as an incentive to
find empirical counterexamples, that is, examples of actual change in the
opposite direction of grammaticalization. The most common label for the
opposite direction has been ‘degrammaticalization.’ Thus, a huge literature
on presumptive examples of degrammaticalization has developed over
time. However, most of these examples are not really the mirror image of
grammaticalization. We already mentioned the creation of hyperonyms,
such as ism from stoicism and communism etc., and word class conversion,
such as up in the ups and downs. As Haspelmath (: ) noted, “most
cases of ‘degrammaticalization’ that are cited in the literature do not show
the reversal of grammaticalization, but something else.”
Furthermore, degrammaticalization must not be confused with mere
‘retraction,’ that is, a development in which some linguistic form
develops some new grammatical use, but the grammatical use is even-
tually lost and only the original lexical, or the less grammatical use
remains. That is, the linguistic form under discussion ‘pulls back’ or
‘retreats’ in its overall range of meanings and functions.
At this point, it is helpful to get to learn about the notion of the
‘grammaticalization cline’ in order to better understand some of
the discussion on degrammaticalization. A typical grammaticalization
cline, that is, the path that a linguistic item eventually follows in
grammaticalization in terms of morphosyntax, is represented in ()
(cf. Norde : ).
() content item > grammatical word > clitic > inflectional affix (> Ø)
The idea is that a content item first becomes a grammatical word, then
a clitic, and then an inflectional affix before eventually ending up being
lost from the language. This is indeed a development that often takes
place in grammaticalization, although it will take so much time that in
most cases we can only observe part of that cline in historical data.
However, the problem with the cline is that it is morphology-centric
while the core of grammaticalization, as presented in this book, is
semantic/functional. This is why we have not presented this cline so
far, although it is such a common and sometimes also useful concept.
Additionally, with respect to morphology, the status of ‘clitics’ is
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
contested (cf. Börjars and Vincent : ). Nevertheless, we need to
introduce the cline here, because a large part of the literature on
degrammaticalization is based on developments claimed to be counter-
directional to it.
For Haspelmath (: –), true degrammaticalization would be
“a change that leads from the endpoint to the starting point of a
potential grammaticalization and also shows the same intermediate
changes.” To distinguish genuine counterexamples like these from the
deluge of alleged degrammaticalizations, he proposes the term ‘anti-
grammaticalization.’ For Willis (: ) and Börjars and Vincent
(: ), degrammaticalization should crucially display (i) the
reverse morphological shift from affix to grammatical word and then
content word, and (ii) semantic changes from grammatical/functional
to lexical. And both should (iii) take place as a gradual process and not
as a willful and abrupt intervention of the speaker, as in the case of the
creation of hyperonyms, for example.
Examples of ‘antigrammaticalization’ or genuine ‘degrammaticaliza-
tion’ in this constrained view are hard to come by. Haspelmath (:
) does acknowledge a total of eight cases from the whole literature up
to that point, but even most of those would seem to be primarily
degrammaticalizations in a morphological sense (i), but not in a func-
tional sense (ii). Askedal () acknowledges not a single of them as
a true example of degrammaticalization, and Börjars and Vincent
(: ) acknowledge only a single example, namely the change of
Old Church Slavonic indefinite pronoun něčǐto to Bulgarian nešto
‘thing.’ This lexeme did not undergo a large counterdirectional change
in terms of morphology, but it did change from grammatical word to
lexical word.
Another interesting candidate for a potentially genuine degramma-
ticalization is the change of Pennsylvania German wotte (Burridge
) from auxiliary to main verb. Wotte is the subjunctive preterite
form of the modal verb welle ‘will,’ comparable and etymologically
related to would as the preterite form of will in English. The modal
verbs in this American German variety are generally more grammatica-
lized than the Standard German verbs, but slightly less grammaticalized
than the English modals. Similar to their English counterparts, they are
subject to a number of constraints such as a defective inflection, lack of
a passive or imperative, and they can, with few exceptions, only be used
with infinite clause complements (German modal verbs also permit
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DEGRAMMATICALIZATION (ANTI-GRAMMATICALIZATION)
nouns and finite clause complements). We would therefore expect it to
be used in sentences such as ():
() Pennsylvania German (Burridge : )
*Ich wott kumme.
‘I wanted to come.’
However, instead it is only used as a lexical verb with the meaning
‘wish’ in sentences with a nominal complement like (), where it is
even in imperative form, or with a finite verb complement as in ().
() Pennsylvania German (Burridge : )
Wott mal hart fer sell.
Wotte once hard for that
‘Wish hard for that.’
() Pennsylvania German (Burridge : )
Er ist juscht am wotte, er kennt noch eens vun
He is just at wishing he could still one of
die Ebbel hawwe.
the apples have
‘He is just wishing he could have one more of the apples.’
In (), wotte is found as an infinitive in the progressive construction
am V-en, which would be impossible (imperative) or at least problem-
atic (am V-en) even for Standard German modal verbs, which are less
grammaticalized than the English modals. Burridge (: –)
suggests that the reason for the idiosyncratic development of wotte is
that two competing subjunctive forms, wette and wotte happened to
develop, and a functional differentiation ensued: while wette assumed
the functions of the regular grammatical form, wotte developed the
idiosyncratic lexical use through contexts of politeness, where a pret-
erite subjunctive of a verb of volition expressed a modest wish.
Börjars and Vincent (: ) are hesitant to acknowledge this
change as a genuine degrammaticalization because of the lack of
historical data for Pennsylvania German, and Haspelmath (: )
misses a well-established grammaticalization path in the opposite
direction. Both reasons for rejection seem harsh, since it is quite
common in grammaticalization research to acknowledge grammatica-
lizations even without accurate historical data, or if there is no other
example of exactly the same path. In the case of wollen, cognate of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
English will, we deal with a very old verb common to Indo-European,
which originally may have denoted something like ‘choose’ or ‘wish’
(cf. Grimm /vol. : ). In the case of the meaning of ‘wish’ for
wotte, a reversion to the original lexical meaning would be implied.
If we are less strict about ‘degrammaticalization’ and admit a larger
range of phenomena that are partially de-grammaticalizing in one
sense or the other, Norde () is a rich source for a very systematic
classification and description of such changes. She suggests that de-
grammaticalization will never be a mirror image of grammaticalization,
but occurs with some frequency if defined as follows:
Degrammaticalization is a composite change whereby a gram in a specific context
gains in autonomy or substance on more than one linguistic level (semantics, mor-
phology, syntax or phonology). (Norde : )
The term ‘autonomy’ already reveals an important source of Norde’s
analysis, namely Lehmann’s six parameters of grammaticalization. Recall
that Lehmann ( []) arranged the three criteria of ‘weight,’ ‘cohe-
sion,’ and ‘variability’ in paradigmatic and syntagmatic dimension, leading
to six parameters (see Section .). On this view then, grammaticalization
is loss of ‘autonomy’ of a ‘linguistic sign’ along these parameters. A gain in
‘autonomy’ conversely would mean degrammaticalization. Norde com-
bines the parameters with the idea of three different levels of observation:
‘content,’ ‘content-syntactic,’ and ‘morphosyntactic,’ and proposes three
types of degrammaticalization, as in ().
() Three types of degrammaticalization, according to Norde (:
, : –)
i. Degrammation: A function word in a specific context is
reanalyzed as a member of a major word class, acquiring
the morphosyntactic properties which are typical of that
word class and gaining in semantic substance.
ii. Deinflectionalization: An inflectional affix in a specific con-
text gains a new function, while shifting to a less bound
morpheme.
iii. Debonding: A bound morpheme in a specific linguistic
context becomes a free morpheme.
An example for ‘degrammation’ was already provided, namely Penn-
sylvania German wotte, which is a paradigm case of degrammaticalization
for Norde (, ). Norde () presents four more, one of which is
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DEGRAMMATICALIZATION (ANTI-GRAMMATICALIZATION)
Bulgarian nešto, the only case also acknowledged by Börjars and Vincent
() (see above).
The best known case of ‘deinflectionalization’ is the English s-geni-
tive, one of five cases of deinflectionalization provided by Norde ().
It is described in some historical detail in Tabor and Traugott ()
and Rosenbach (). Old English -s was a genitive case inflection that
alternated with other inflections. When a noun phrase was genitive,
each of the elements of the noun phrase was marked as genitive, as
in ():
() Old English (ca. , Aelfric Hom ) (Tabor and Traugott
: )
ðæs cyning:es sweoster Ecgfrid:es
the: king: sister: Ecgfrid:
‘the sister of Ecgfrid the king’
In the late twelfth century, examples can be found in which the genitive
is only marked once in the noun phrase, namely on the possessor head,
as in ():
() Middle English (late twelfth century, Seinte Katerine) (Rosen-
bach : )
þ was te deoules budel
that was the- devil: door.man
‘who was the devil’s door-man’
Finally, -s came to be attached not to the possessor head, but to the last
element of a possessive phrase, even if it is not the possessor but only a
postmodifier, as in ():³
() Early Modern English (sixteenth century, Madox) (Rosenbach
: )
the king of Portingales ship
the king of Portugal: ship
‘the ship of the king of Portugal’
³ Note that the actual history of -s is complicated by the fact that there apparently was a
transitory stage at which not -s but his or ys were attached to the last element of the noun
phrase. This his or ys may have either been the possessive article his, in which case there is
no straightforward development from genitive case marker -s to phrase marker, or a
phonetically strengthened version of -s (cf. Rosenbach : –).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
In other words, genitive -s changed from a genitive noun marker to a
possessive phrase marker. Further, it has also be claimed that it changed
from inflection to clitic. Major arguments in favor of deinflectionaliza-
tion, as an instance of degrammaticalization, could be as follows:
() Depararadigmatization in Lehmann’s sense took place, that is,
something that was once the member of the paradigm of noun declen-
sion now became a phrase marker, () deobligatorification took place,
since -s is not obligatory any more on every word of a noun phrase,
() the “structural scope” (better: structural size) in Lehmann’s sense
expanded instead of shrinking, and () the degree of attachment of the
genitive to the noun phrase became weaker, as -s became clitic-like.
Three of the four points may be of limited validity: Both the fact that
-s is not part anymore of a paradigm and that it is not marked anymore
on every element of a noun phrase are largely due to developments
external to -s: English noun declension in general faded away, so that -s
can no longer be a member of an inflectional paradigm. Likewise, case
marking of all other elements of the noun phrase except nouns has
disappeared. Furthermore, it has been questioned that Modern English
-s is a clitic (cf. Norde : ), and lastly, the usefulness of the
category ‘clitic’ in a grammaticalization cline as such has also been
questioned (Börjars and Vincent : ). Nevertheless, it does
seem that morphologically degrammaticalization has taken place in
that the morpheme -s has gained instead of losing autonomy.
On the other hand, it is difficult to see that -s degrammaticalized in a
functional sense. Quite the contrary, in expanding its scope (also in the
general linguistic logical-semantic sense), it functionally underwent
grammaticalization.
Finally, we have debonding, which is presented with a full ten
examples by Norde (: –). This shows that ‘debonding’
may be the change that is most likely to occur among the three types
of degrammaticalization. Also, ‘debonding’ is, as a rule, a purely mor-
phological change (cf. Norde : ). One of the best known cases
from the literature—and also included by Norde ()—are Japanese
clause connectives that developed from clause-internal clitics to clause-
initial adverbs. For example, the causal connective kara was initially an
ablative case marker. Like many case markers in Japanese, it also
developed a connective function, first in combination with other
words, and then alone. () illustrates the ablative use, and () the
connective use.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. DEGRAMMATICALIZATION (ANTI-GRAMMATICALIZATION)
() Middle Japanese, sixteenth century (Higashiizumi : )
Yoso=kara
afar=
‘from a distance’/ ‘from outside’
() Middle Japanese, c. (Higashiizumi : )
Muyoo=na koto=o yuu=kara . . . .
unnecessary= word= say=
‘Since you say something unnecessary, . . . .’
In the nineteenth century then, kara came to be used sentence-initially,
either attached to the copula da, or as a standalone, as in ():
() Early Modern Japanese (, Daisen sekai gakuya sagashi)
(cited from NKD :)
Geshi~gerō=no te=ni kakat.te,
servant-commoner= hand= fall.
nanor.u=de=mo nee=to
name.== not.be=
omo.u=mo muri=dya gozar-an.u.
think. = unreasonable= be[]--
Kara wasi=ga sitikudoku mibu=o akas.u=sa.
I= pesky standing= unveil.=
‘It is only reasonable that you don’t want to say your name if you
have fallen into the hand of a commoner. So, I will go ahead and
tell you my own standing.’
Sentence-initial connectives are preceded by a pause, and the preceding
word can have any marking, while the marking of words preceding
sentence-internal connectives is restricted. They furthermore can be
used as discourse markers. Example () may be ambiguous between
functioning as a logical connection between two propositions or as
discourse marker, that is, giving a reason for the following speech act
rather than sentence. The discourse marker function becomes especially
clear when the connective is eventually used turn-initially as in ().
() Modern Japanese; constructed (Matsumoto : )
Dakara chuui~si~nasai=tte it.ta==da.
take.care~do~do[]= say.==
‘That’s why I told you to be careful.’
Morphologically, kara and other Japanese connectives clearly under-
went a development towards greater ‘autonomy,’ that is, counter to the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
expected directionality of grammaticalization. However, functionally
speaking, kara has developed increasingly wider scope, from scope over
an NP to scope over a clause, and then over a whole sentence, reaching
beyond sentence boundaries as a discourse marker. Therefore, func-
tionally it has rather continuously undergone grammaticalization
before being coopted as a discourse marker (see Chapter ).
The discussion above has shown that genuine degrammaticalizations,
in the sense of the reverse of grammaticalization are rare if existent at all.
However, if one focuses only on specific aspects or parts of the process,
they are easier to come upon. In particular, it seems that degrammatica-
lization in a purely morphological sense, that is, towards a greater mor-
phological autonomy of a morpheme, is relatively “cheap.” However, the
core universal aspect of grammaticalization is functional. Morphology,
although it is the aspect of grammaticalization that is the easiest to observe
and describe, is language-specific and not essential for the process
(cf. Narrog c). In some languages, there is very limited morphological
development in grammaticalization (cf. Bisang , ).
In any case, as is true for most aspects of grammaticalization that we
have discussed in this book, the scope of the phenomenon of degram-
maticalization, and even its very existence, will vary widely depending
on one’s definitions of grammaticalization and degrammaticalization
and the criteria applied.
Further reading
Literature on degrammaticalization is plentiful, even beyond what has been
cited above. Norde () is the standard reference. Some particularly
remarkable contributions not yet cited include Giacalone Ramat (),
discussing the boundaries of grammaticalization, Heine (b), providing
a comprehensive picture of degrammatialization in relation to grammatical-
ization and other processes, Brinton () on English subject clitics as
potential examples of degrammaticalization, Schlüter () with a detailed
corpus study on marginal modal dare as a possible example of degramma-
ticalization, Norde () with a detailed breakdown of Lehmann’s para-
meters with respect to degrammaticalization, Trousdale and Norde () on
degrammaticalization from the specific perspective of Construction Gram-
mar, and Viti’s () attempt to filter out ‘authentic’ cases of degrammati-
calization from ‘non-authentic’ ones and discussing the potential influence of
language type on degrammaticalization.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. E X A P T A T I O N /R E G R A M M A T I C A L I Z A T I O N
10.3 Exaptation/regrammaticalization
The term ‘exaptation’ was introduced into linguistics by Lass ()
from biology (Gould and Vrba ). There, it referred to seemingly
superfluous or redundant structures in organisms that were utilized for
new purposes:
[O]rganisms—and I suggest, other historically evolved systems, like language—may in
their structure show a certain amount of bricolage; they are to some extent jury-rigged
or cobbled together, and the remnants of old structures can be recobbled into
new ones. (Lass : )
An example brought up by Lass () is the Dutch adjective ending -e,
which attached to adjectives based on the gender of the following noun.
In Afrikaans, gender was lost, and so we would expect that accordingly
the ending -e would either be dropped or generalized to all adjectives.
What happened instead was that now certain adjectives categorially
(always) would have -e, namely morphologically and morphophone-
mically complex ones, while simple monosyllabic adjectives would be
without -e. Lastly, with only a few adjectives, both presence and absence
of -e became available, with different meanings. For example, ’n enkel
man would mean ‘a solitary man,’ while ’n enkel-e man would mean ‘a
single (unmarried) man.’ Thus, due to the loss of gender, the suffix -e
surrendered its original grammatical function, but instead of getting
eliminated, speakers assigned a new function to the ending. Another
nice example is from English dialects, where the plural past form were
of the verb be has become a negative marker (Willis ). In these
dialects, the last remaining person–number distinction with regular
verbs in Modern English, the third person singular -s (I sing, she
sing-s) has become obsolete. Instead of preserving the number distinc-
tion between was and were, speakers use was in non-negated contexts
and were, that is, weren’t, in negation, for all persons (e.g. I was, we was;
I weren’t, we weren’t).
As exaptation has garnered some attention, various properties have
been ascribed to it. These are mainly (a) unexpectedness of the new
function—that is, gap between the previous and the new function, (b)
novelty of the new function, and (c) junk status of the input (cf. van de
Velde and Norde ). We consider (c) as the core of the concept.
‘Unexpectedness’ (a), or the perception of a gap, is in historical
linguistics most often the artifact of comparing the initial state and
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
the end state of a change without paying attention to the actual process
or the context in which the change took place. For example, in the case
of were, where there is an apparent large gap between person–number
as the source and negation as the target, Willis (: ) argues that
the new function is in fact very well motivated in the context of
grammar: Speakers of English are used to stem allomorphy occuring
with negation. For example, do is pronounced much differently in do
than in don’t, the same holds for can and can’t and even more so for will
and won’t. So, construing weren’t as a polarity counterpart to was, when
were had no apparent function anymore, makes sense to acquirers of
the language. In most cases, on a closer view, there is a semantic
connection or motivation in discourse. It is obvious that speakers try
to make sense of extant linguistic material that has become functionally
opaque for purposes related to the original function.
Novelty (b) of the new function is another possible definitory restric-
tion of the concept but there is no logical necessity to introducing it.
Among the two changes discussed above, Afrikaans -e would qualify as
an exaptation because it introduced a distinction into the language that
was not yet expressed. In contrast, English dialectal weren’t would be
excluded because negation is not a novel function in English. But this
exclusion would seem arbitrary and confusing with respect to other
changes. Would, then, grammaticalization leading to a new category
also be exaptation? It would further be un-economical since one would
have to create yet another concept for cases such as dialectal weren’t.
The concept of ‘novelty’ as such is also problematic because categories
might already exist in a language but not with exactly the same seman-
tic properties, or it may have been expressed “only” pragmatically or
lexically. We will therefore stick to ‘junk’ as the core concept. However,
‘junk,’ a term used by Lass () himself, is a pejorative label intro-
duced for rhetorical purposes that has caused negative reactions. Scho-
lars have furthermore contested that it is possible to claim for any
structure in language that it is totally useless and ‘junk’ (e.g. Smith
). Better terms may be ‘functionally opaque’ (Narrog , )
or ‘obsolescent’ (Willis , ). Morphological material has lost a
regular and clear form–function association, often because the catego-
rical distinction it expressed has become obsolete. Or it may have lost
the ‘competition’ to express a certain grammatical category to an
eventually more productive or newly emergent form, losing productiv-
ity and being absorbed into a limited number of lexical items. This may
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. E X A P T A T I O N /R E G R A M M A T I C A L I Z A T I O N
not only occur at the end of a ‘grammaticalization cline,’ as suggested
by Traugott (: ), but at any point in grammaticalization.
If the functionally opaque material ‘recycled’ for a new grammatical
function was clearly grammatical at some point before, we can speak of
‘regrammaticalization’ (cf. Greenberg : ).⁴ However, in some
cases we do not have enough historical knowledge to be sure that it
already had grammatical status at a previous stage. Therefore it makes
sense to retain the term ‘exaptation’ and apply the term ‘regrammati-
calization’ to those cases where a previous grammatical function is
unambiguously documented.
Thus far, it may seem that ‘exaptation’ only refers to some very rare
and quirky cases of development of grammatical material. However,
the extent to which exaptation occurs may be larger than expected.
Most research focusing on grammaticalization-related processes of
language change has focused on issues that seemed theoretically espe-
cially promising, that is, on distinguishing grammaticalization from
lexicalization, and on finding examples of genuine degrammaticaliza-
tion as a challenge to grammaticalization theory. In contrast, exapta-
tion/regrammaticalization, which is not as common as lexicalization
and not counterdirectional to grammaticalization as degrammaticali-
zation would potentially be, has been flying under the radar. Probably
this process is also more common in languages with rich morphology.
For Japanese, for example, Narrog () presented six cases of ex-
aptation from Japanese language history.
One case is a lexicalized transitivization pattern becoming regular and
productive as a causative. These lexicalized transitivization patterns are
likely to be the fossilized remnants of a former grammatical pattern:
Erstwhile productive causative suffixes become fossilized and lexicalized
as part of verb stems. They end up as phonological elements of a transitive
verb in intransitive–transitive verb pairs, or of the ditransitive counterpart
to a transitive verb.⁵ Since the erstwhile suffixes have been absorbed into the
lexicon, they do not participate further in grammaticalization. This is what
⁴ There is also the similar term ‘refunctionalization’ (cf. Smith ).
⁵ Note that a terminological difference between ‘causative’ and ‘transitive/intransitive’ is
made here. Some approaches, especially formal approaches employing lexical conceptual
structure (LCS) would not necessarily make such a distinction. However, at least in
Japanese, a clear borderline can be drawn between lexical patterns of transitive/intransitive
verbs and productive causative morphology. Furthermore, in typological studies, Nichols
et al. () have established the terms ‘transitivizing’ and ‘detransitivizing’ for transitivity
pairs. These are the terms used here.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
has been found in quite a few languages, for example in Lahu (Matisoff
), Tamil (Pedersen ), and Germanic (the -jan verbs; e.g. Kemmer
and Verhagen ; García García ).
In Old Japanese, verbs whose stem ended in -s-/-se- marked an
additional argument, that is, they were the transitive in an
intransitive–transitive verb pair, or the ditransitive in a transitive–
ditransitive pair. Thus, pair () is intransitive–transitive, while pairs
() and () are transitive–ditransitive.
() ok- ‘rise’ vs. okos- ‘raise’
() mi- ‘see’ vs. mise- ‘show’
() ap- ‘fit, meet’ vs. apase- ‘fit (something) in / adapt (something) to’
These transitive verbs are probably based on a productive formation in
Proto-Japanese or earlier with the verb se- ‘do’, that is, ‘rise’ plus ‘do’ is
rendered ‘raise’ (see Kuteva et al. , DO > CAUSATIVE). In Old
Japanese, these formations were already morphologically and function-
ally opaque and fused into lexical items. For one thing, many of these
verbs can no longer be segmented into stem and suffix according to
productive rules of morpheme concatenation in Old Japanese. This is,
for example, true for ok- ‘rise’ vs. okos- ‘raise’ in () above. For another
thing, the pattern was already restricted to a lexically determined set
of verbs and competed with other lexically determined patterns of
(in)transitivization. Thirdly, many of these verbs had developed idio-
syncratic meanings and uses which do not constitute simple causativi-
zation of the intransitive (for example, an emperor can okos- ‘raise’
soldiers, but soldiers do not ok- ‘rise’ in a corresponding meaning).
Overall, this morphological pattern is no longer a productive part of
grammar and has become part of the lexicon. As such it has become, to
some extent, opaque or obsolete but not entirely meaningless.
Decisively, then, this pattern of verbs ending in -se- was the likely source
of the regular productive causative suffix verb -(s)ase- in Late Old Japanese
with the allomorphs -ase-, as in () and -sase-, as in () and ():
() omof- ‘think’ vs. omof-ase- ‘make/let think’
() s- ‘do’ vs. se-sase- ‘make/let do’
() mi- ‘look’ vs. mi-sase- ‘make/let look’; also
mise- ‘show’ vs. mise-sase- ‘make/let show’
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. E X A P T A T I O N /R E G R A M M A T I C A L I Z A T I O N
The lexicalized verb pairs as in () to () are the only likely source for
the new grammatical suffix that is known to us, and there is also a
plausible scenario for this development. Many of the transitive verbs
ended in the sequence /ase/.⁶ Apase- ‘fit in, adapt to’ as in () is an
example. The vowel /a/ between the verb stem ap- ‘fit, meet’ and se- is
most likely originally part of a proto-Japanese vowel stem often ending
in /a/, rather than the ending, but the frequent sequence /ase/ probably
encouraged a reanalysis as ap-ase-. This form already accounts for the
majority of Japanese verbs, whose stem ends on a consonant, and to
which -ase- came to be suffixed productively from Late Old Japanese
on. We further need the insertion of an /s/ in front of -ase- to form the
allomorph -sase- suffixed to vowel-stem verbs. As a result, we even
obtain a complex causative verb from a transitive mi-sase- ‘make/let
see’ based on a transitive verb that competes with an already extant
ditransitive verb ‘show’ (cf. ()), and a causative from a ditransitive
verb, mise-sase ‘let see’ (). Morphologically, it seems that accretion of
phonological material in the form of pseudo-reduplication took place.⁷
The process hypothesized here is one of ‘secretion’ in the sense of
Jespersen (: –; cf. also Wischer : –).
With respect to the functional side of the change, we know that
Japanese had a productive causative -(a)sime- in Old Japanese, which
was already in decline at that time, and later only survived in particular
written registers of the language (and in some dialects). The issue
whether there is a “need” to express some category is notoriously
contentious, but if there was such a need, -(s)ase- clearly emerged to
replace the older -(a)sime- as the productive causative, and filled that
need. This is what it has served as until Contemporary Japanese.
Crucially for our purposes, it came in as an exaptation from fossilized
lexicalized material on the way to demise.
One thing that is spectacular about this case of exaptation is that
a similar development is known from a few other languages as
well. In Sanskrit, a development of the -áya-formations of the
older Vedic period, which are a lexical transitivization pattern, to the
Classical Sanskrit morphological causative suffix -áya- is known
⁶ With the slashes “//” we indicate that we consider this as a phoneme sequence and not
a morpheme.
⁷ One might naively suspect that another ‘do’-verb was inserted but this would have led
to the form -sise- and not -sase-, so it does seem that we are dealing with a quasi-
reduplication of the /s/ in front of the -ase-sequence here.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
(Jamison , ). The Vedic -áya-formations were originally tran-
sitive counterparts to intransitive verbs, and a semantically irregular
and functionally opaque pattern. According to Jamison, the expansion
of the pattern and its ascent to morphological productivity started from
verbs of motion, consumption, perception, and enjoyment, that is,
verbs that are grammatically construed as transitive verbs, but are
semantically intermediate between transitivity and intransitivity. In
other words, the transitivization of the experiencer verbs was reinter-
preted as a causativization of a transitive verb, and this causativization
then spread to other verbs.
A similar development can even be found in one more entirely
unrelated group of languages. In a number of North American lan-
guages, a lexical instrumental pattern has been the source of productive
causative prefixes. Mithun () has described this process in some
detail for the Shoshone causative prefix ma- (cf. Mithun : , ,
). According to her, Proto-Uto-Aztecan *maʔa ‘hand’ first lexica-
lized into Tümpisa ma- ‘with the hand’; for example ma-sungkwa’ah
‘feel with the hand.’ This in turn later became the source of the
Tümpisa grammatical causative ma- as in ma-tukwiihwa ‘let go out/
put out’ (not necessarily involving hand action). Mithun suggests that it
was exactly the stage of lexicalization of the pattern, in which the
original meaning and function of the prefix grew opaque, that made
the further development to causative possible. She writes that,
Because of the large inventory of causative verbs containing hand-action prefixes . . . ,
the implication of causation in these verbs was ultimately reinterpreted as the primary
meaning of the prefixes. The reinterpretation could not have taken place without the
tight bridging context provided by lexicalization. . . . ” (Mithun : –)
The last example here is a case where the source of the exaptation is
known to have been grammatical at an earlier point in history. Thus we
are dealing here with genuine ‘regrammaticalization.’ The later evidential
-rasi- was in Late Middle Japanese an adjectival ending on a few nouns,
and most likely a fossilized remnant of the Late Old Japanese productive
inferential particle rasi. It denoted the factual appearance (i.e. appearance
coinciding with fact as perceived by the speaker) of some property. For
example, airasi-, based on the noun ai ‘love’ meant ‘cute,’ and dokurasi-
based on the noun doku ‘poison’ meant ‘poisonous.’ In extension it could
also mean ‘typicality,’ as in otokorasi-, based on the noun otoko ‘man,’ and
meaning ‘manly’ (cf. Narrog a; Section ..).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. E X A P T A T I O N /R E G R A M M A T I C A L I Z A T I O N
Much later, in the eighteenth century, this ending spread to nouns
other than those few that were lexically determined, denoting not only a
factual appearance but, reversely, false appearance. Under the newly
available reading, otokorasi- would now not mean ‘manly’ but ‘out-
wardly appearing like a man, but actually being something else (e.g. a
woman).’ It was probably this extension in meaning and use, and the
new ambiguity that led to a perception of -rasi- as something increas-
ingly independent from the lexical stem to which it was attached.
Furthermore, while the two extremes of factual appearance and false
appearance both denote a positive (vs. negative) fact, and are therefore
non-modal, the fact that it could be used for both these areas of
factuality positioned it well to encroach on the in-between area of the
factually undetermined, that is, modality and evidentiality. -rasi- thus
came to be used for denoting a semblance, as in makoto-rasi- (‘truth’ +
-rasi-) ‘appearing to be truthful.’ Expressions such as ‘appearing to be
truthful’ are bound to be used as the predicate of a clause (e.g. ‘this
appears to be truthful’). Thus, -rasi- eventually became associated with
clause level predication, and spread to word classes other than nouns.
First, these were sentence adverbs like soo ‘to be so’ (e.g. soo=rasi-
‘seems to be so’), and finally also adjectives and verbs.
In terms of morphology, -rasi- must be classified as a ‘particle’ at this
point, that is, as a clitic able to attach to various word classes. It is
therefore not a noun suffix anymore. It thus moved towards greater
morphosyntactic independence. In terms of function, by the nineteenth
century it had become an evidential marker with scope over entire
clauses, and sentences, instead of just over a noun, as in example ().
() Early Modern Japanese (Iroha Bunko –; Yuzawa :
)
Ano XX-saN=wa, onusi=ni yoppodo ki=ga
(name)-= you= very feeling=
ar.u=rasi.i
be.=.
‘That XX seems to have feelings for you.’
The new particle rasi- moreover developed hearsay uses. With respect
to exaptation and regrammaticalization, the former ending that was
fossilized and absorbed into a handful of nouns, was recycled to
gradually become a fully productive evidential marker. The presence
of the Classical Japanese particle rasi in literary classics and in written
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
documents based on literary classics may also have supported this
process. Educated speakers knew the classical inferential particle and
how it was used in classical Japanese.
Narrog has collected an unpublished database of Japanese with more
than cases of grammaticalizations, compared to which six cases of
exaptations, is not a high number, or to be precise, only about .
percent of the changes leading to the creation of grammatical material.
Why is exaptation/regrammaticalization less frequent than grammati-
calization? The answer is rather obvious. The lexical material available
in any language is a practically infinite source for the expression of
grammatical categories, that is, for grammaticalization. In contrast,
erstwhile grammatical material that has become functionally opaque
or fossilized is rather limited. Furthermore, if ‘expressiveness’ is indeed
one of the crucial factors in grammaticalization as it has traditionally
been held to be (cf. Section .), the relatively short and functionally
opaque forms available for exaptation are not effective source material.
They do have the advantage, though, of being already entrenched.
Interestingly, in Japanese there is a tendency to back up the expressive-
ness of such lexicalized material through partial reduplication of
the phonological material (cf. causatives above; Narrog ). Despite
the relative scarcity, exaptations may still be much more common than
genuine degrammaticalizations, that is, cases of degrammaticalization
which are also functional. No such case has yet been established for
Japanese, unlike exaptation, for which six cases have been documented.
Furthermore, one scholar of Japanese language history claims that
exaptation was even the prevalent mode for the creation of grammatical
material in proto-Japanese, instead of grammaticalization. Koyanagi
(: –) calls the phenomenon “promotion to function word”
(shōkaku kinōgoka) and claims that proto-Japanese had word forma-
tion as a rich resource for grammaticalizations, instead of lexical words
in later historical times. This is entirely a matter of reconstruction then,
and refers to developments such as that discussed above, of the lexical
transitive verb endings becoming regular causative suffix verbs. Janda
() has dedicated a monograph to ‘defunct’ paradigms and cate-
gories in Slavic languages that gained new functionality and Smith
(, ) identified a large number of cases of ‘refunctionalization’
of paradigms in the nominal domain in Romance languages and
beyond. Overall, exaptation is a process that is still vastly understudied
compared to others.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. OTHER -IZATIONS
Further reading
While most relevant literature has already been cited above, we should like to
point out that there is full paper collection has been dedicated to the topic
(Norde and Van de Velde ) and that Lass, the original proponent of the
concept, also discussed it in some detail in Lass (). Furthermore, besides
the labels already mentioned, exaptation or overlapping processes are also
known as ‘secretion’ (Harnisch , ) and ‘functional renewal’ (Brinton
and Stein ). Haiman () presents a broader concept of exaptation
including exaptation into the lexicon. Kailuweit () and Rosemeyer
() focus on the conditions for ‘refunctionalization,’ trying to distinguish
it from ‘exaptation.’
10.4 Other -izations
In this section we will briefly introduce other terms and labels for
processes that have been related to grammaticalization in previous
research.
The term phonologization refers to a process whereby allophones of
one phoneme develop into phonemes on their own. The process may
further feed into morphology. A well-known example is the develop-
ment of front rounded vowels /ø/ and /y/ in German, which originally
were contextual variants of /o/ and /u/ with /i/ in the following syllable.
Phonologization is only of limited relevance for grammaticalization. It
does lead to a more distant relationship between what were originally
two forms of one word; for example fall/fell, drink/drench.
Phonogenesis is the degrading of erstwhile morphemes into purely
phonological elements of a word (cf. Hopper ). One example
already mentioned under ‘lexicalization’ would be the word-final /r/
in Modern English near, which was originally the comparative adjective
ending -(e)r, that can still be found in a number of adjectives such as
strong/stronger. Another example would be the /s/ in craftsman, which
was originally a genitive -s. In order to reduce terminology one could
also speak of phonogenesis as ‘phonologization from above’ vs. ‘pho-
nologization (from below),’ but in contrast to ‘phonologization (from
below),’ the result is extant phonological material and not a new
phoneme.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
The term morphologization has been used in two senses: Within the
study of grammaticalization, it has been used for morphological change
from an independent word to a bound morpheme, that is, the core of the
potential morphological development accompanying grammaticaliza-
tion (e.g. Hopper and Traugott : , : –). More frequently,
however, in the tradition of morphology-centered linguistic studies, this
term has been used in a much broader sense, namely generally for the
development of morphological material from various sources (e.g.
Joseph and Janda ; Gaeta ; Klausenburger ; Joseph ;
Andersen ).⁸ These are primarily phonological and secondarily
syntactic sources. The original use of the term morphologization is for
phonological operations gaining morphological value (cf. Gaeta :
). For example, we mentioned under ‘phonologization’ that phonol-
ogy may feed into morphology: The German umlauts [ø] and [y] not
only gained phonemic status as /ø/ and /y/ (phonologization) but also
became markers of plural, as in Mutter ‘mother’–Mütter ‘mothers.’
Morphologization from syntax includes morphologization during gram-
maticalization but may also include cases of lexicalization like that of
German heute ‘today’ from hiu tagu ‘this day’ (Joseph : ).
Demorphologization is reversely either the degradation of morpho-
logical patterns to phonology (in this case it would be roughly the same
as phonogenesis above),⁹ or from a more bound to a more independent
morpheme such as a clitic or independent word, in which case it would
roughly refer to degrammaticalization in a morphological sense.
Transcategorization, not to be confused with ‘transcategoriality’
(Section .), is a shift of a category from one word class to another;
for example during from verb participle to preposition, and lateral shift
is a change in a category that does not involve change towards a more
(or less) grammaticalized category on a grammaticalization cline.
Joseph () gives as examples shifts of inflectional endings from
active voice to nonactive voice and vice versa in Greek. In the area of
modality, necessity may change to possibility and vice versa without a
⁸ “[Morphologization is] a set of developments by which some element or elements in a
language that are not a matter of morphology at one stage come to reside in a morpho-
logical component” (Joseph : ).
⁹ “When a morpheme loses its grammatical-semantic contribution to a word but retains
some remnant of its original form, and thus becomes an indistinguishable part of a word’s
phonological construction, I shall speak of the resulting phonological material as morpho-
logical residue, and of the process itself as demorphologization” (Hopper : ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONCLUSION
difference in degree of grammaticalization (cf. Narrog a;
Section ..).
Delexicalization is the reverse of lexicalization. Speakers analyze
words as the concatenation or fusion of morphemes that are often
not their actual source; for example ham-burger from hamburger,
which originally comes from Hamburg (place name) + -er (nominal
affix), and create new word formation based on their analysis. More
broadly, delexicalization can be understood as a type of remotivation
(Harnisch and Krieger ), whereby speakers ascribe more structure
and analytical meaning to a word or construction than it actually has
and reanalyze it accordingly. Relexicalization, in contrast, would refer
precisely to a case of morphological degrammaticalization (demorpho-
logization), in which an erstwhile morpheme becomes a lexeme.
Retraction (Haspelmath ) was already mentioned as a term
referring to the loss of a further advanced grammatical function that
a linguistic item developed, with only the less advanced function(s)
remaining.
Regrammation is a term that has been proposed by Andersen ()
for what we refer to as ‘secondary grammaticalization’ in this book
(Section .), while pragmaticalization refers to the development of
discourse markers as something different from grammaticalization. We
have attached enough importance to this process to dedicate the whole
of Chapter to it.
10.5 Conclusion
In this chapter, we have discussed in some detail the three processes
that are either theoretically or empirically most relevant for grammat-
icalization, namely, lexicalization, degrammaticalization, and exapta-
tion/regrammaticalization each in a section of their own, and others
very briefly in Section .. Morphologization is also an empirically and
theoretically important concept, but in its narrow definition as the
typical morphological process from more to less independent mor-
phemes, it is already part of the study of grammaticalization, and, in
its broad definition as any change that leads to new morphology, it does
not have a well-defined relationship with grammaticalization. Pragma-
ticalization is a theoretically and empirically important concept and
also relevant to grammaticalization but is afforded a chapter of its own
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
(Chapter ). Lexicalization, degrammaticalization, and exaptation/re-
grammaticalization are in a well-defined and relevant relationship to
grammaticalization. This relationship can be represented as in
Figure ..
grammaticalization
words and phrases
elements of grammar
(degrammaticalization)
lexicalization fossilization/obsolescensce
(vs. delexicalization) (vs. exaptation/regrammaticalization)
lexical entries
Figure 10.1 Lexicalization, degrammaticalization, and exaptation vis-à-vis grammaticalization
Figure . shows that words and phrases can either lexicalize or
grammaticalize. Lexicalization and grammaticalization are the ‘ele-
phants’ in the room with respect to change with words and syntactic
phrases as the source. This is indicated through the regular arrows. In
case of grammaticalization, once a word or construction has entered the
domain of grammar, there is a long potential ‘runway’ of categories along
which the semantic scope of the category widens, and the item further
grammaticalizes, for example from a resultative to anterior, then perfec-
tive and simple past, or from desire to intention to future and then a
complementizer (cf. Bybee et al. : , ). Morphological gram-
maticalization is optional. That is, the typical morphological ‘grammat-
icalization cline’ (cf. Section .) may or may not accompany the
development. At any stage, an element of grammar can become obsolete.
Often it may entirely disappear, like most person/number endings of
English verbs or case endings of English nouns. It then goes to “zero,” a
case which is not represented in Figure .. It may also fossilize and be
absorbed into the lexicon, either as a lexical entry itself or as a phono-
logical part of a lexical entry (‘phonogenesis’). These are the most
common developments. The lexicon is the largest reservoir for linguistic
material and acts like a huge sink, eventually attracting regular and
transparent combinations of words and morphemes. Individual lexical
items once more become the input for syntactic combination. This is
indicated by the large arrow on the left.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONCLUSION
However, sometimes processes take place in the opposite direction.
The opposite processes are much less common and regular, and are
therefore indicated by dotted lines and the terms in brackets. Theoret-
ically the most interesting, but empirically probably the rarest, is
degrammaticalization. It is usually confined to the morphological
aspect, that is, a non-central aspect of grammaticalization, and lasts
for only one step—in contrast to grammaticalization, which can con-
tinue for long stretches of development often along cross-linguistically
common grammaticalization paths. This is why the degrammaticaliza-
tion arrows are represented as shorter than the grammaticalization
arrow. Delexicalization and its scope is basically outside the range of
this book but has been included in the figure for the sake of complete-
ness. If delexicalization is also labeled as ‘exaptation’ (cf. Haiman ),
then exaptation could be used as the unified term for all processes
coming back from the lexicon either to the syntactic combinations of
words, or to grammatical elements. Exaptation brings back obsolescent
or obsolete grammatical elements (or elements that have the appear-
ance of erstwhile grammatical elements) to grammar. It is probably
more common than degrammaticalization, at least it is more common
than genuine degrammaticalization in a functional sense. However, it
has received less attention in the literature probably because it does not
have the same theoretical appeal as degrammaticalization and the
category labels under which largely the same phenomena have been
discussed (exaptation, regrammaticalization, refunctionalization, func-
tional renewal, secretion, remotivation, partially also morphologiza-
tion) have been splintered to an excessive degree.
Further reading
Other authors who try to provide an integrated picture of grammaticalization
with other processes include Lehmann (, ), Wischer (, ,
), Heine (b), Brinton and Traugott (, ), and Norde and
Beijering ().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
BEYOND GRAMMATICALIZATION
Discussion points
() Take the example of a common word formation affix from English or
another language that you are familiar with; for example English -able
(rechargeable), or -ty (immunity). Would you consider them lexicalizations
or grammaticalizations? What would be the arguments in favor of the one
or the other?
() In your view, should the term ‘degrammaticalization’ be restricted to the
exact reversal of grammaticalization, or should partial counterdevelop-
ments also be allowed to count as ‘degrammaticalizations’? How would
one stance or the other impact your concept of grammaticalization?
() Exaptation is a term borrowed from evolutionary biology. In your view, is
there justification for using a term from biology for linguistic phenomena
as discussed in the chapter, and if yes, what would be a justification?
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
11
Discourse markers
11.1 What are discourse markers?
11.2 Problematic features of discourse markers
11.3 Hypotheses on the development of discourse markers
11.4 Conclusions
Discussion points
Grammaticalization theory is by now widely accepted as one of the
main tools of historical linguistics. But there are also problems with its
application, as has been pointed out in a number of publications, and in
, a book was devoted precisely to this issue, entitled The Limits of
Grammaticalization (Giacalone Ramat and Hopper ). Some of the
main problems were discussed in Chapter , relating most of all to the
unidirectionality principle, which is a cornerstone of the theory.
One of the problems surfacing in the course of the last decades was
the status of discourse markers (henceforth: DMs): Is their rise and
development the result of grammaticalization, like that of other gram-
matical forms, such as tense markers, case markers, conjunctions, etc.?
This problem, which has generated some amount of research output
(see, e.g., Brinton ; Heine et al. ), has been discussed contro-
versially. It was argued in particular that grammaticalization is inade-
quate as a tool to account for the history of DMs and an alternative
mechanism was proposed, namely ‘pragmaticalization.’
In , a volume on the development of DMs was published as a
special issue of the journal Linguistics (Volume , ). Observing that
there had been ‘an explosion of research on discourse markers’ in the
preceding two decades, the editors asked the contributors to the
volume to comment on the question of which of the main hypotheses
that had been voiced in previous work was best suited to account for
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
this development: grammaticalization, pragmaticalization, or perhaps
‘simple’ semantic change? The outcome of this project, which had a
noteworthy impact on ensuing research, is summarized by the editors
Degand and Simon-Vandenbergen () thus:
. . . we had not expected these contributions to be so unanimous in their rejection of
pragmaticalization as a substitute term for the processes described. All authors plead
for grammaticalization as the best possible explanation for the developments de-
scribed, on condition that grammar is given a wider sense . . . .
(Degand and Simon-Vandenbergen : )
As straightforward as this comment may be, however, it does not solve
the problem of how the history of DMs is to be described, as we will show
in the present chapter. To this end, Section . provides a short intro-
duction on what DMs are about. The problems that a student of gram-
maticalization is faced with when analyzing the history of DMs are then
discussed in Section ., while Section . presents an overview of the
main hypotheses that have been proposed to deal with these problems.
11.1 What are discourse markers?
The following example, taken from Traugott (a: ), may be of help
to illustrate the problem involved.
() a. She spoke well.
b. Well, she spoke.
The item well in (a) is a manner adverb modifying the meaning of the
verb. The homophonous item well in (b), by contrast, is not a syntactic
constituent of the clause, and it is frequently set off prosodically from
the clause, typically marked in writing by a comma. And it also differs
from well in (a) in its meaning: Rather than contributing to the
semantic content of the clause, it has been described as doing ‘meta-
textual work’ (Traugott a: ).
The item well in (b) is commonly described as a DM. It seems to be
fairly uncontroversial that the DM well developed historically out of
adverbial uses of well like that in (a) above. The question is: When
and, more importantly, how did this happen? The DM well can be
traced back to Old English, it is well attested for instance in the form
wel la or wella in King Alfred’s Anglo-Saxon Version of Boethius De
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. WHAT ARE DISCOURSE MARKERS?
consolatione philosophiae (Jucker : ). Does this mean that the
DM had already arisen in Old English, or even earlier?
As difficult as it is to answer this question, it is even more difficult to
establish how DMs such as well in (b) came to be what they are: Are
they the result of grammaticalization, as has frequently been claimed? It
is this question that is the subject of the following paragraphs.
The term ‘discourse marker’ is used in a wide range of senses and for
quite a number of different phenomena, extending from monosyllabic
interjection-like particles to clausal expressions. Also called discourse
particles, pragmatic markers, or discourse connectives, DMs have
become known under a large number of different names, and they
have been the subject of many studies.
Some authors do not make a distinction between discourse markers and
pragmatic markers, while others do (e.g. Fraser : ). Some authors
argue that DMs can be described exhaustively with reference to conven-
tional linguistic taxonomy, calling them adverbs or adjuncts, a ‘subclass of
adverbials,’ ‘connecting adverbs,’ linking adverbials, ‘speaker-oriented or
conjunctive sentence adverbs,’ or ‘connective adjuncts.’ Other authors
again maintain that DMs are elusive to conventional categories of gram-
mar and must be understood and described in their own right. This is the
predominant stance of students of DMs, and it is also adopted here.
One problem with definitions of DMs is that quite a few of them are
not highly specific. This also applies to the classic definition by Schiffrin
(: ), for whom discourse markers are “sequentially dependent
elements which bracket units of talk.” In the present work they are
defined as follows (see Heine et al. for elaboration):
Discourse markers are (a) invariable expressions which are (b) syntactically independent
from their environment, (c) typically set off prosodically from the rest of the utterance,
and (d) their function is metatextual, relating an utterance to the situation of discourse,
that is, to the organization of texts, speaker-hearer interaction, and/or the attitudes of
the speaker.
This definition differs from definitions proposed in the tradition of
Schiffrin (), which stipulate that DMs signal some kind of rela-
tionship between clauses or utterances,¹ because in a number of their
¹ This applies, for example, to the following definition by Traugott (b: ): “By a
DM I mean a metatextual marker that signals some kind of relationship between clauses/
utterances.”
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
uses, DMs do not signal such a relationship. This seems to apply, for
example, to uses of the English DM well in utterances such as the
following where no relationship of this kind needs to be involved: My
performance yesterday was not really, well, outstanding.
DMs have been classified, on the one hand, as a sub-type of prag-
matic markers (Fraser ) and, on the other hand, as a sub-type of
parentheticals (see the contributions in Dehé and Kavalova ) or, as
we will say here, of theticals (Kaltenböck et al. ). What distin-
guishes them from many other (paren)theticals is that they are largely
or entirely invariable, that is, they are formulaic, and they typically do
not allow internal modification (see Section ..). The items listed in
() are amongst the ones most commonly discussed and are the least
controversial items classified as English DMs, and it is these items that
are the topic of the present chapter.
() Paradigm English DMs
after all, anyway, as it were, besides, however, indeed, in fact, instead,
I mean, now, okay, so, then, I think, well, what else, you know, you see
But not all uses of these items really qualify as DMs. As we saw already
in (), for example, it is only the use of well in (b) that qualifies as a
DM, not that of (a). A characteristic of many DMs is in fact that they
have homophonous counterparts that are not DMs.
Further reading
Discourse markers, also called pragmatic markers, constitute a wide field of
research. The following are a few publications providing an insight into what
they are about: Schiffrin (, ), Brinton (, b: , , ),
Jucker (: , ), Jucker and Ziv (: –), Hansen (a, b),
Schourup (), Fischer (), Barth-Weingarten and Couper-Kuhlen
(: ), Aijmer (), Traugott and Dasher (: –), Dostie
(), Kaltenböck (: ), Dér (), and Heine et al. ().
11.2 Problematic features of discourse markers
For students of grammatical change, an interest in DMs arose mainly in
the s when some scholars suggested that such items are hard to
reconcile with established principles of grammaticalization. It is most
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. PROBLEMATIC FEATURES OF DISCOURSE MARKERS
of all changes such as the ones in () that are commonly found in the
development of DMs and provide a challenge to an understanding of
DMs in terms of grammaticalization.
() Changes commonly found in the development of discourse
markers
a. From syntactic constituent of the sentence to syntactically
unattached status
b. From prosodically integrated to non-integrated or less inte-
grated status
c. From meaning as part of the sentence to meaning outside the
sentence
d. From sentence function to metatextual function
e. From positionally constrained to largely unconstrained
placement
The catalog in () does not exhaust the number of features that have
been pointed out, most of all in the literature on pragmaticalization (see
Section .. below), but they are particularly hard to reconcile with
observations commonly made in grammaticalization. They are looked
at in more detail in Sections ..–... But before we go into such
details, Section .. deals with a more general problem concerning
the status of DMs.
11.2.1 Do discourse markers belong to grammar?
We noted in Section . that DMs are also called ‘pragmatic markers,’
and there have been some debates on what DMs actually are. Questions
that have been asked include, predominantly, the ones in ().
() Questions
a. Are DMs are part of grammar?
b. And if yes, how is grammar to be defined?
These questions have generated a number of discussions, most of all in
work that distinguishes in some form or other between a narrow or
restricted and a wider or expanded view of grammaticalization (see, for
example, Degand and Evers-Vermeul : , ). When reviewing
the literature on this issue as well as the way DMs are conventionally
treated in linguistic work it would seem that the most reasonable
answer to the two questions in () can be phrased as follows at the
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
present state of research: DMs are part of grammar and, accordingly,
grammar is best defined in way that includes DMs. Obviously, this
answer is not entirely satisfactory considering all the complex theory-
dependent issues that have been raised in connection with (). The
main reasons for the answer can be summarized as follows.
First, the answer is in accordance with, or is not contradicted by,
most of what has been written on this issue. For some scholars, the
proposal entails the need to adopt “a more comprehensive view of what
constitutes grammar” (Degand and Simon-Vandenbergen : ),
or “a broad definition of grammar” (Traugott and Trousdale :
). For others, such a need does not exist. This applies, for example,
to the framework of Discourse Grammar, where DMs are treated as
belonging to the component of Thetical Grammar rather than to
Sentence Grammar (Heine et al. ; Heine ). This distinction
corresponds largely to that proposed by Haselow (: ) between
‘macrogrammar,’ which refers to structural relations beyond isolated
sentences, and ‘microgrammar,’ referring to sentence-internal strucural
relations, respectively.
Second, the answer is also in accordance with the way DMs are
traditionally treated, for example, in standard grammars of English
such as those by Quirk et al. (: –), Biber et al. (:
–), or Huddleston and Pullum (: ff.), where DMs are
analyzed as a part of grammar, even if they may not be referred to with
the term ‘discourse markers.’
And finally, classifying DMs as belonging to a domain other than
‘grammar,’ for example to ‘pragmatics,’ would require DMs to be con-
sistently set off from ‘grammar.’ To our knowledge, this has so far not
been done. In the absence of more detailed evidence to the contrary we
follow those scholars who classify DMs as belonging to ‘grammar,’
ignoring the question of how exactly ‘grammar’ is to be defined.
11.2.2 From syntactic constituent of the sentence to
syntactically unattached status
DMs have been described as being typically associated with the follow-
ing features: They are syntactically isolated, extrasyntactic (Brinton
: ) or independent and outside the main clausal structure
(Brinton : ). For some authors, syntactic independence, detach-
ment, or freedom is therefore one of the most conspicuous features of
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. PROBLEMATIC FEATURES OF DISCOURSE MARKERS
DMs. Accordingly, the growth of DMs has been described as involving
“an increase in syntactic freedom instead of syntactic fixation” (Norde
: ; Beijering ).
Movement out of the morphosyntax of a sentence is a change that
stands in stark contrast to what can commonly be observed in gram-
maticalization, specifically to the parameter of decategorialization (see
Section .), and especially to the changes associated with external
decategorialization summarized in (), repeated here for convenience
from Chapter (a–d).
() Salient changes associated with external decategorialization
a. Loss of independence as an autonomous form.
b. Increasing dependence on some other form.
c. Increasing obligatoriness of use.
d. Loss of the ability to be moved from its canonical position to
other positions in the sentence in ways that are characteristic
of the corresponding non-grammaticalized source item.
In sum, rather than losing syntactic freedom, independence from
surrounding items, and positional variability, as commonly found in
grammaticalization, linguistic items becoming DMs move in the exact
opposite direction.
11.2.3 From prosodically integrated to non-integrated or less
integrated status
In a number of studies it has been pointed out that DMs tend to be
prosodically set off from their host utterance, often occurring “in an
independent breath unit carrying a special intonation and stress pat-
tern” (Traugott a: ). The hedge ‘tend to be’ indicates that distinct
prosody is not consistently found, or can even be missing in DMs.
What is obvious, however, is, first, that DMs are more likely to be
separated prosodically from their environment than the expressions
from which they are historically derived. And secondly, items such as
the adverb well in (a) or (a) may well be interpreted as DMs once
they are prosodically set off from the rest of a sentence, cf. (b).
() English
a. Paula lives well in Bagdhad.
b. Paula lives, well, in Bagdhad.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
In this respect, DMs differ from what would be expected in grammat-
icalization: As was discussed in Chapter , grammaticalizing items tend
to undergo erosion, losing suprasegmental features such as stress, tonal
distinctions, or intonation, and they may be adapted to adjacent pho-
nological units (Section ., ()). Loss of prosodic independence may
in fact lead “to an item’s inability to form a prosodic word of its own”
(Haspelmath : ), and Wichmann () concludes:
[G]rammaticalization involves not only (and not always) the attrition of phonetic
substance but more importantly the loss of prosodic prominence with concomitant
loss of independence in intonational structure. (Wichmann : )
In short, rather than leading from prosodically non-integrated to
integrated items, DMs are likely to move in the opposite direction. As
noted above, however, distinct prosody is not consistently found, or can
even be missing in DMs, and it may even happen that frequently
occurring DMs lose their prosodic distinctiveness, as has been ob-
served, for example, with the English markers of course and sorry
(Wichmann : –).
11.2.4 From meaning as part of a sentence to
meaning outside the sentence
When a DM evolves, it is no longer part of the meaning of the sentence,
and it does not affect the propositional content of a sentence or an
utterance (e.g. Jucker : ). Not being part of the meaning of a
sentence explains why the use of DMs is described as ‘optional’ in the
sense that “the content of an utterance is not altered if the DM is
removed” (Frank-Job : ). In accordance with these observa-
tions, DMs are commonly described as being non-truth-conditional,
they are not part of sentence questions, they cannot become the focus of
a cleft sentence, and they are not in the scope of a negated sentence
(cf. Espinal ).
Meanings arising in grammaticalization, by contrast, shape the
meaning of sentences, be that within a phrase or a clause, or between
clauses, but not normally beyond the level of sentences. For example, in
many languages, meaningful expressions used to structure noun
phrases or adverbial phrases, such as demonstratives, nominalizing
markers, adpositions, or case markers, are commonly grammaticalized
to markers of clauses combining like complementizers or relative clause
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. PROBLEMATIC FEATURES OF DISCOURSE MARKERS
markers (see Kuteva et al. for examples). However, such processes
are essentially restricted to the meaning content of sentences.
11.2.5 From sentence function to metatextual function
The functions of DMs have been described as introducing a higher-
level speech act, as being metatextual, metacommunicative, discourse-
interactional, as operating on the textual or discourse level, to serve as
text structuring devices at different levels of discourse, or as contribut-
ing to the interpretation of an utterance rather than to its propositional
content (e.g. Frank-Job : ; Fraser : ). And in work on
Relevance Theory, the function of DMs is described as procedural and
portrayed as serving to guide the hearer’s linkage of an utterance to an
appropriate context (e.g. Blakemore : ). Such features contrast
sharply with those captured in accounts that have been proposed for
grammaticalization processes of any kind, in particular in cross-
linguistic typological accounts (e.g. Kuteva et al. ). Rather than
being metatextual, functions arising in grammaticalization shape the
meaning of sentences, be that within a phrase or a clause, or between
clauses.
11.2.6 From positionally constrained to
largely unconstrained placement
Being semantically and syntactically unattached, DMs are, ideally,
unconstrained in their placement, that is, they may be found in various
slots of a sentence. Accordingly, Hansen (: ) notes that DMs
(pragmatic markers in her terminology) “frequently have great freedom
of position within the host utterance, and are thus syntagmatically
variable,” and Tabor and Traugott (: ) note that DMs “occur
in multiple positions in the clause.”
Positional mobility of the kind commonly found in DMs is at
variance with what can be observed in grammaticalization, which
typically leads to a reduction of placement options. Accordingly,
Brinton and Traugott (: ) conclude that Lehmann’s (
[]: –) notion of fixation or loss of syntagmatic variability is
problematic for DMs since they are “often quite moveable.” At the
same time there are also constraints on the placement of DMs,
and these constraints are in accordance with the particular discourse
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
functions expressed by them. The functionally preferred position is that
at the left periphery of a sentence, followed by the right periphery, and
less commonly within the sentence.
Overall, however, there is converging evidence to suggest that DMs
are in general distinctly more flexible in their placement than the
expressions from which they are etymologically derived. This observa-
tion is hard to reconcile with observations made in grammaticalization,
according to which the development from lexical to grammatical ex-
pressions almost invariably entails reduction or loss in positional
variability.
11.2.7 Conclusions
In sum, there are problems with the view that the rise of DMs can be
reduced to grammaticalization. What is more, the features discussed
above are not the only ones suggesting that the grammaticalization
hypothesis is in need of reconsideration, as we will see in Section ...
At the same time there are also features that are more readily
compatible with a general understanding of grammaticalization. This
applies in particular to the notion of procedural meaning, which con-
strains the computations needed to parse and comprehend an utter-
ance. Procedural meaning, contrasting with conceptual meaning, is
widely agreed to be a paradigm feature of DMs, but it is in the same
way considered to characterize grammaticalization.
Nevertheless, the features listed in Sections .. to .. are the
ones that are particularly hard to account for when using an interpre-
tation of DMs purely in terms of grammaticalization. The question to
be looked into in the next section is how such features are dealt with in
reconstructing the rise and development of DMs.
11.3 Hypotheses on the development of discourse markers
The discussion in Section . suggests that DMs are not only hard to
reconcile with but even contradict what has commonly been found
cross-linguistically in grammaticalization. A number of proposals were
made to account for this fact, but most of them can be reduced to the
hypotheses in ().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. HYPOTHESES ON THE DEVELOPMENT OF DISCOURSE MARKERS
() Hypotheses on the development of discourse markers
a. They are the result of grammaticalization.
b. They are the result of pragmaticalization.
c. They are the result of lexicalization.
d. They are the result of cooptation followed by grammaticalization.
The hypothees can be referred to, respectively, as the grammaticaliza-
tion hypothesis, the pragmaticalization hypothesis, the lexicalization
hypothesis, and the cooptation hypothesis. In the following paragraphs,
each of these hypotheses is looked at in turn.
Most of the studies dealing with the hypotheses discussed below use
DMs in English and a few other European languages, as well as Japanese
and Korean. Languages in other parts of the world, such as Australia,
Africa, or the Americas, have so far played hardly any role in this debate.
11.3.1 The grammaticalization hypothesis
Work carried out over the last decades has been based most commonly
on the assumption that the rise and/or development of DMs is the
result of grammaticalization (for references, see Further reading).
Proponents of the grammaticalization hypothesis do not all share the
same view of how discourse markers evolved. Rather, the two main
stances in () can be distinguished.
() Kinds of grammaticalization distinguished to account for dis-
course markers
a. A different kind of grammaticalization.
b. An expanded notion of grammaticalization.
A position in accordance with (a) is that of Wischer (), who
argues that processes relating to pragmaticalization should be assigned
to one subtype of grammaticalization that deals with text and discourse
level phenomena. For a number of proponents of (b), two views of
grammaticalization need to be distinguished, referred to, respectively,
as the ‘narrow’ and the ‘wider’ views. According to the ‘narrow,’ or the
‘restricted’ view, grammaticalization is compatible with the criteria
proposed by Lehmann ( []), and this view is said to have
problems to account for DMs. The ‘wider’ view, by contrast, takes care
of DMs, and perhaps also of other kinds of discourse material (e.g.
Degand and Evers-Vermeul ).
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
A more pronounced proposal relating to this distinction is made in
the study of Traugott and Trousdale (: –). Rejecting the term
‘pragmaticalization,’ they distinguish two kinds of approaches, referred
to as ‘GR’ (grammaticalization as restriction) and ‘GE’ (grammatical-
ization as extension). According to them, in GR approaches, grammat-
icalization is viewed as reduction and increase in dependency. As
examples of such approaches, studies published before are cited,
but there are also some more recent studies, such as Haspelmath ()
and Boye and Harder (). In GE approaches, by contrast, gram-
maticalization is claimed to be viewed as expansion. Traugott and
Trousdale () argue that there are a number of features distinguish-
ing the two approaches, and these are principally features characteriz-
ing DMs. For a critical discussion of Traugott and Trousdale (), see
Heine (a).
As we will see in Sections .. and .., the problems mentioned
in this section have also been a major issue in alternative approaches to
understanding the evolution of DMs.
Note, however, that doubts have been raised more recently by the
main proponent of the grammaticalization hypothesis on whether this
hypothesis is in fact appropriate to account for the rise of DMs. Thus,
Traugott (a: ), had argued earlier that “[t]o treat [the develop-
ment of DMs; a.n.] as a case of something other than grammaticaliza-
tion would be to obscure its similarities with the more canonical clines.”
Now she no longer thinks that the rise of DMs is a case of grammat-
icalization (Elizabeth Traugott, p.c. of March , ).
11.3.2 The pragmaticalization hypothesis
The observation that grammaticalization theory has problems with
explaining the nature of DMs has induced some authors to propose
an alternative process, generally referred to as pragmaticalization.
Pragmaticalization has been defined as
. . . the process by which a syntagma or word form, in a given context, changes its
propositional meaning in favor of an essentially metacommunicative, discourse inter-
actional meaning. (Frank-Job : )
Grammaticalization—it is argued—describes the emergence of
sentence-internal grammatical markers while pragmaticalization de-
scribes that of DMs, that is, of text structuring devices at different levels
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. HYPOTHESES ON THE DEVELOPMENT OF DISCOURSE MARKERS
of discourse. Following Norde (: ), pragmaticalization can be
defined more concisely as the development of DMs. On this view, then,
the presence of DMs presupposes prior pragmaticalization.
On the pragmaticalization hypothesis, lexical or other elements
develop into DMs without involving grammaticalization or via a kind
of grammaticalization that differs from ‘canonical’ grammaticalization
in some way, the result being markers “mainly serving as text structur-
ing devices at [non-sentential] levels of discourse” (Erman and
Kotsinas : ). Some authors in fact argue that the main reason
for distinguishing pragmaticalization as a distinct mechanism is
semantic-pragmatic (e.g. van Bogaert : ), but this view is not
shared by all. It is mostly the features listed in () that have been
adduced in favor of pragmaticalization.
() Features claimed to distinguish pragmaticalization from
grammaticalization
a. Syntactic isolation.
b. Lack of fusion.
c. Peculiar grammatical status.
d. Increase in structural scope.
e. Optionality.
f. Non-truth conditionality.
Three of the six features listed () relate to what was described in
Section . as ‘movement out of the morphosyntax of a sentence.’
This means that DMs are at variance with the grammaticalization
parameter of decategorialization (Section .), which would predict
syntactic integration rather than isolation, increasing fusion rather
than lack of fusion, and syntactic and grammatical fixation rather
than increasing syntactic freedom (cf. Norde : ; Beijering ).
(d) is described as a development ‘from restricted structural scope
to scope beyond the sentence.’ And finally, (e) and (f) have been
covered in Section . with reference to the development from mean-
ing as part of a sentence to metatextual function: ‘Optionality’ in (e)
refers in particular to the fact that the meaning of an utterance is not
altered if the DM were removed (Frank-Job : ). And not being
part of the meaning of a sentence also implies that DMs do not
contribute to its truth conditionality.
Other features mentioned by students of pragmaticalization are
that DMs do not belong to the categories traditionally considered
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
‘grammatical,’ that they do not become part of a paradigm, and that
“they do not become rule-governed.”
But there are also features that are, as it seems, compatible with a
grammaticalization analysis. One is high frequency of use, which is also
widely held to be a paradigm feature of grammaticalized items. Fur-
thermore, Frank-Job (: ) observes that pragmaticalized items
are characterized by ‘co-occurence in contiguity,’ that is, a DM can
co-occur with a corresponding sentence grammar unit in the same
utterance. To take the example that we had in (), reprinted here as
(), a possible response to (a) could be (b), where the first
instance of well is a DM and the second an adverb. But much the
same can also be found in grammaticalization, for example, when the
English future tense marker be going to co-ocurs with its lexical source,
the verb to go, as in ().
() a. She spoke well.
b. Well, yes, she spoke well.
() He’s going to go home soon.
On account of such features, Ocampo (: ) argues that gram-
maticalization is movement towards syntax and morphology whereas
DMs move “precisely to the opposite end: outside of syntax and
towards discourse,” and Norde (: ) concludes that “movement
towards discourse is genuinely different from movement towards
grammar, and the two are therefore best kept separate.”
11.3.3 The lexicalization hypothesis
In a few studies it is also argued that lexicalization is the main, or one of
the mechanisms leading to the development of DMs. Thus, according
to Wischer (), the early English DM methinks was the result of
both grammaticalization and lexicalization, in that a construction con-
sisting of a recipient subject and an impersonal verb was fossilized,
becoming unproductive and being stored as a whole in the lexicon.
A stronger form of the lexicalization hypothesis is proposed by Fischer
(b: –), who argues that in non-standardized languages, epi-
stemic parentheticals are likely to undergo morphological bonding and
to form one lexical unit in the course of time, turning into lexical units.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. HYPOTHESES ON THE DEVELOPMENT OF DISCOURSE MARKERS
For a critical review of the lexicalization hypothesis, see Brinton
(: –, ).
11.3.4 The cooptation hypothesis
A fourth perspective on the development of DM evolved more recently,
inspired by research on parentheticals, that is, linguistic expressions
that are inserted fairly freely within a sentence, but also at the left or
right periphery of a sentence without forming a participant of the
sentence (see the contributions in Dehé and Kavalova ). In the
framework of Discourse Grammar (Kaltenböck et al. ; Heine et al.
), parentheticals are referred to in short as theticals, their use is
illustrated in the constructed examples of (), where theticals are
printed in bold.
() Constructed examples of theticals
a. This is, if I may say so, not exactly what I had expected.
b. This is, and I ask for your understanding, not exactly what
I had expected.
c. This is, please forgive me if I am impolite, not exactly what
I had expected.
Theticals, like if I may say so in (a), are expressions that are not
syntactic parts of the sentence, they tend to be set off prosodically, for
example by small pauses and/or a special intonation contour, they
express meanings that are not part of the sentence meaning, and their
function is typically metatextual: Rather than to the meaning of the
sentence, they relate to speaker–hearer interaction, speaker attitudes,
and/or the organization of texts.
DMs, such as the ones in (), exhibit much the same features as
theticals. Compare the examples in () with those of (), where
instead of theticals, the DMs well, as it were, and in fact are used.
() Constructed examples of discourse markers
a. This is, well, not exactly what I had expected.
b. This is, as it were, not exactly what I had expected.
c. This is, in fact, not exactly what I had expected.
The units printed in bold in () and () exhibit essentially the
features listed in (). Accordingly, DMs have been treated as a subclass
of parentheticals or theticals.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
() Features typically exhibited by theticals (Kaltenböck et al. )
a. They are not part of the syntax of the sentence.
b. They are as a rule not integrated prosodically.
c. Their meaning is not part of the sentence meaning.
d. They have metatextual functions.
e. They are fairly flexible in their placement.
On account of such observations it is argued by Heine et al. () that
features such as those discussed in Section . do not constitute a
problem for reconstructing the development of DMs—on the contrary,
they are to be expected since DMs arise as theticals. But this approach
raises a number of questions, in particular the following: (a) How do
theticals arise? and (b) how do DMs differ from other theticals, such as
the ones in ()? The answers proposed by Heine et al. () are based
on the hypotheses in ().
() Hypotheses on the development of DMs (Heine ; a)
a. Theticals arise via cooptation.
b. Once coopted, some theticals develop further into DMs via
subsequent grammaticalization.
Cooptation is defined as a discourse strategy (Heine et al. ), more
specifically as a cognitive-communicative operation whereby a chunk
of sentence grammar,² such as a clause, a phrase, a word, or any other
text piece, is deployed for use on the metatextual level of discourse
processing, thereby turning into a thetical, either interpolated in an
utterance or placed at its periphery, or else used as a syntactic standa-
lone. This operation, sometimes perceived as one leading from ‘gram-
matical’ to ‘pragmatic’ uses of a linguistic expression, resembles to
some extent notions such as discoursivization, leading, for example,
from adjective to discourse particle (Ocampo : ), and category
change (Dostie (), both phrased in terms of pragmaticalization (see
Section ..):
The term [pragmaticalization] refers to a process of linguistic change in which a full
lexical item . . . or grammatical item . . . changes category and status and becomes a prag-
matic item, that is, an item which is not fully integrated into the syntactic structure of the
utterance and which has a textual or interpersonal meaning. (Dostie : )
² On the notion of Sentence Grammar vs. Thetical Grammar see Kaltenböck et al.
() and Heine et al. (); see also below.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. HYPOTHESES ON THE DEVELOPMENT OF DISCOURSE MARKERS
In accordance with its new status, the coopted unit is now an autono-
mous information unit, set off from the clause syntactically, prosodi-
cally, and semantically—that is, it is a thetical exhibiting the features in
(). No longer serving a sentence grammatical function, it is a device
of metatextual planning, its main function being that of relating the text
to the situation of discourse. Thus, an English adverb like sadly in (a)
can be coopted as a thetical, as in (b) (cf. Swan : ). As a result,
it no longer modifies the meaning of the sentence or a part of it; rather,
its meaning now relates to the situation of discourse, in this case more
specifically to the attitudes of the speaker.
() a. He sang sadly while she wept.
b. Sadly, Alex lost the election.
A coopted unit may be used once and never again. But it may as well be
used recurrently, and in specific cases this can mark the beginning of a
gradual process of grammaticalization, eventually leading to the rise of
DMs. This process is in accordance with the parameters described in
Chapter and can be sketched as follows:
Context extension: The coopted expression is extended to new con-
texts where it assumes specific functions of discourse processing. These
functions are essentially metatextual, relating to the overall situation of
discourse.
Desemanticization: In the same way as the new discourse functions
are foregrounded, the lexical meaning of the evolving DMs is back-
grounded and may eventually disappear entirely. Thus, in its gramma-
ticalized uses, the DM well that we presented in (b) exhibits little of its
former semantics of an evaluative adverb, now serving exclusively as
functions of discourse organization.
Decategorialization: It is largely restricted to internal decategoriali-
zation, in that the expression loses its ability to be inflected, to take on
derivational morphology or modifiers, and it also loses internal mor-
phological boundaries (see Section ., ()). In this way, collocations
like I think, you know, or any way undergo univerbation, gradually
turning into the invariable DMs I think, you know, and anyway,
respectively. But the external status of the evolving DM is hardly
affected by decategorialization—that is, the grammaticalizing DMs
retain the syntactic independence they inherit from their earlier use
as non-grammaticalized theticals.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
Erosion: DMs may but need not lose part of their phonetic sub-
stance, with one partial exception: Frequently used DMs can lose their
prosodic distinctiveness—to the extent that they are integrated in the
intonation contour of adjacent pieces of discourse. For example, rather
than being prosodically separated, Kaltenböck (: ) found the
DM I think in the following text piece to form one pitch contour unit
with the preceding noun Monday, separating Monday I think from the
preceding tone unit by pitch change and a pause.
() The House knows that this matter may be debated on the
Queen’s speech specifically tomorrow and again on uh Monday
I think. (sb--; Kaltenböck : )
To be sure, not all of the parameters necessarily apply in a given case.
For example, internal decategorialization leading to univerbation is
relevant to DMs such as I think, what else, or you know but not to so,
then, or well, since they consist of one morpheme only.
In sum, on the cooptation hypothesis (Heine et al. ), DMs are
the result of a two-stage history, sketched as in (). Evidence for () is
provided in Heine (: –) and Heine et al. ().
() Hypothesized development of discourse markers (Heine :
–)
(grammaticalization >) cooptation > grammaticalization
In accordance with (), the morphosyntactic, phonological and
semantic features of DMs cannot be reduced to or be explained with
reference to only one of these two mechanisms—be that grammatical-
ization or cooptation: On the one hand, DMs exhibit features such as
the ones listed in Section . that bear witness to their genesis as
(paren)theticals. On the other hand, they also exhibit the features of
grammaticalization just mentioned.
In sum, grammaticalization may and frequently does accompany the
entire history of a DM from its rise to its present usage. Cooptation, by
contrast, involves only a short phase in this development, but this phase
can be held responsible for salient grammatical properties of DMs, such
as the ones listed in ().
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
. CONCLUSIONS
Further reading
Discourse markers have so far remained a fairly marginal field in grammat-
icalization studies. We therefore present a more extensive list of bibliograph-
ical references than in other chapters to familiarize the reader with this field.
In many studies it is argued that discourse markers are largely or entirely the
result of grammaticalization. The following represent paradigm examples of
this hypothesis: Thompson and Mulac (), Traugott (a), Auer
(), Nakayama and Ichihashi-Nakayama (), Günthner (,
), Barth-Weingarten and Couper-Kuhlen (), Traugott and Dasher
(), Günthner and Mutz (), Rhee (), Auer and Günthner (),
Brinton and Traugott (), Brinton (b, ), van Bogaert (), and
Diewald (a; b). A ‘wider’ view of grammaticalization is discussed
principally in Traugott (a), Hansen (b), Lenker (), Traugott
and Dasher (), Brinton and Traugott (: –), Prévost (),
Diewald (a; b), van Bogaert (), and Degand and Evers-Vermeul
(: ). For a critical review of attempts to reduce the development of
discourse markers to grammaticalization, see Heine (a).
The pragmaticalization hypothesis is presented in particular in Erman and
Kotsinas (), Aijmer (), Günthner (), Dostie (), Frank-Job
(), Ocampo (), Brinton (: –), Hansen (: , ), Norde
(: –), Claridge and Arnovick (: –), Arroyo (), and
Beijering (: –). Lexicalization as a factor involved in the development
of discourse markers is argued for by Wischer () and Fischer (b); for
a critical review of the lexicalization hypothesis, see Brinton (: –,
). The cooptation hypothesis is expounded in particular in Heine (),
and Heine et al. (, ). For critical comments on this hypothesis, see
Degand and Evers-Vermeul (: ) and Brinton (: ).
11.4 Conclusions
The four hypotheses discussed in this chapter are seemingly incompat-
ible with one another. As a matter of fact, however, there is a remark-
able similarity between them, in that they all claim, in some form or
another, that the development of DMs involves a development towards
discourse. The differences among them chiefly relate to the question of
whether this development is restricted to grammaticalization of some
kind (the grammaticalization hypothesis), whether it involves but is
not restricted to grammaticalization (the cooptation hypothesis), or
whether it is idiosyncratic to the extent that it needs to be treated
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
DISCOURSE MARKERS
separately and referred to with a distinct term (the pragmaticalization
hypothesis).
The rise of DMs thus requires an account that takes care of the
features which are incompatible with or contradict principles of gram-
maticalization, in particular the ones discussed in Section .. Propo-
nents of the grammaticalization hypothesis have so far not provided
such an account. This is different to the other hypotheses examined.
Proponents of the cooptation hypothesis, in particular, argue that these
are exactly the features that have been identified in research on extra-
clausal constituents, or on parentheticals, or theticals.
Which of the hypotheses surveyed is correct is an empirical issue, in
that the hypotheses can be verified or falsified by means of conclusive
historical evidence. Some kind of evidence is presented in Heine et al.
(), but more detailed reconstruction studies are urgently needed.
What all this research suggests is that DMs will remain a challenge for
students of grammaticalization.
Discussion points
() What do you think is the advantage or disadvantage of defining discourse
markers through a combination of distributional and semantic properties,
instead of defining them purely in terms of their distribution or their
semantics?
() Which properties of discourse markers, do you think, are problematic for
an analysis based on grammaticalization? Which might be more readily
reconcilable?
() Describe the difference between grammaticalization and pragmaticaliza-
tion. Do you think the term pragmaticalization is needed? Give your
reasons for one position or the other.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
12
Summary and conclusion
There is by now a wealth of publications on grammaticalization,
extending from articles to books and contributions to handbooks on
language structure and language change. But around the turn of the
millennium, there was massive criticism challenging grammaticalization
theory (see below under Further reading). A number of weaknesses and
inconsistencies found in previous analyses of grammaticalization were
pointed out, and attention was drawn to areas of research that had been
neglected or ignored in earlier work. In particular, the more general
claims represented in (i) to (iv) have been made time and again:
(i) Not all instances of grammatical change are due to
grammaticalization.
(ii) Grammaticalization is not unidirectional.
(iii) Grammaticalization is not a distinct process.
(iv) ‘Grammaticalization theory’ is not a theory.
In earlier work, the impression was in fact conveyed that any changes
affecting the grammatical structure of a language can be subsumed under
the heading of ‘grammaticalization.’ The challenge in (i) was therefore a
timely reminder that grammatical change also involves factors that are
not covered by grammaticalization, and more research is needed to deal
with these factors in detail. It is especially the later stages of grammatical
change that may pose problems or be elusive to parameters of grammat-
icalization such as the ones discussed in Chapter , as aptly pointed out
especially by Norde ().
With regard to (ii), doubts as to whether grammaticalization is truly
a unidirectional process have in fact been raised throughout the history
of this field of research, and a number of examples contradicting the
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
unidirectionality hypothesis have been identified. We have dealt with
this issue in Sections . and ..
In accordance with (iii) it has been argued that grammaticalization is
not a distinct process because the main parameters (or mechanisms)
characterizing it, that is, desemanticization, decategorialization,
erosion, as well as context extension (see Chapter ) can be also be
identified in other kinds of linguistic change (Newmeyer : –;
Campbell ; Janda ).
There are a couple of reasons why such a position does not seem to
be justified. First, the main task of grammaticalization theory is to
provide explanations for why grammatical forms arise and develop,
and it is these four parameters that have been found to be material to
jointly achieving such explanations (see Section .). Thus, irrespective
of how one wishes to define a ‘distinct process,’ these mechanisms and
their interaction are part of one and the same explanatory framework.
And second, grammaticalization, as conceived here, is first and
foremost a semantic process. Being extended to new contexts, linguistic
expressions lose parts of their meaning but acquire new, grammatical
meanings invited by the context (context extension). And by having
acquired grammatical meanings, these forms tend to increasingly
diverge from their old uses, losing morphosyntactic properties charac-
teristic of the old uses (decategorialization). They tend to be used more
frequently, their use becomes increasingly predictable, and they tend to
lose part of their phonetic substance (erosion). Accordingly, what used
to be a free form having lexical meaning may end up as a bound form
expressing some schematic function. Thus, even if some of their effects
can also be found in other kinds of linguistic change, the four mechan-
isms are needed to account for processes of grammaticalization.
In sum, to the extent that these parameters jointly provide a tool for
describing and explaining the rise, development, and structure of
functional categories through space and time, and to understand why
these categories are structured the way they are, there seems to be
justification for viewing grammaticalization as a distinct process.
Finally, with regard to (iv), it was argued that ‘grammaticalization
theory’ is not a theory (see especially Newmeyer : –; Campbell
). For most researchers this is not, and has never been, an issue, since
their interest is simply in describing grammatical change and the impli-
cation it has for a better understanding of language. Whether their work
deserves or needs to be described as a theory is not considered by them to
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
be of major concern. Other researchers again describe grammaticalization
as a theory in their work; see below under Further reading.
As understood here, grammaticalization theory can be taken to
provide an explanation for certain linguistic phenomena (Narrog and
Heine : ), and its hypotheses can be tested by means of empirical
evidence:
Grammaticalization theory is neither a theory of language nor of language change; its
goal is to describe grammaticalization, that is, the way grammatical forms arise and
develop through space and time, and to explain why they are structured the way they are.
(Heine a: )
Furthermore, as Song (: ) notes, “the fact that some, not other,
changes giving rise to grammaticalization and constraints on what can
and cannot grammaticalize exist does call for the need for grammati-
calization theory” (see also Haspelmath a). We are aware, how-
ever, that there are also a number of other, equally relevant stances on
this issue.
In the course of this book, we have given an overall introduction on
grammaticalization and subsequently focused on a number of issues
that we consider to be particularly relevant for an introduction, even if
the list of issues treated is by no means exhaustive.
The first five chapters discussed the fundamentals of the study of
grammaticalization. In Chapter , we defined grammaticalization as
having both a functional and a formal aspect, with the functional aspect
as the core. Furthermore, we distinguished primary grammaticaliza-
tion, as the change from lexical categories to grammatical (functional)
categories, from secondary grammaticalization, as the extension to
further grammatical categories. In some approaches only the change
from lexical to grammatical is labeled as grammaticalization, and
subsequent changes as something else, but we perceive a directionality
of change that goes beyond the initial stages of grammaticalization.
This directionality of change is characterized by an expansion of scope
and a development towards discourse orientation, as the aggregate of
speaker-, hearer-, and textual orientation. Unidirectionality is an
important feature of grammaticalization. Generally, lexical categories
develop into grammatical categories and not vice versa. Likewise,
within the further extension from one grammatical to the next gram-
matical category, there is also clear directionality. For example, aspec-
tual categories will develop into tense categories, or modal categories
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
into subordinated moods but not vice versa. As was later seen in
Chapter , unidirectionality can be empirically challenged, but chal-
lenges have not been entirely successful if one strictly requires the exact
reversal of grammaticalization as a condition for a successful challenge.
Grammaticalization can be taken as an explanation for a wide range
of grammatical phenomena. Some cases that we reviewed involved
word order, morpheme order, properties of grammatical categories at
certain stages of their development, and polysemy/multifunctionality.
Lastly, we discussed that grammaticalization can essentially be stud-
ied in two ways, namely diachronically, on the basis of historical
language data, and synchronically, through comparison of related
constructions in one language or across a well-defined set of structur-
ally similar or genetically related languages. The diachronic approach is
empirically superior but can only be applied to the small number of the
world’s languages that are historically well-documented, and even with
these languages it faces various challenges. Reconstruction tries to
combine diachronic data, if available, and synchronic data, to fill in
the blanks in the history of a language where historical data are not
available, making use of methodologies established in historical lin-
guistics. Grammaticalization has become an important tool in recon-
struction, and the topic is one of the focal points in this volume that is
emphasized throughout the chapters.
In order to identify a linguistic phenomenon as ‘grammaticalization’
a set of criteria or parameters is needed. Criteria are also instrumental
in identifying in more detail in which respect some linguistic item is
grammaticalized and what stage of the process it has reached. Chapter
provided an overview of such criteria, which have also been proposed
under the terms ‘parameters’ and ‘principles.’ In Chapter , the key
parameters of grammaticalization were discussed in great detail. The
parameters are () context extension, () desemanticization, () de-
categorialization, and () erosion. These parameters are also the tools of
reconstruction, as they allow us to reversely determine the likely former
function and form of a grammatical item, or its predecessor.
Chapter elaborated on the overall directionality of semantic change
in grammaticalization. We first discussed the term ‘bleaching’ that has
been favored by some researchers but has come under fire from others,
and suggested that ‘abstraction’ may be a more suitable term. We then
hypothesized that grammatical categories develop in the direction
of more ‘discourse orientation.’ This term subsumes three tendencies,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
namely orientation towards the speaker, the hearer, and finally text and
discourse itself. The phenomena known as ‘subjectification’ and ‘inter-
subjectification’ in grammaticalization research only relate to minor
parts of this overall tendency, namely the gain in speaker-centered
evaluative meanings in early stages of grammaticalization and the
expression of the speaker’s attention to the addressee’s self-image.
The order between the three tendencies has yet to be fully established
empirically, but evidence suggests that textual, or proper discourse
orientation is a later or last stage phenomenon. Furthermore, while
early stage speaker orientation can be identified with the expression of
speaker attitudes, in later stages that involve increasingly abstract
meanings, speaker deixis takes its place. The claim that later stages of
grammaticalization involve objectification basically refers to the fact
that expressive speaker-oriented meanings tend to fade in secondary
grammaticalization, and more abstract speaker-deictic and textual
functions take its place.
Chapter , the last chapter of the block of ‘fundamentals’ chapters,
discussed gradualness and various concepts of stages in grammatical-
ization. We suggest that grammaticalization is best captured as a
gradual process, even if alternative conceptualizations, for example, as
a gradient, are possible. From a wider perspective, stages can be organ-
ized into grammaticalization chains and also semantic maps. In terms
of semantics, as grammaticalization proceeds in steps and changes, one
linguistic form accumulates more meanings and functions. The reason
is that the original meanings and functions rarely disappear immedi-
ately, but rather, they are prone to persist. The resulting polysemy and
multifunctionality has been variously described in terms of ‘polysemy,’
‘transcategoriality,’ and ‘heterosemy.’ We argued that ‘heterosemy’ is a
particularly useful term. Lastly, we discussed ‘cycles’ in grammaticali-
zation, that is, cases in which the same grammatical category is re-
created when an older expression fades and a new expression for the
same category grammaticalizes, or the older expression is ‘reinforced’
by new material. The best-known case of this is negation. But not all
categories that develop in grammaticalizations are renewals of older
ones. Sometimes, genuine ‘innovation’ takes place.
A second block of chapters, from Chapter to Chapter , dealt with
motivations for and constraints on grammaticalization. Chapter dis-
cussed in some detail mechanisms of, and motivations for grammati-
calization that have been proffered in the research literature. Reanalysis
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
and analogy have been traditionally labeled as ‘mechanisms’ only, and
both have been subject to considerable controversy. In contrast, the
language system, communicative needs, inferences and context, cogni-
tion and conceptualization, frequency of use, and discourse have been
mostly identified as ‘motivations,’ but some of them also alternatively
as ‘mechanisms’ or ‘filters.’ None of them has been undisputed in its
relevance for grammaticalization, and which one of them is considered
as decisive heavily depends on the theoretical stance taken towards the
study of grammaticalization. We suggest that grammaticalization can
have multiple motivations. None of them needs to exclude the others.
However, we consider language structure as guiding and constraining
grammaticalization rather than motivating it.
‘Borrowing’ is traditionally one of the three mechanisms of gram-
matical change. Also traditionally, ‘external’ language change through
language contact is distinguished from internal change. In Chapter
we took a grammaticalization perspective on so-called ‘external’ lan-
guage change and revealed that the replication of structures of a contact
language (the ‘model language’) is quite common and often happens
via the same mechanisms through which purely ‘language-internal’
grammaticalization takes place. In fact, it is often not easy to decide
whether a grammaticalization was motivated ‘internally’ or ‘externally.’
We suggested a number of criteria for grammaticalization through
language contact and for deciding which language was the model lan-
guage. These are also important tools for reconstruction, since language
change through language contact is quite common in the languages of
the world.
Finally, in Chapter , we first returned to the idea that language
structures guide or constrain grammaticalization. We provided some
examples for extant language structures guiding, or serving as a model
for, newly developing structures. This can be most easily observed in
the morphological and phonological structures of grammaticalization,
such as so-called ‘isolating’ languages grammaticalizing isolating struc-
tures, but also holds for functional domains. We can see this in the fact
that languages tend to bolster their extant grammatical categories
rather than constantly developing new ones. Alternatively, principles
of grammaticalization can also bear on the structural features of a
language. Well-known examples are the so-called suffixing preference,
and morpheme order in complex words. Furthermore, grammaticali-
zation is a decisive factor in how specific grammatical categories are
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
expressed in a specific language. And, it almost trivially contributes to
the formation of new analytic language structures to replace older
synthetic ones.
Chapter is a standalone chapter that discussed a number of theo-
retical frameworks that have adopted the study of grammaticalization
and have thereby developed very specific perspectives on grammatical-
ization. We saw that in generative grammar, grammaticalization is
understood in terms of upward reanalysis in hierarchical clause struc-
ture, in consonance with our idea that grammaticalization implies
scope expansion and economy. In Functional Discourse Grammar as
well, although the theoretical background is entirely different, gram-
maticalization is also conceptualized as a development upward (or to
the outside of) hierarchical clause structure. Functional Discourse
Grammar was the first theoretical framework to make this claim.
Variationist Theory differs from these frameworks in that it does not
apply a different analysis of grammar to the phenomenon, but focuses
on important factors in grammaticalization that traditionally tend to be
overlooked. In particular, it views grammaticalization in terms of
competition between variants of expression for the same functional
domain, and analyzes the use contexts of each variable. Cognitive
Grammar takes a unique synchronic view of grammaticalization
entirely in terms of semantic change, specifically ‘subjectification.’
This subjectification is highly compatible with what we consider as
speaker orientation in terms of speaker deixis, and contrasts with the
more common view of subjectification in terms of the expression of the
speaker’s attitude. Lastly, we examined the relationship between gram-
maticalization and Construction Grammar, which is a challenging one,
since Construction Grammar essentially denies the distinction between
the lexicon and grammar, which is definitory for grammaticalization.
We discussed a few attempts to reconcile grammaticalization theory
and Construction Grammar.
Chapters and constituted the part of this book that dealt
with issues ‘beyond’ grammaticalization. These were mainly lexicaliza-
tion, degrammaticalization, and exaptation (regrammaticalization) in
Chapter , and the development of discourse markers, also known as
pragmaticalization in Chapter . Degrammaticalization and the devel-
opment of discourse markers, in particular, have been posited as
challenges to grammaticalization theory. Degrammaticalization, in a
strict sense as the mirror image of grammaticalization, hardly exists or
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUMMARY AND CONCLUSION
does not exist at all, as mentioned in the remarks on Chapter above.
On the other hand, partial developments in the opposite direction of
grammaticalization, especially purely formal (morphological) ones, do
exist, and the less restrictive we are in our definition of degrammatica-
lization, the more ‘degrammaticalizations’ can be identified. Exaptation
(regrammaticalization) has received far less attention than degramma-
ticalization, but may in fact be a more common and even a more
regular progress. The development of discourse markers is a process
that is in principle different from grammaticalization but incorporates
elements of grammaticalization.
While grammaticalization as such is a linguistic phenomenon, we
have outlined grammaticalization in this book as a theory. We argued
in particular that it can uniquely explain a variety of language phenom-
ena. Grammaticalization can be understood as a ‘mid-level’ explanation
in that grammaticalization itself is driven by deeper motivations such as
cognition and communication. However, it is a useful explanation
because linguists usually do not have direct access to those deeper
factors, and even if they do, such factors may play out in many
alternative ways, not only those that are actually realized in a grammar.
Further reading
Around the turn of the millennium, the question of whether grammaticaliza-
tion can be defined as a distinct process has generated some research activity;
see especially Newmeyer (), Joseph (), Campbell (), Campbell and
Janda (), Janda (), and Norde (, , , ). This activity
has also identified a range of grammatical changes claimed to be at variance
with the unidirectionality hypothesis; see also Joseph and Janda (), Ramat
(), Frajzyngier (), and especially Newmeyer (: ff.)
While the term ‘grammaticalization theory’ is avoided by a number of
researchers, others, including Haspelmath (a: ), Heine (a), Song
(), Noël (), Gisborne and Patten (), Narrog and Heine (,
), Reinöhl and Himmelmann (), and Heine (a), have used it
affirmatively.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
Glossary
The following list presents terms which are in some way or another relevant
for describing phenomena of grammaticalization as discussed in this book.
The list is based on more detailed descriptions in established dictionaries of
linguistics and in the literature on grammaticalization. It only aims to provide
brief characterizations of common uses of the terms, and not exhaustive
descriptions or definitions. For more details and bibliographical references
see the main text of this book as well as the references provided therein.
ability, see PI-possibility
ablative: ‘(away) from’; also ‘from above/below/inside’; case indicating a
spatial source participant; direction from, source; temporal ‘from,’ ‘since’;
indicating a temporal (source) participant (e.g. from morning to evening)
affix: a bound morpheme added to a root or a stem. See also prefix, suffix
agent: e.g. ‘by’; semantic role of a participant that instigates or performs the
action described by the main verb
agglutinating: a mechanism of word formation where a grammatical item
typically having a separate form and a separate function is affixed to a root
morpheme and to other grammatical items, e.g. Swahili tumefika ‘we have
arrived’ (<tu- ‘we’, me- perfect, fika ‘arrive’). Cf. Chapter . See also inflec-
tional, isolating
agglutinating language: a language using mainly or exclusively agglutinating
word formation. Cf. Chapter . See also agglutinating
agreement: correspondence of person, number, gender, or class etc. between
two or more sentence elements
analogical leveling: elimination of alternations in a morphological paradigm
in analogy to the more common form pattern
analogy: a change in a linguistic expression under the influence of semantic-
ally, formally, or functionally related expressions, resulting in some respect
from an attempt to make some expression more similar to other expres-
sions. Cf. Chapter .
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
analytic language: language in which each meaning or function is expressed
by a clearly separate word or morpheme. Cf. Chapter .. Opposite of
synthetic language
andative: ‘motion thither’; verbal category for a movement away from the
speaker or deictic center; also called itive, translocative. See also venitive
anterior, see perfect
antigrammaticalization: a change leading from the endpoint to the starting
point of a grammaticalization process, exhibiting in reverse order the same
intermediate stages as the latter. Cf. Chapter .
applicative: marker or process that increases the valency of a verb by one
participant. Depending on the language concerned, the marker or process
may express a range of functions such as ablative, benefactive, comitative,
goal, locative, patient, and recipient
argument: a noun phrase whose function is determined by a verb
automation: a process whereby a less schematic construction gradually
becomes alienated from its more schematic parent construction
autonomy, syntactic: the degree of syntactic integration of a given item. Cf.
Chapters ,
auxiliary (verb): a verb used with a main verb to express grammatical func-
tions of tense, aspect, modality, etc.
avertive: a category of verbal aspect denoting ‘almost, nearly’, that is, an
action or event that was on the verge of taking place but did not take
place. See also proximative
benefactive: ‘for,’ ‘for the benefit of ’; () semantic role or case of a participant
in whose favor an action or event takes place; () construction indicating
that the action of the main verb is for the benefit of or on behalf of some
participant
bleaching: loss of concrete aspects of the meaning of a linguistic item, leaving
behind a more generalized or abstract meaning. Cf. Chapter .
bondedness: the intimacy with which a sign is connected with another sign
to which it bears a syntagmatic relation. Cf. Chapters ,
borrowing: a process whereby phonological substance or form–meaning units
(morphemes or combinations of them) are transferred from one language to
another. Cf. Chapter . See also replication
boundness: a particular form may be more or less attached to another form.
For example, affixes are tighter bound to word stems than clitics. Cf.
Chapters ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
bridging context: involves the extension of a linguistic form to a new context
which invites a new meaning (the target meaning), co-existing with the old
meaning (the source meaning)—thereby giving rise to ambiguity in that
context. Cf. Chapters , .. See also context extension, switch context
causative: ‘cause to be,’ ‘cause to do’; voice category which adds an external
agent to the argument structure
cause: semantic role of circumstance leading to, or triggering a certain result.
Linguistically often expressed through case, adpositional phrases such as
‘because of,’ ‘since,’ ‘on account of,’ or ‘therefore.’ See also reason
chain of grammaticalization: a continuum of interlocking uses of a gram-
maticalizing form extending from the source to the target form
change of state: ‘become,’ ‘turn into’; inchoative, ingressive. See also
resultative
chunking the fusion of linguistic forms that frequently co-occur. See also
ritualization, univerbation
classifier: classificatory formative; a general term referring to the specific
system of formatives that consists of quantifiers, repeaters, and noun
classifiers proper
cline (of grammaticalization): usually refers to morphological change accom-
panying grammaticalization from lexical word to affix; cf. Chapter ..
Sometimes used as a synonym for grammaticalization chain. Cf.
Chapter .
clitic: a morpheme which is more bound than a word but less bound than an
affix. See also proclitic
coalescence: the gradual increase in boundness or tightness of a combination
of linguistic elements. Cf. Chapters ., .. See also chunking, morpholo-
gization, univerbation
comitative: ‘(together) with’; semantic role, in some languages a case
comparative: . A construction in which two entities are compared to each
other with one serving as the standard of comparison; . In some languages
a morphological form of adjectives and adverbs used for comparison
complementizer: ‘that’; word or dependent morpheme introducing comple-
ment clauses
compositionality, morphological: the degree of formal transparency or opa-
city of a compositional linguistic item
compositionality, semantic: the extent to which the meaning of an expres-
sion can be derived or constructed from its component parts
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
concessive: clause construction of two clauses that are in a semantic relation-
ship of contradiction or contrast; ‘despite the fact that,’ ‘even though’ or
‘but’
conditional: clause construction of two clauses that are in a semantic rela-
tionship in which the proposition of one clause is a condition for the
proposition in the other clause; ‘if’
conjunction: e.g. ‘and,’ ‘or,’ ‘but,’ etc.; word used for conjoining clauses; clause
connective, sentence connective
construction: a symbolic pairing of form and meaning at different levels of
abstractness and atomicity that can be organized in a taxonomy
constructional change: (i) a process involving a sequence of changes in the
form and meaning poles of a construction; (ii) a process involving changes in
form or meaning but not both. Cf. constructionalization
constructionalization: a sequence of changes in the form and meaning poles
of a construction whereby new formal configurations come to serve par-
ticular functions, and to encode new meanings. Cf. Chapter .. See also
constructional change
contact-induced grammaticalization: a grammaticalization process that
has been triggered or accelerated by language contact. Cf. Chapter
context extension: use of a linguistic expression in new contexts triggering
the rise of a new meaning, eventually also of a new functional category. Cf.
Chapter .. See also bridging context, switch context, conventionaliza-
tion stage, extension
conventionalization stage: a linguistic form which arose via (>) context
extension and no longer needs to be supported by the context which gave
rise to it. Cf. bridging context, switch context
cooptation: a productive operation whereby a chunk of sentence grammar,
such as a word, a phrase, a reduced clause, a full clause, or some other piece
of text, is deployed for use on the metatextual level of discourse processing,
thereby turning into a thetical. Cf. Chapter ... See also thetical
copula: ‘be’; predicate used in propositions of the type ‘X is (a) Y’; identifying
copula, classifying copula. See also locative copula
creole: a language that has evolved as a rule from a (>) pidgin learned by
children as their first language, thereby acquiring an extended lexicon and
grammar. As a result, creoles are in a number of ways indistinguishable
from other languages. Cf. pidgin
cycle, also morphological cycle: a cyclic form of evolution, typically involving
morphosyntactic or semantic-pragmatic material, leading to the renewal of
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
a grammatical category of the same kind. Cf. Chapter .. See also
reinforcement, renewal, spiral
dative: case that typically marks a third participant, such as a recipient
de-volitive: a grammaticalized form derived from a verb meaning ‘want’
debonding: a composite change whereby a bound morpheme in a specific
linguistic context becomes a free morpheme. Cf. Chapter .
decategorialization: loss of morphological and/or syntactic properties that a
linguistic expression experiences in the course of grammaticalization. Cf.
Chapter .
definite: ‘the’; definite article; nominal determiner. Cf. indefinite
degrammaticalization: a composite change whereby in a specific context a
grammatical form gains in autonomy and/or substance on at least two of
the three levels semantics, morphosyntax, and phonology. Cf. Chapter .
degrammation: a composite change whereby in a specific context a function
word is reanalyzed as a member of a major word class, gaining both
semantic substance and morphosyntactic properties characteristic of that
word class. Cf. Chapter .
deinflectionalization: a composite change whereby in a specific linguistic
context an inflectional affix gains a new function and shifts to a less
bound type. Cf. Chapter .
delexicalization: the reverse of lexicalization, where speakers re-analyze
words as the concatenation or fusion of morphemes that often are not
their actual source; e.g. ham-burger from hamburger (from German Ham-
burger ‘[something] from Hamburg’). Cf. Chapter .. See also lexicalization
demonstrative: ‘this/these,’ ‘that/those’; nominal determiner or pronoun
demorphologization: either the degradation of morphological patterns to
phonology, or from a more bound to a more independent morpheme
such as a clitic or independent word. Cf. phonogenesis, degrammaticaliza-
tion. Cf. Chapter .
deontic (modality), see D-necessity
desemanticization (or bleaching): loss of semantic properties that a
linguistic expression experiences in the course of grammaticalization.
Cf. Chapter .
discourse marker (or pragmatic marker): invariable expressions which are
semantically and syntactically independent from their environment, typic-
ally set off prosodically from the rest of the utterance, and their function is
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
metatextual, relating a text to the situation of discourse, e.g. anyway, indeed,
in fact, now, well, etc.
discourse orientation: orientation towards speaker and hearer, and towards
organizing speech or discourse itself. Usually increases in grammatical-
ization. Cf. Chapter .. See also speaker orientation, hearer orientation,
textual orientation
discursively primary, see primary, discursively
discursively secondary, see secondary, discursively
distal: ‘far away’; spatial distance, e.g. expressed by demonstratives. Cf.
proximal
divergence: a lexical form undergoes grammaticalization but survives as an
autonomous lexical element, occurring side by side with its grammatical-
ized variant. Cf. Chapter
dual: number category referring to no more and no less than two items
durative: ‘keep doing’; verbal aspect
elative: ‘too,’ as in too much, too big, etc.
emphatic: expression of emphasis or contrast
endocentric (construction): a syntactic construction which belongs to the
same form class/category (i.e. shows the same distribution) as one or
more of its constituents
epistemic (modality): type of modality that denotes a necessity or possibility
based on the speaker’s knowledge and beliefs about the state of affairs
expressed in the proposition
ergative: case marking the agent argument of a transitive verb in ergative
languages
erosion (or phonetic reduction): loss of phonetic features that a linguistic
expression experiences in the course of grammaticalization
evidential: category, typically of the verb, that refers to information about the
evidence on which a report is based
evidential, inferred: category, typically of the verb, indicating information
acquired through inductive or deductive inference
evidential, reported: category, typically of the verb, indicating that the evi-
dence for the proposition is based on hearsay
exaptation: as used in grammaticalization theory, the term stands for a
change whereby a grammatical category that has lost its semantic and/or
formal distinctiveness is put to new functional uses. See Chapter .
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
exclusive: ‘we excluding you’; a distinction made typically in first person
plural pronouns, which excludes the hearer. Cf. inclusive
explanation, external: an explanation where a set of facts is derived as a
consequence of principles outside the domain concerned, which typically—
though not necessarily—is that of grammar. This is the case, for example,
when linguistic phenomena are explained by means of extra-linguistic
phenomena
explanation, internal: an explanation where a set of facts is considered to be a
consequence of the deductive structure of a particular theory of grammar,
or where a given phenomenon is explained with reference to other phe-
nomena belonging to the same domain
extension: a mechanism which results in changes in the surface manifestation
of a pattern which does not involve immediate or intrinsic modification of
underlying structure. See also context extension
external explanation, see explanation, external
focus: grammatical function ‘highlighting’ some participant of a sentence,
presenting that participant as new information
frequency, see token frequency, type frequency
future: ‘will,’ ‘be going to’; tense indicating that the speaker predicts an event
to occur later than the time of the speech event
future, de-volitive: a future tense category historically derived from a verb for
‘want’
grammaticalization, paired, see paired grammaticalization
habitual: ‘do habitually’; verbal aspect for an event occurring habitually or
usually and on different occasions
hearer orientation: orientation of a linguistic expression towards the
addressee; e.g. honorifics. It may increase in the grammaticalization of
some categories. Especially in advanced grammaticalization, it becomes
hearer deixis (anchoring in the second person); e.g. imperatives. Cf.
Chapter .. See also speaker orientation, discourse orientation
heterosemy: obtains when within a single language two or more meanings
that are historically related, in the sense of deriving from the same ultimate
source, are borne by reflexes of the common source element that belong in
different morphosyntactic categories
honorific: marker of honorific reference or deference
imperative: form of the verb that expresses directives, commands, and
requests
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
imperfective: verbal aspect denoting that an event is viewed as unbounded
temporally. See also perfective
impersonal: construction without openly expressed agent; the agent may be
not identifiable or simply irrelevant
implicature, conversational: any meaning implied or expressed by, and
inferred or understood from, the utterance of a sentence which is meant
without being part of what is strictly said. It is one of the manifestations of
inferences. See also inference
inchoative: . phasal aspect marking the beginning of a state or event. .
Stands for spontaneous action not depending on a volitional actor
inclusive: ‘we including you’; a distinction typically made within first person
plural pronouns which includes the hearer. See also exclusive
indefinite: ‘a, an’; indefinite article; nominal determiner. See also definite
indefinite pronoun: ‘something,’ ‘someone,’ etc., including impersonal pro-
nouns, like French on or German man
inference: conclusion drawn on the basis of a set of propositions, either
logically or from context or from everyday reasoning. See also implicature,
conversational
inflection: consistent or even obligatory change of word form corresponding
to specific syntactic and semantic functions of a word, typically found in
verbs, adjectives, and nouns, e.g. keep vs. kept, goose vs. geese). Cf.
Chapter . See also agglutinating, isolating
inflectional language: a language using mainly inflectional word formation.
Cf. Chapter . See also inflectional
ingressive: synonymous with inchoative
innovation: a mechanism whereby a new grammatical form is created for
which there was no earlier conventionalized equivalent in the language
concerned. See also renewal
instrument(al): ‘with,’ ‘by means of ’; semantic role, and in some languages a
case indicating the function of a participant as an instrument
integrity: the amount of semantic and phonological substance needed for a
given form to maintain its identity
intensifier: ‘-self’, as in The king himself, The king did it himself; frequently
referred to as emphatic reflexive, identifier
intention, see B-necessity
interjection: wow!, Boy!, a typically short emotive exclamation
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
internal explanation, see explanation, internal
intersubjectification: increase in intersubjectivity . See also subjectification
intersubjectivity: . In philosophy, psychology, and part of linguistics, shared
ground between speaker/writer and hearer/reader; . In some grammat-
icalization studies, attention towards the addressee’s face needs. See also
subjectivity
isolating (or analytic): a mechanism of word formation where each grammat-
ical category is represented by a separate word, that is, where words have no
internal morphological structure. Cf. Chapter . See also agglutinating,
inflectional
isolating language: a language using mainly or exclusively isolating word
formation. Cf. Chapter . See also isolating
iterative (or repetitive): ‘do repeatedly’; aspectual category for action that is
done repeatedly or an event occuring repeatedly
lateral shift: a diachronic change from one category to another that does not
involve change towards a more (or less) grammaticalized category on a
grammaticalization chain. See also transcategorization
layering: new layers of structure arising via grammaticalization may coexist
with older layers within a broad functional domain of a language, with the
two, or more kinds of layers interacting with one another. Cf. Chapter
lexicalization: a process by which linguistic units enter the lexicon. More
specifically, it is a change whereby in certain linguistic contexts speakers
use a syntactic construction or word formation as a new contentful form
with formal and semantic properties that are not completely derivable or
predictable from the constituents of the construction or the word formation
pattern. Cf. Chapter .
locative: semantic role, or in some languages, case indicating a locative
participant
locative copula: ‘be at,’ ‘be somewhere’; predicate used in propositions of the
type ‘X is (located) at Y’
loss-and-gain model: a model of grammatical change according to which
meaning change in grammaticalization involves both losses of old material
and gains of new materials. Cf. Chapter .
map, semantic, see semantic map
mechanism: a particular constellation of factors causing one state to change
into another
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
metatextual: the function of a marker, such as a discourse marker, as being
anchored in and relating the meaning of a sentence to the situation of
discourse. In more general terms, a metatextual marker refers to a state-
ment whose topic is the text itself
model language: a language that provides the source for a transfer of lin-
guistic material from one language to another. Cf. Chapter . See also
replica language
morpheme: a minimal form unit with a meaning. See also phoneme
morphological cycle, see cycle
morphologization (in grammaticalization): a diachronic process leading
from an independent word to a bound morpheme. The term relates to
the core of the potential morphological development accompanying gram-
maticalization. See also coalescence
morphology: the branch of grammar devoted to the structure of words
morphosyntax: the branch of grammar involving simultaneously morph-
ology and syntax
B-necessity (Intention): ‘intend to’; boulomaic necessity
D-necessity (Obligation): ‘have to,’ ‘should,’ ‘must’; the validity of a propos-
ition is relativized as being contingent on an obligation or societal norms;
deontic necessity
E-necessity (Probability): ‘it is likely that’; the validity of a proposition is
relativized as being contingent on the speaker’s assessment of its likelihood;
the likelihood is high; epistemic necessity
negation: ‘not,’ ‘no’
neuter: a gender category that is neither feminine nor masculine
nominalizer: marker used to derive nouns from verbs, adjectives, and other
categories; for example, English ‐er, ‐ization
non-specific marker: use of an indefinite article to refer to a participant in
discourse whose referential identity neither the hearer nor the speaker
knows or cares to know. See also specific marker
obligation, see D-necessity
obligatorification: the use of a linguistic form in some specific context
becomes obligatory rather than being optional. Cf. Chapter
obligatoriness: a particular expression can be required to differing extents,
being required in some environments. Cf. Chapters ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
paired grammaticalization: a situation where two neighboring languages in
contact share two grammaticalization processes for the same general gram-
matical function. Cf. Chapter
paradigmaticity: the formal and semantic integration both of a paradigm as a
whole and of a single subcategory into the paradigm of its generic category.
Cf. Chapter .
parenthetical: a linguistic expression that is inserted fairly freely within a
sentence, but also at the left or right periphery of a sentence, without being
integrated in the sentence. See also thetical
passive: a grammatical voice in which the agent is suppressed or demoted
and the event is often centered on the perspective of the recipient or patient
of the verb
past: tense indicating that an event or state is temporally located before the
moment of speech
patient: semantic role of a participant that is the undergoer of the action
denoted by the verb; direct object
perfect: verbal aspect indicating that a past event is relevant to the situation at
reference time; anterior
perfective: verbal aspect indicating that an event is viewed as bounded
temporally. See also imperfective
persistence: relates the meaning and function of a grammatical form to its
history as a lexical form
phoneme: a minimal distinctive sound unit. See also morpheme
phonetic reduction, see erosion
phonogenesis: a process whereby erstwhile morphemes are degraded into
purely phonological elements of a word
phonologization: a process whereby allophones of one phoneme develop into
phonemes of their own
phonology: the branch of grammar devoted to the sound system of a
language
pidgin: a language that has a reduced lexicon and grammatical structure,
resulting typically from extended contact between groups of people with
no language in common
plural: number category indicating two or more referents. See also dual
polysemy: one linguistic form having two or more related meanings, that is,
there is significant overlap in semantic content between the meanings. See
also heterosemy, homonymy
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
polysemy copying: a process whereby in situations of language contact a
combination of related meanings associated with one form is copied analogic-
ally from one language (the model language) into another language (the
replica language), that is, it is ‘translated’ in another language. Cf. Chapter .
possibility, see C-, E-possibility
C-possibility (root possibility): ‘can’; the validity of a proposition is relativ-
ized as being contingent on enabling circumstances and conditions, e.g.
You can get that book in our bookstore; also known as circumstantial or
situational possibility
E-possibility: ‘it is possible that’; the validity of a proposition is relativized as
being contingent on the speaker’s assessment of its likelihood; the likeli-
hood is not high; epistemic possibility. See also potential
PI-possibility (ability): ‘can, be able’ (participant-internal possibility)
pragmatic marker, see discourse marker
pragmatic strengthening: a process whereby in a situation of semantic
ambiguity of an utterance pragmatic inferences about that utterance serve
to strengthen one of the available interpretations. The result can be that,
over time, an old meaning is replaced by a new one. Cf. Chapter .
pragmaticalization: a process leading to the rise of (>) discourse markers
prefix: an affix which precedes a root or a stem. See also affix, suffix
present: tense that is simultaneous or overlapping with the moment of speech
presentative marker: use of an indefinite article to refer to a new participant
in discourse presumed to be unknown to the hearer and this participant is
taken up as definite in subsequent discourse. See also non-specific marker,
specific marker
primary, discursively: a linguistic form, such as a noun or a demonstrative,
that can be focalized or addressed in discourse, e.g. It is this that I wanted to
tell you. See also secondary, discursively*
primary grammaticalization: the development from lexical to grammatical
meanings or forms. Cf. Chapter .. See also secondary grammaticalization
probability, see E-necessity
procedural meaning: encodes information about computations rather than
about representations, that is, on how an utterance is to be processed rather
than conceptually expressed
proclitic: a clitic which is attached to the beginning of a word. See also clitic
progressive: ‘be doing’; aspect indicating that an action or event in progress
at reference time
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
proximative: a category of verbal aspect denoting ‘be about to,’ ‘be on the
verge of doing.’ See also avertive
purpose: ‘in order to,’ ‘so that’; () semantic role for the purpose of an action.
() Relationship between two clauses in which one clause indicates the
purpose of the event or action described in the other clause
reanalysis: a change that assigns a new underlying structure to a surface
sequence without overtly modifying that sequence
reason: . In complex clauses, proposition adduced by a speaker to support a
conclusion in the other clause. Can be marked in English with conjunctions
such as because or since. . Semantic role used synonymously with cause
recipient: ‘to’; semantic role of a typically human goal or recipient; indirect
object, dative
reflexive: ‘self ’, as in I saw myself in the mirror; where the patient is the same
referent as the agent (i.e. the two have identical reference)
reinforcement: a mechanism whereby a grammatical marker A is supported
by some additional expression B ‘strengthening’ the meaning of A. See also
cycle, renewal
relative pronoun: ‘who,’ ‘which,’ ‘that’; marker introducing relative clauses
renewal: a mechanism whereby an existing expression of a grammatical
category is replaced by a new expression, with the result that the new
expression assumes a function that is similar to the old one. See also
cycle, innovation, reinforcement
renovation, see substitution
repetitive, see iterative
replica language: a language that is the target of a transfer of semantic or
structural linguistic material from one language to another. See also model
language
replication: a contact-induced process whereby semantic and/or structural
linguistic material is transferred from one language to another. See also
borrowing
reported evidential, see evidential, reported
resultative: ‘having reached a new state’; verbal aspect. See also change-of-
state
retraction: loss of a further advanced grammatical function that a linguistic
item developed, with only the less advanced function(s) remaining
ritualization: term from ethnology, referring to the development of rituals
and routines through frequent repetition of the same behavior; in
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
linguistics, seen as the driving force behind grammaticalization by scholars
who emphasize the language processing aspect of grammaticalization. Cf.
Chapter .. See also chunking, routinization
root possibility, see C-possibility
routinization: a process by which a linguistic sign or sequence is made to be
used more frequently, thereby progressively ruling out alternatives and
choices. See also ritualization
scope: a logical-semantic concept, defined by an asymmetric (exocentric)
relationship between an operator and an operand. By the scope of an
operator, such as a connective or a quantifier, is meant that part of a
formula which is within the domain of operation of the operator
scope, structural: the structural size of the construction which a grammatical
form helps to form; specific to Lehmann’s concept of grammaticalization
and fundamentally different from scope (above)
secondary, discursively: a linguistic form, such as an article (the, a), that
cannot be focalized or addressed in discourse. See also primary,
discursively
secondary grammaticalization: the development from grammatical mean-
ings or forms to even more grammaticalized meanings or forms. Cf.
Chapter .. See also primary grammaticalization
semantic map: two- (potentially even three-)dimensional representations of
degree of similarity of a specific range of linguistic categories
serial verb: a verb in a construction in which two or more verbs sharing the
same subject or agent are juxtaposed and together express the equivalent of
a verb phrase with just one verb in other languages
serial verb construction: a monoclausal construction consisting of multiple
independent verbs with no element linking them and with no predicate–
argument relation between the verb
source, semantic role; see ablative
source (of grammaticalization): a linguistic form or construction giving rise
to a process of grammaticalization. See also target (of grammaticalization)
speaker orientation: orientation of a linguistic expression towards the
speaker; e.g. attitudinal use of adverbs and adjectives, said to increase in
grammaticalization. Especially in later stages of grammaticalization, it takes
the form of speaker deixis (e.g. tense). Cf. Chapter .. See also hearer
orientation, discourse orientation
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
specialization: the narrowing of choices that characterizes an emergent gram-
matical construction
specific marker: use of an indefinite article to refer to a participant in
discourse known to the speaker but presumed to be unknown to the hearer;
See also indefinite, non-specific marker
speech act orientation: term formerly used for discourse orientation
speech community: a group of individuals who share a set of norms regard-
ing the use of a language
spiral: elaboration on the concept of cycle. While a ‘cycle’ rests on a metaphor
of exact renewal of a specific linguistic category, or state of a language, the
metaphor of a spiral emphasizes that the renewed category is similar but
not identical to the one it replaced. Cf. Chapter .
subjectification: . Traugottian sense: a process whereby a meaning becomes
increasingly based in the speaker’s subjective belief state or attitude towards the
proposition; related to subjectivity ; . Langackerian (Cognitive Grammar)
sense: the realignment of some relationship from the objective to the subjective
axis, or a gradual process of progressive attenuation in which an objective
relationship fades away, leaving behind a subjective relationship that was
originally immanent in it; related to subjectivity . See also speaker orientation
subjectivity: in linguistics: . Expression of the speaker’s attitude or commit-
ment to the content of a proposition. . Speaker deixis, in particular implicit
speaker deixis (e.g. spatial expressions such as here, there; tense). . Refer-
ring to information that is not shared with the addressee. Cf. Chapter ..
See also intersubjectivity, subjectification
subordinator: general marker introducing adverbial clauses, having a range
of adverbial functions, such as presenting locative, temporal, causal, and/or
conditional functions
substitution (or renovation): a grammatical form A falls into disuse and is
replaced by a largely equivalent form B. See also cycle, renewal
suffix: an affix which follows a root or a stem. See also affix, prefix
switch context: the use of a linguistic form in a new context highlighting a
new meaning (the target meaning) which is incompatible with the earlier
meaning of that form (the source meaning). Cf. Chapter .. See also
bridging context
syntactic variability, see variability
syntagmatic: refers to the relationship between linguistic items in linear
sequence, e.g. word formation or sentence formation. Contrasts with the
term paradigmatic, which refers to the relationship between linguistic items
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
that can replace each other in a specific position in a linear sequence. Not
identical with syntactic. Term used by Lehmann. Cf. Chapter .
synthetic language: language in which two or more meanings and grammat-
ical functions are often fused into one morpheme, and morphemes are
often not separated by clear boundaries. Opposite of analytic language. Cf.
Chapter .
target (of grammaticalization): a linguistic form or construction being the result
of a grammaticalization process. See also source (of grammaticalization)
thetical: a linguistic expression that is syntactically separate, typically set off
prosodically, and its meaning relates to speaker–hearer interaction, speaker
attitudes, and/or the organization of texts, rather than to the meaning of the
sentence it is associated with. See also parenthetical
token frequency (in grammaticalization): refers to the frequency with which
an individual form may be used and becomes relevant when grammatical-
ization has kicked in. Cf. Chapter .. See also type frequency
topic: the theme of a sentence, that is, what the sentence is about, as ‘X’ is in
‘as for X’
transcategoriality: one and the same kind of linguistic form is used simul-
taneously on two or more different planes of linguistic organization based
on some regular pattern. English round is an example of transcategoriality,
occurring as an adjective (a round table), a preposition (He walked round
the corner), a verb (The boat will round the buoy soon), an adverb (I went
round to the shop), or a noun (It’s my round). Cf. Chapter .
transcategorization: a shift of a category from one word class to another; e.g.
considering from verb participle to preposition. See also lateral shift
type frequency (in grammaticalization): the type and number of contexts in
which a linguistic form can be used, e.g. the number of verbs with which an
aspect marker co-occurs. Cf. Chapter .. See also token frequency
uniformitarian principle, or uniformitarianism: a hypothesis according to
which the same kind of linguistic phenomena and changes that we observe
in the present must have existed at some time in the past which is not
accessible to direct observation, unless there is empirical evidence to the
contrary
univerbation: a process whereby a collocation of two or more words or
morphemes loses its internal morphological boundaries and merges into
a new, invariable word, e.g. English forget-me-not, anyway). Cf. Chapter ..
See also chunking, coalescence
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
GLOSSARY
variability, syntactic: the flexibility of a linguistic item, that is, the number of
syntactic slots that an element may occupy. For example, items moving
towards bound status lose in syntactic variability; syntagmatic variability
venitive: ‘motion hither,’ ‘motion towards’; verbal category for a movement
towards the speaker or deictic center; ventive. See also andative
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
References
Aaron, Jessi Elena and Rena Torres Cacoullos. . Quantitative measures of
subjectification: A variationist study of Spanish salir(se)*. Cognitive Lin-
guistics /, –.
Abraham, Werner. . The grammaticalization of the German modal
particles. In: Traugott, Elizabeth C. and Bernd Heine (eds.) Approaches to
Grammaticalization, Volume , –. Amsterdam: Benjamins.
Abraham, Werner. . Einleitung zum Thema dieses Bandes. Grammatika-
lisierung und Reanalyse: Einander ausschließende oder ergänzende
Begriffe? Folia Linguistica Historica /–, –.
Abraham, Werner. . How far does semantic bleaching go? About gram-
maticalization that does not terminate in functional categories. In: Faar-
lund, Jan Terje (ed.) Grammatical Relations in Change, –. Amsterdam:
Benjamins.
Abraham, Werner. . Deutsche Syntax im Sprachenverleich: Grundlegung
einer typologischen Syntax des Deutschen, . Tubingen: Stauffenburg.
Abraham, Werner. . Methodische Überlegungen zu Grammatikalisier-
ung, zykli-schem Wandel und dem Wechsel von Analytik zu Synthetik—
und zyklisch weiter zu Analytik. In: Bittner, Dagmar and Livio Gaeta (eds.)
Kodierungstechniken im Wandel. Das Zusammenspiel von Analytik und
Synthese im Gegenwartsdeutschen, –. Berlin: de Gruyter.
Abraham, Werner. . Illocutive force is speaker and information source
concern. What type of syntax does the representation of speaker deixis
require? Templates vs. derivational structure? In: Abraham, Werner and
Elisabeth Leiss (eds.) Modality and Theory of Mind Elements Across Lan-
guages, –. Berlin: Mouton de Gruyter.
Abraham, Werner. . Strong modality and truth disposability in syntactic
subordination: What is the locus of the phase edge validating modal
adverbials? Studia Linguistica /, –.
Adamson, Sylvia. . From empathetic deixis to empathetic narrative:
Stylisation and (de)subjectivisation as processes of language change. In:
Stein, Dieter and Susan Wright (eds.) Subjectivity and Subjectivisation:
Linguistic Perspectives, –. Cambridge: Cambridge University Press.
Adamson, Sylvia. . A lovely little example: word order options and
category shift in the premodifying string. In: Fischer, Olga, Anette Rosen-
bach, and Dieter Stein (eds.) Pathways of Change: Grammaticalization in
English, –. Amsterdam: Benjamins.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Adone, Dany and Ingo Plag (eds.). . Creolization and Language Change.
Tübingen: Niemeyer.
Ahn, Mikyung. . The English causal adposition thanks to: a grammatical-
ization perspective. In: Linguistic Society of Korea (eds.) Current issues in
unity and diversity of languages: collection of the papers selected from the
CIL , held at Korea University in Seoul, on July –, , –.
Seoul: Linguistic Society of Korea.
Ahn, Mikyung. . From ‘favor’ to ‘cause’: the English causal adposition thanks
to and its Korean counterpart tekpwuney. Language Sciences , –.
Aijmer, Karin. . I think—an English modal particle. In: Swan, T. and
O. J. Westvik (eds.) Modality in Germanic Languages: Historical and
Comparative Perspective, –. Berlin: Mouton de Gruyter.
Aijmer, Karen. . English Discourse Particles: Evidence from a Corpus.
Amsterdam: Benjamins.
Aikhenvald, Alexandra Y. . Areal diffusion in northwest Amazonia: the
case of Tariana. Anthropological Linguistics , –.
Aikhenvald, Alexandra Y. . Language Contact in Amazonia. Oxford:
Oxford University Press.
Aikhenvald, Alexandra Y. . Mechanisms of change in areal diffusion: new
morphology and language contact. Journal of Linguistics , –.
Aikhenvald, Alexandra Y. . The grammaticalization of evidentiality. In:
Narrog, Heiko and Heine, Bernd (eds.) Handbook of Grammaticalization,
–. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. . Areal diffusion and the limits of grammat-
icalization: An Amazonian perspective In: Narrog, Heiko and Bernd Heine
(eds.) Grammaticalization from a Typological Perspective, –. Oxford:
Oxford University Press.
Aikhenvald, Alexandra Y. and Robert M. W. Dixon (eds.). . Areal Diffu-
sion and Genetic Inheritance: Problems in Comparative Linguistics. Oxford:
Oxford University Press.
Aikhenvald, Alexandra Y. and R. M. W. Dixon (eds.). . Grammars in
Contact: A Cross-Linguistic Typology. Oxford: Oxford University Press.
Ambrazas, Vytautas. . Lithuanian Grammar. Vilnius: Baltos Lankos.
Andersen, Henning. . Understanding linguistic innovations. In: Breivik,
Leiv Egil and Ernst Håkon Jahr (eds.) Language Change: Contributions To
the Study of Its Causes, –. Berlin: Mouton de Gruyter.
Andersen, Henning. a. Actualization and the (uni)directionality of change.
In: Andersen, Henning (ed.) Linguistic Change in Progress: Papers from a
Workshop Held at the th International Conference on Historical Linguistics,
Vancouver, B.C., August , –. Amsterdam: Benjamins.
Andersen, Henning. b. Markedness and the theory of change. In: Ander-
sen, Henning (ed.) Linguistic Change in Progress: Papers from a Workshop
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Held at the th International Conference on Historical Linguistics, Van-
couver, B.C., August , –. Amsterdam: Benjamins.
Andersen, Henning. . Grammation, regrammation, and degrammation.
Diachronica /, –.
Andersen, Henning. . From morphologization to demorphologization In:
Luraghi, Silvia and Vit Bubenik (eds.) Continuum Companion to Historical
Linguistics, –. New York: Continuum.
Anderson, Lloyd B. . The perfect as a universal and as a language specific
category. In: Hopper, Paul J. (ed.) Tense–Aspect: Between Semantics and
Pragmatics, –. Amsterdam: Benjamins.
Anderson, Lloyd B. . Evidentials, paths of change, and mental maps:
typologically regular asymmetries. In: Chafe, Wallace and Marianne
Mithun (eds.) Evidentiality: The Linguistic Coding of Epistemology,
–. Norwood: Ablex.
Ansaldo, Umberto. . Comparative Constructions in Sinitic: Areal Typ-
ology and Patterns of Grammaticalization. Ph.D. dissertation, Stockholm
University.
Ansaldo, Umberto. . Surpass comparatives in Sinitic and beyond: typ-
ology and grammaticalization. Linguistics /, –.
Ansaldo, Umberto and Lisa Lim. . Phonetic absence as syntactic prom-
inence: Grammaticalization in isolating tonal languages. In: Fischer, Olga,
Muriel Norde, and Harry Perridon (eds.) Up and Down the Cline—the
Nature of Grammaticalization, –. Amsterdam: Benjamins.
Ansaldo, Umberto, Walter Bisang, and Pui Yiu Szeto. . Grammatical-
ization in isolating languages and the notion of complexity. In: Narrog,
Heiko and Bernd Heine (eds.) Grammaticalization from a Typological
Perspective. –. Oxford: Oxford University Press.
Arcodia, Giorgio Francesco. . Grammaticalisation with coevolution of
form and meaning in East Asia? Evidence from Sinitic. Language Sciences
, –.
Arends, Jacques, Pieter Muysken, and Norval Smith. (eds.) . Pidgins and
Creoles: An Introduction. (Creole Language Library, .) Amsterdam:
Benjamins.
Ariel, Mira. . Pragmatics and Grammar. Cambridge: Cambridge University
Press.
Arkadiev, Peter and Timur Maisak. . Grammaticalization in the
North Caucasian languages. In: Narrog, Heiko and Bernd Heine (eds.)
Grammaticalization from a Typological Perspective. –. Oxford:
Oxford University Press.
Arroyo, José Luis Blas. . From politeness to discourse marking: The
process of pragmaticalization of muy bien in vernacular Spanish. Journal
of Pragmatics , –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Askedal, John Ole. . ‘Degrammaticalization’ versus typology. Reflections on
a strained relationship. In: Eythórsson, Thórhallur (ed.) Grammatical Change
and Linguistic Theory: The Rosendal Papers, –. Amsterdam: Benjamins.
Athanasiadou, Angeliki, Costas Canakis, and Bert Cornillie (eds.). . Sub-
jectification: Various Paths to Subjectivity. Berlin: Mouton de Gruyter.
Auer, Peter. . The pre-front field position in spoken German and its
relevance as a grammaticalization position. Pragmatics , –.
Auer, Peter and Susanne Günthner. . Die Entstehung von Diskursmar-
kern im Deutschen: Ein Fall von Grammatikalisierung? In: Leuschner,
Torsten, Tanja Mortelmans, and Sarah De Groodt (eds.) Grammatikalisierung
im Deutschen, –. Berlin: Walter de Gruyter.
Awolaye, Yiwola. . Reflexivization in Kwa languages. In: Dimmendal,
Gerrit J. D. (ed.) Current Approaches to African Linguistics, vol. , –.
Dordrecht, Cinnaminson: Foris Publications.
Baker, Mark. . Incorporation: A Theory of Grammatical Function Chan-
ging. Chicago: Chicago University Press.
Baker, Philip and Anand Syea. (eds.) . Changing Meanings, Changing
Functions: Papers Relating to Grammaticalization in Contact Languages.
London: University of Westminster Press.
Barðdal, Jóhanna, Elena Smirnova, Lotte Sommerer, and Spike Gildea (eds.).
. Diachronic Construction Grammar. Amsterdam: Benjamins.
Barth-Weingarten, Dagmar and Elizabeth Couper-Kuhlen. . On the
development of final though: A case of grammaticalization? In: Wischer,
Ilse and Gabriele Diewald (eds.) New Reflections on Grammaticalization,
–. Amsterdam: Benjamins.
Beijering, Karin. . Expressions of Epistemic Modality in Mainland Scan-
dinavian: A Study Into the Lexicalization-Grammaticalization-Pragmatica-
lization Interface. Ph.D. dissertation, Rijksuniversiteit Groningen.
Beijering, Karin. . The lexicalization–grammaticalization–pragmaticaliza-
tion interface. The case of Mainland Scandinavian jeg tror. In: Smith,
Andrew D. M., Graeme Trousdale, and Richard Waltereit (eds.) New
Directions in Grammaticalization Research, –. Amsterdam: Benjamins.
Bergqvist, Henrik. . lntersubjectification revisited: a cross-categorical
perspective. In: Guentchéva, Zlatka (ed.) Epistemic Modalities and Eviden-
tiality in Cross-Linguistic Perspective, –. Berlin: Mouton de Gruyter.
Bergs, Alexander and Gabriele Diewald (eds.). . Constructions and Lan-
guage Change. Berlin: Mouton de Gruyter.
Bhat, D.N.S. . The Prominence of Tense, Aspect and Mood. Amsterdam:
Benjamins.
Biber, Douglas. . Modal use across registers and time. In: Curzan, Anne
and Kimberly Emmons (eds.) Studies in the History of the English Language
II: Unfolding conversations, –. Berlin: Mouton de Gruyter.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Biber, Douglas and Bethany Gray. . Grammatical change in the noun
phrase: the influence of written language use. English Language and Lin-
guistics /, –.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward
Finegan. . Longman Grammar of Spoken and Written English. London:
Longman.
Bisang, Walter. . Grammaticalization without coevolution of form and
meaning as an areal phenomenon in East and mainland Southeast Asia: the
case of tense-aspect-mood (TAM). In: Bisang, Walter, Nikolaus Himmel-
mann, and Björn Wiemer (eds.) What Makes Grammaticalization? A Look
from its Components and its Fringes, –. Berlin: Mouton de Gruyter.
Bisang, Walter. . Grammaticalization and the areal factor: The
perspective of East and mainland Southeast Asian languages. In: López-
Couso, Maria Jóse, and Elena Seoane (eds.) Rethinking Grammatical-
ization: New Perspectives, –. Amsterdam: Benjamins.
Bisang, Walter. . Grammaticalization in Chinese. A construction-based
account. In: Traugott, Elizabeth Closs and Graeme Trousdale (eds.)
Gradience, Gradualness and Grammaticalization, –. Amsterdam:
Benjamins.
Bisang, Walter. . Grammaticalization and typology. In: Narrog, Heiko
and Heine, Bernd (eds.) Handbook of Grammaticalization, –. Oxford:
Oxford University Press.
Bisang, Walter. . Grammaticalization in Chinese: A cross-linguistic per-
spective. In: Xing, Janet Zhiqun (ed.) A Typological Approach to Grammat-
icalization and Lexicalization: East Meets West, –. Berlin: De Gruyter
Mouton.
Bisang, Walter, Andrej Malchukov, Iris Rieder, Linlin Sun, Marvin Martiny,
and Svenja Luell. . Position paper: Universal and areal patterns in
grammaticalization. In Bisang, Walter and Andrej Malchukov (eds.) Gram-
maticalization Scenarios: Areal Patterns and Cross-Linguistic Variation,
–. Berlin: De Gruyter.
Blakemore, Diane. . Relevance and Linguistic Meaning: The Semantics
and Pragmatics of Discourse Markers. Cambridge: Cambridge University
Press.
Boland, Johanna H.G. . Aspect, Tense, and Modality. Theory, Typology,
Acquisition. Ph.D. thesis. Universiteit van Amsterdam.
Boretzky, Norbert. . Kreolsprachen, Substrate und Sprachwandel.
Wiesbaden: Otto Harrassowitz.
Boretzky, Norbert and Birgit Igla. . Balkanische (südosteuropäische)
Einflüsse im Romani. In: Hinrichs, Uwe and Uwe Büttner (eds.) .
Handbuch der Südosteuropa-Linguistik, –. Wiesbaden: Harrassowitz.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Börjars, Kersti and Nigel Vincent. . Grammaticalization and unidirec-
tionality. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook
of Grammaticalization, –. Oxford: Oxford University Press.
Börjars, Kersti, Nigel Vincent, and George Walkden . On constructing a
theory of grammatical change. Transactions of the Philological Society /,
–.
Boye, Kasper and Peter Harder. . A usage-based theory of grammatical
status and grammaticalization. Language , –.
Bradley, Peter T. and Ian Mackenzie. . Spanish. An Essential Grammar.
London: Routledge.
Breban, Tine. . Grammaticalization, subjectification and leftward move-
ment of English adjectives of difference in the noun phrase. Folia Linguis-
tica /, –.
Breban, Tine. . Refining secondary grammaticalization by looking at
subprocesses of change. Language Sciences , –.
Breban, Tine and Kristin Davidse . The history of very: the directionality
of functional shift and (inter)subjectification. English Language and Lin-
guistics /, –.
Breban, Tine and Svenja Kranich. . Introduction to What happens after
grammaticalization? Secondary grammaticalization and other late stage
processes. Language Sciences , –.
Brems, Lieselotte, Lobke Ghesquière, and Freek Van de Velde (eds.) Intersub-
jectivity and Intersubjectification in Grammar and Discourse: Theoretical
and Descriptive Advances. Amsterdam: Benjamins.
Brinton, Laurel J. . Pragmatic Markers in English: Grammaticalization
and Discourse Functions. Berlin: Mouton de Gruyter.
Brinton, Laurel J. . From matrix clause to pragmatic marker: The history
of look-forms. Journal of Historical Pragmatics , –.
Brinton, Laurel J. . Subject clitics in English. A case of degrammaticaliza-
tion? In: Lindquist, Hans and Christian Mair (eds.) Corpus Approaches to
Grammaticalization in English, –. Amsterdam: Benjamins.
Brinton, Laurel J. . The development of I mean: implications for the study
of historical pragmatics. In: Fitzmaurice, Susan and Irma Taavitsainen
(eds.) Methods in Historical Pragmatics, –. Berlin: Mouton de Gruyter.
Brinton, Laurel J. a. “Where grammar and lexis meet”. Composite predicates
in English. In: Seoane, Elena and Maria Jose Lopez-Couso (eds.) Theoretical
and Empirical Issues in Grammaticalization, –. Amsterdam: Benjamins.
Brinton, Laurel J. b. The Comment Clause in English: Syntactic Origins
and Pragmatic Development. Cambridge: Cambridge University Press.
Brinton, Laurel J. . The development of I mean: Implications for the study
of historical pragmatics. In: Fitzmaurice, Susan M. and Irma Taavitsainen
(eds.) Methods in Historical Pragmatics, –. Berlin: Mouton de Gruyter.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Brinton, Laurel J. . “The ghosts of old morphology”. Lexicalization or (de)
grammaticalization? In: Davidse, Kristin, Tine Breban, Lieselotte Brems,
and Tanja Mortelmans (eds.) Grammaticalization and Language Change:
New Reflections, –. Amsterdam: Benjamins.
Brinton, Laurel. . The Evolution of Pragmatic Markers in English: Path-
ways of Change. Cambridge: Cambridge University Press.
Brinton, Laurel J. and Dieter Stein. . Functional renewal. In: Andersen,
Henning (ed.) Historical Linguistics : Selected Papers from the th
International Conference on Historical Linguistics, Los Angeles, –
August , –. Amsterdam: Benjamins.
Brinton, Laurel J. and Elizabeth Closs Traugott. . Lexicalization and
Language Change. (Research Surveys in Linguistics.) Cambridge: Cam-
bridge University Press.
Brinton, Laurel J. and Elizabeth Closs Traugott. . Lexicalization and gram-
maticalization all over again. In: Salmons, Joseph C. and Shannon Dubenion-
Smith (eds.) Historical Linguistics : Selected Papers From the th Inter-
national Conference on Historical Linguistics, –. Amsterdam: Benjamins.
Bruyn, Adrienne. a. Relative clauses in early Sranan. In: Arends, Jacques
(ed.) The Early Stages of Creolization, –. Amsterdam: Benjamins.
Bruyn, Adrienne. b. Grammaticalization in Creoles: The Development of
Determiners and Relative Clauses in Sranan. Amsterdam: IFOTT (Institute
for Functional Research into Language and Language Use).
Bruyn, Adrienne. . On identifying instances of grammaticalization in
Creole languages. In: Baker, Philip and Anand Syea (eds.) Changing Mean-
ings, Changing Functions: Papers Relating to Grammaticalization in Con-
tact Languages, –. London: University of Westminster Press.
Bruyn, Adrienne. . What can this be? In: Schmid, Monika S., Jennifer
R. Austin, and Dieter Stein (eds.) Historical Linguistics : Selected Papers
from the th International Conference on Historical Linguistics, Düsseldorf,
– August , –. Amsterdam: Benjamins.
Budts, Sara and Peter Petré. . Reading the intentions of be going to: On the
subjectification of future markers. Folia Linguistica Historica , –.
Burridge, Kate. . From modal auxiliary to lexical verb: The curious case of
Pennsylvania German wotte. In: Hogg, Richard M. and Linda van Bergen
(eds.) Historical Linguistics . Selected Papers From the th Interna-
tional Conference on Historical Linguistics, Manchester, August . Vol-
ume : Germanic Linguistics, –. Amsterdam: Benjamins.
Burridge, Kate. . From modal auxiliary to lexical verb: The curious case of
Pennsylvania German wotte. In: Hogg, Richard M. and Linda von Bergen
(eds.) Historical Linguistics : Selected Papers from the th Inter-
national Conference on Historical Linguistics, Manchester, August ,
Volume : Germanic Linguistics, –. Amsterdam: Benjamins.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Bybee, Joan L. . Morphology: A Study of the Relation between Meaning
and Form. Amsterdam: Benjamins.
Bybee, Joan L. . The diachronic dimension in explanation. In: Hawkins,
John A. (ed.) Explaining Language Universals, –. Oxford: Blackwell.
Bybee, Joan L. a. Mechanisms of change in grammaticization: The role of
frequency. In: Joseph, Brian D. and Richard D. Janda (eds.) The Handbook
of Historical Linguistics, –. Oxford: Blackwell.
Bybee, Joan L. b. Cognitive processes in grammaticalization. In: Toma-
sello, Michael (ed.) The New Psychology of Language, Volume II, –.
Mahwah: Erlbaum.
Bybee, Joan L. . From usage to grammar: The mind’s response to repe-
tition. Language /, –.
Bybee, Joan L. . Language universals and usage-based theory. In: Chris-
tiansen, Morten H., Chris Collins, and Shimon Edelman (eds.) Language
Universals, –. Oxford: Oxford University Press.
Bybee, Joan L. . Usage-based theory and grammaticalization. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Bybee, Joan L. . Language Change. Cambridge: Cambridge University Press.
Bybee, Joan L. and William Pagliuca. . Cross linguistic comparison and
the development of grammatical meaning. In: Fisiak, Jacek (ed.) Historical
Semantics, Historical Word-Formation, –. Berlin: Mouton Publishers.
Bybee, Joan L. and William Pagliuca. . The evolution of future meaning.
In: Ramat, Anna Giacalone, Onofrio Carruba, and Guiliana Bernini (eds.)
Papers from the VIIth International Conference on Historical Linguistics,
–. Amsterdam: Benjamins.
Bybee, Joan L. and Sandra Thompson. . Three frequency effects in syntax.
Proceedings of the Twenty-third Annual Meeting of the Berkeley Linguistics
Society, –.
Bybee, Joan L., William Pagliuca, and Revere D. Perkins. . On the
asymmetries in the affixation of grammatical material. In: Croft, William
et al. (eds.) Studies in Typology and Diachrony: Papers Presented to Joseph
H. Greenberg on his th Birthday, –. Amsterdam: Benjamins.
Bybee, Joan L., William Pagliuca, and Revere D. Perkins. . Back to the
future. In: Traugott, Elizabeth C. and Bernd Heine (eds.) Approaches to
Grammaticalization, Volume , –. Amsterdam: Benjamins.
Bybee, Joan L., Revere D. Perkins, and William Pagliuca. . The Evolution
of Grammar: Tense, Aspect and Modality in the Languages of the World.
Chicago: University of Chicago Press.
Campbell, Lyle. . What’s wrong with grammaticalization? Language
Sciences /–, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Campbell, Lyle. . Historical Linguistics: An Introduction. nd edition.
Cambridge, MA: MIT Press.
Campbell, Lyle and Richard Janda. . Introduction: conceptions of gram-
maticalization and their problems. Language Sciences /–, –.
Carlier, Anne, Walter de Mulder, and Béatrice Lamiroy. . Introduction:
The pace of grammaticalization in a typological perspective. Folia Linguis-
tica /, –.
Cennamo, Michela. . Paths of grammaticalization in Romance voice
systems. Paper presented at the symposium on “Areal Patterns of
Grammaticalization and Crosslinguistic Variation in Grammaticalization
Scenarios”, Mainz, – March .
Chafe, Wallace. . Florescence as a force in grammaticalization. In: Gildea,
Spike (ed.) Reconstructing Grammar. Comparative Linguistics and Gram-
maticalization, –. Amsterdam: Benjamins.
Chomsky, Noam A. . The Minimalist Program. Cambridge, MA: MIT Press.
Chor, Winnie. . From nonsubjective to subjective to intersubjective—a
pathway of semantic change in grammaticalization: the case of faan in
Cantonese. In: Ishiguro, Teruhiro and Kang-kwong Luke (eds.) Grammar
in Cross-Linguistic Perspective: The Syntax, Semantics, and Pragmatics of
Japanese and Chinese, –. Bern: Peter Lang.
Claridge, Claudia and Leslie Arnovick. . Pragmaticalisation and
discursisation. In: Jucker, H. and Irma Taavitsainen (eds.) Historical
Pragmatics, –. Berlin: De Gruyter Mouton.
Claudi, Ulrike and Bernd Heine. . On the metaphorical base of grammar.
Studies in Language /, –.
Coates, Jennifer. . The Semantics of the Modal Auxiliaries. London:
Croom Helm.
Collins, Peter. . Modals and Quasi-Modals in English. Amsterdam: Rodopi.
Comrie, Bernard. . Morphology and word order reconstruction: Prob-
lems and prospects. In: Fisiak, Jaced (ed.) Historical Morphology, –.
The Hague: Mouton.
Comrie, Bernard. . Language Universals and Linguistic Typology. nd
edition. Oxford: Blackwell.
Comrie, Bernard. . Typology and reconstruction. In: Jones, Charles (ed.)
Historical Linguistics: Problems and Perspectives, –. London: Longman.
Comrie, Bernard. . Perspectives on grammaticalization. In: Ohori, Toshio
(ed.) Studies in Japanese Grammaticalization: Cognitive and Discourse
Perspectives, –. Tokyo: Kurosio Publishers.
Cornillie, Bert. . The shift from lexical to subjective readings of Spanish
prometer ‘to promise’ and amenazar ‘to threaten’. A corpus-based account.
Pragmatics /, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Cornillie, Bert. . On modal grounding, reference points, and subjectifica-
tion: The case of the Spanish epistemic modals. Annual Review of Cognitive
Linguistics , –.
Cornillie, Bert. . On the grammaticalization and (inter)subjectivity of
evidential (semi-)auxiliaries in Spanish. In: Seoane, Elena and María José
López-Couso (eds.) Theoretical and Empirical Issues in Grammatical-
ization, –. Amsterdam: Benjamins.
Cornillie, Bert and Nicole Delbeque (eds.). . Topics in Subjectification
and Modalization. Special Issue. Belgian Journal of Linguistics .
Couper-Kuhlen, Elizabeth. . Grammaticalization and conversation. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Craig, Colette G. . Ways to go in Rama: a case study in polygramma-
ticalization. In: Traugott, Elizabeth C. and Bernd Heine (eds.)
Approaches to Grammaticalization, Volume , –. Amsterdam:
Benjamins.
Croft, William. . Explaining Language Change: An Evolutionary Approach.
London: Longman.
Cysouw, Michael. . The asymmetry of affixation. In: Gärtner, Hans-
Martin et al. (eds.) Puzzles for Krifka. (http://www. zas.gwz-berlin.de/pub-
lications/–-puzzles-for-krifka/), –.
Dahl, Östen. (ed.) . Tense and Aspect in the Languages of Europe. Berlin:
Mouton de Gruyter.
Dahl, Östen. . Inflationary effects in language and elsewhere. In: Bybee,
Joan L. and Paul J. Hopper (eds.) Frequency and the Emergence of Linguistic
Structure, –. Amsterdam: Benjamins.
Dahl, Östen. . The Growth and Maintenance of Linguistic Complexity.
Amsterdam, Philadelphia: Benjamins.
Dahl, Östen. . Grammaticalization in the languages of Europe. In: Narrog,
Heiko and Bernd Heine (eds.) Grammaticalization from a Typological
Perspective. –. Oxford: Oxford University Press.
Dávalos, Jimena Tena. . The end of a cycle: grammaticalization of the
future tense in Mexican Spanish. In: Hengeveld, Kees, Heiko Narrog, and
Hella Olbertz (eds.) The Grammaticalization of Tense, Aspect, Modality and
Evidentiality, –. Berlin: Mouton de Gruyter.
Davidse, Kristin and Simon de Wolf. . Lexicalization and grammatical-
ization: the development of idioms and grammaticalized expressions with
no question. Text & Talk /, –.
Davidse, Kristin, Lieselotte Brems, and Liesbeth De Smedt. . Type noun
uses in the English NP: a case of right to left layering. International Journal
of Corpus Linguistics /, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Davidse, Kristin, Lieven Vandelotte, and Hubert Cuyckens (eds.). . Subjectifica-
tion, Intersubjectification and Grammaticalization. Berlin: De Gruyter Mouton.
De Mulder, Walter and Anne Carlier. . The grammaticalization of defin-
ite articles. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford Hand-
book of Grammaticalization, –. Oxford: Oxford University Press.
De Smet, Hendrik. . Analyzing reanalysis. Lingua , –.
De Smet, Hendrik. . Grammatical interference. Subject marker for and
the phrasal verb particles out and forth. In: Traugott, Elizabeth Closs and
Graeme Trousdale (eds.) Gradience, Graduality and Grammaticalization,
–. Amsterdam: Benjamins.
De Smet, Hendrik. . Does innovation need reanalysis? In: Coussé, Evie
and Ferdinand von Mengden (eds.) Usage-Based Approaches to Language
Change, –. Amsterdam: Benjamins.
Degand, Liesbeth and Jacqueline Evers-Vermeul. . Grammaticalization or
pragmaticalization of discourse markers?: More than a terminological issue.
Journal of Historical Pragmatics /, –.
Degand, Liesbeth and Benjamin Fagard. . Alors between discourse and
grammar: The role of syntactic position. Functions of Language /, –.
Degand, Liesbeth and Anne-Marie Simon-Vandenbergen. . Introduction:
Grammaticalization and (inter)subjectification of discourse markers. Lin-
guistics /, –.
Dehé, Nicole. . Parentheticals in Spoken English: The Syntax–Prosody
Relation. Cambridge: Cambridge University Press.
Dehé, Nicole and Yordanka Kavalova (eds.) . Parentheticals. Amsterdam:
Benjamins.
DeLancey, Scott. . Grammaticalization and the gradience of categories.
In: Bybee, Joan et al. (eds.) Essays on Language Function and Language
Type: Dedicated to T. Givón, –. Amsterdam: Benjamins.
DeLancey, Scott. . Lexical Categories. Lecture of his Santa Barbara
Lectures on Functional Syntax. The LSA Summer Institute, UC Santa
Barbara. (www.uoregon.edu/~delancey/sb/functional_syntax.doc).
DeLancey, Scott. . Grammaticalization and syntax. A functional view. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Denison, David. . Syntax. In: Romaine, Suzanne (ed.) The Cambridge
History of the English Language. Vol. IV. –, –. Cambridge:
Cambridge University Press.
Dér, Csilla Ilona. . On the status of discourse markers. Acta Linguistica
Hungarica /, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Detges, Ulrich. . Time and truth: the grammaticalization of resultatives
and perfects within a theory of subjectification. Studies in Language ,
–.
Detges, Ulrich. . How cognitive is grammaticalization? The history of the
Catalan perfet perifràstic. In: Fischer, Olga, Muriel Norde, and Harry
Perridon (eds.) Up and Down the Cline—the Nature of Grammaticalization,
–. Amsterdam: Benjamins.
Detges, Ulrich and Waltereit, Richard. . Grammaticalization vs. reanaly-
sis: A semantic-pragmatic account of functional change in grammar.
Zeitschrift für Sprachwissenschaft , –.
Devos, Maud and Jenneke van der Wal (eds.). . Grammaticalization and
(Inter)subjectification in African Languages. Africana Linguistica
(Special issue).
Diewald, Gabriele. . Eine Einführung in Sein und Werden grammatischer
Formen. Tübingen: Niemeyer.
Diewald, Gabriele. . A model for relevant types of contexts in grammat-
icalization. In: Wischer, Ilse and Gabriele Diewald (eds.) New Reflections on
Grammaticalization, –. Amsterdam: Benjamins.
Diewald, Gabriele. . The catalytic function of constructional restrictions
in grammaticalization. In: Verhoeven, Elisabeth et al. (eds.) Studies on
Grammaticalization, –. Berlin: Mouton de Gruyter.
Diewald, Gabriele. . Konstruktionen und Paradigmen. Zeitschrift für
germanistische Linguistik , –.
Diewald, Gabriele. . On some problem areas in grammaticalization studies.
In: Stathi, Katerina, Elke Gehweiler, and Ekkehard König (eds.) Grammat-
icalization: Current Views and Issues, –. Amsterdam: Benjamins.
Diewald, Gabriele. a. Grammaticalization and pragmaticalization. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Diewald, Gabriele. b. Pragmaticalization (defined) as grammaticalization
of discourse functions. Linguistics /, –.
Diewald, Gabriele and Elena Smirnova. . Paradigmatic integration. The fourth
stage in an expanded grammaticalization scenario. In: Davidse, Kristin, Tine
Breban, Lieselotte Brems, and Tanja Mortelmans (eds.) Grammaticalization
and Language Change: New Reflections, –. Amsterdam: Benjamins.
Dik, Simon C. a. The Theory of Functional Grammar—Part : The
Structure of the Clause. nd revised edition. Edited by Kees Hengeveld.
Berlin: Mouton de Gruyter.
Dik, Simon C. b. The Theory of Functional Grammar, Part : Complex
and Derived Constructions. nd revised edition. Edited by Kees Hengeveld.
Berlin: Mouton de Gruyter.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Do-Hurinville, Thành. . Les parties du discours en vietnamien: gramma-
ticalisation et transcatégorialité. Bulletin de la Société Linguistique de Paris
/, –.
Do-Hurinville, Thành and Huy Linh Dao. . La transcategorialité: une
histoire de limite sans limite. Bulletin de la Société Linguistique de Paris
/, –.
Do-Hurinville, Thành and Sylvie Hancil. . La transcatégorialité à travers
les langues. University of Rouen, Typescript.
Dong, Xiufang. . Lexicalization in the history of the Chinese language. In:
Xing, Janet Zhiqun (ed.) Newest Trends in the Study of Grammaticalization
and Lexicalization in Chinese, –. Berlin: Mouton de Gruyter.
Dostie, Gaetane. . Pragmaticalisation et marqueurs discursifs: Analyse
sémantique et traitement lexicographique. Brussels: De Boeck and Larcier.
Dostie, Gaétane. . Discourse markers and regional variation in French:
A lexico-semantic approach. In: Beeching, Kate, Nigel Armstrong, and
Francoise Gadet (eds.) Sociolinguistic Variation in Contemporary French,
–. Amsterdam: Benjamins.
Dryer, Matthew S. . Prefixing versus suffixing in inflectional morphology.
In: Haspelmath, Martin, Matthew S. Dryer, David Gil, and Bernard Comrie
(eds.) The World Atlas of Linguistic Structures, –. Oxford: Oxford
University Press.
Du Feu, Veronica. . Rapanui. London: Routledge.
Eckardt, Regine. . Grammaticalization and semantic reanalysis. In:
Maienborn, Claudia et al. (eds.) Semantics: An International Handbook of
Natural Language Meaning, –. Berlin: Mouton de Gruyter.
Enfield, N. J. . Heterosemy and the grammar-lexicon trade-off. In:
Ameka, Felix, Allan Dench, and Nicholas Evans (eds.) Catching Language,
–. Berlin: Mouton de Gruyter.
Espinal, M. Teresa. . The representation of disjunct constituents.
Language : –.
Erman, Britt and Ulla-Britt Kotsinas. . Pragmaticalization: The case of ba
and you know. Studier i modern sprakvetenskap , –.
Esseesy, Mohssen. . Typological features of grammaticalization in
Semitic. In: Narrog, Heiko and Bernd Heine (eds.) Grammaticalization
from a Typological Perspective. –. Oxford: Oxford University Press.
Evans, Nicholas and David Wilkins. . The knowing ear: an Australian
test of universal claims about the semantic structure of sensory verbs
and their extension into the domain of cognition. Cologne: Institut für
Sprachwissenschaft.
Evers-Vermeul, Jacqueline, Liesbeth Degand, Benjamin Fagard, and Liesbeth
Mortier. . Historical and comparative perspectives on subjectification:
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
A corpus-based analysis of Dutch and French causal connectives. Linguistics
/, –.
Fagard, Benjamin and Alexandru Mardale. . The pace of grammatical-
ization and the evolution of prepositional systems: Data from Romance.
Folia Linguistica Historica /, –.
Fischer, Kerstin (ed.) . Approaches to Discourse Particles. Amsterdam:
Elsevier.
Fischer, Olga. . Grammaticalisation: Unidirectional, non-reversable? In:
Fischer, Olga, Anette Rosenbach, and Dieter Stein (eds.) Pathways of
Change. Grammaticalization in English, –. Amsterdam: Benjamins.
Fischer, Olga. . Principles of grammaticalization and linguistic reality. In:
Rohdenburg, Günter and Britta Mondorf (eds.) Determinants of Grammat-
ical Variation in English, –. Berlin: Mouton de Gruyter.
Fischer, Olga. a. Morphosyntactic Change: Functional and Formal Per-
spectives. Oxford: Oxford University Press.
Fischer, Olga. b. The development of English parentheticals: A case of
grammaticalization? In: Schendl, Herbert and Ute Smit (eds.) Tracing
English through Time: Explorations in Language Variation: A Festschrift
for Herbert Schendl on the Occasion of his th Birthday, –. Vienna:
Braumüller.
Fischer, Olga. . On analogy as the motivation for grammaticalization.
Studies in Language /, –.
Fischer, Olga. . An analogical approach to grammaticalization. In: Stahl,
Katerina, Elke Gehweiler, and Ekkehard Konig (eds.) Grammaticalization:
Current Views and Issues, –. Amsterdam: Benjamins.
Fischer, Olga. . Grammaticalization as analogically driven change? In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Fischer, Olga. . An inquiry into unidirectionality as a foundational
element of grammaticalization: On the role played by analogy and the
synchronic grammar system in processes of language change. Studies in
Language /, –.
Fischer, Olga and Annette Rosenbach. . Introduction. In: Fischer, Olga,
Anette Rosenbach and Dieter Stein (eds.) Pathways of Change. Grammat-
icalization in English, –. Amsterdam: Benjamins.
Fitzmaurice, Susan. a. The meanings and uses of the progressive con-
struction in an early eighteenth-century English network. In: Curzan,
A. and K. Emmons (eds.) Studies in the History of the English Language
II. Unfolding Conversations, –. Berlin: De Gruyter.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Fitzmaurice, Susan. b. Subjectivity, intersubjectivity and the historical
construction of interlocutor stance: from stance markers to discourse
markers. Discourse Studies /, –.
Fortescue, Michael. . The Abstraction Engine. Extracting Patterns in
Language, Mind and Brain. Amsterdam: Benjamins.
Fortuin, Egbert and Ronny Boogaart. . Imperative as conditional: from
constructional to compositional semantics. Cognitive Linguistics /,
–.
Frajzyngier, Zygmunt. . Grammaticalization of the Complex Sentence: A
Case Study in Chadic. Amsterdam: Benjamins.
François, Alexandre. . Semantic maps and the typology of colexification.
Intertwining polysemous networks across languages. In: Vanhove, Martine
(ed.) From Polysemy to Semantic Change: Towards a Typology of Lexical
Semantic Associations, –. Amsterdam: Benjamins.
Frank-Job, Barbara. . A dynamic-interactional approach to discourse
markers. In: Fischer, Kerstin (ed.) . Approaches to Discourse Particles,
–. Amsterdam: Elsevier.
Fraser, Bruce. . An approach to discourse markers. Journal of Pragmatics
/, –.
Fraser, Bruce. . Pragmatic markers. Pragmatics /, –.
Fraser, Bruce. . What are discourse markers. Journal of Pragmatics ,
–.
Fraser, Bruce. . Topic orientation markers. Journal of Pragmatics ,
–.
Fried, Miriam. . Construction Grammar as a tool for diachronic analysis.
Constructions and Frames /, –.
Fried, Miriam. . Grammaticalization and lexicalization effects in partici-
pial morphology: A Construction Grammar approach to language change.
In: van linden, An, Jean-Christophe Verstraete, and Kristin Davidse (eds.)
Formal Evidence in Grammaticalization Research, –. Amsterdam:
Benjamins.
Gabelentz, Georg von der. [] . Die Sprachwissenschaft: lhre
Aufgaben, Methoden und bisherigen Ergebnisse. nd edition. Leipzig:
Weigel Nachf.
Gaeta, Livio. . Some remarks on analogy, reanalysis and grammaticalization.
In: Giacalone Ramat, Anna and Paul J. Hopper (eds.) The Limits of
Grammaticalization, –. Amsterdam: Benjamins.
Gaeta, Livio. . Analogical change. In: Luraghi, Silvia and Vit Bubenik
(eds.) Continuum Companion to Historical Linguistics, –. New York:
Continuum.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
García García, Luisa. . Germanische Kausativbildung: Die deverbalen jan-
Verben im Gotischen. Heidelberg: Vandenhoeck and Ruprecht.
Gast, Volker and Johan van der Auwera. . What is ‘contact-induced
grammaticalization?’ Evidence from Mayan and Mixe-Zoquean languages.
Typescript, University of Antwerp.
Geis, Michael L. and Arnold Zwicky. . On invited inferences. Linguistic
Inquiry , –.
Georgakopoulos, Thanasis and Stephane Polis. . The semantic map
model: State of the art and future avenues for linguistic research. Language
and Linguistics Compass , : February .
Ghesquière, Lobke. . On the subjectification and intersubjectification
paths followed by adjectives of completeness. In: Davidse, Kristin, Lieven
Vandelanotte, and Hubert Cuyckens (eds.) Subjectification, Intersubjectifi-
cation and Grammaticalization, –. Berlin: Mouton de Gruyter.
Ghesquière, Lobke. . The Directionality of (Inter)subjectification in the
English Noun Phrase: Pathways of Change. Berlin: Mouton de Gruyter.
Ghezzi, Chiara and Piera Molinelli. . Deverbal pragmatic markers from
Latin to Italian (Lat. QUAESO and It. prego): The cyclic nature of func-
tional developments. In: Ghezzi, Chiara and Piera Molinelli (eds.) Discourse
and Pragmatic Markers from Latin to the Romance Languages, –.
Oxford: Oxford University Press.
Giacalone Ramat, Anna. . Testing the boundaries of grammaticalization.
In: Giacalone Ramat, Anna and Paul J. Hopper (eds.) The Limits of
Grammaticalization, –. Amsterdam: Benjamins.
Giacalone Ramat, Anna and Caterina Mauri. . The grammaticalization of
coordinating interclausal connectives. In: Narrog, Heiko and Bernd Heine
(eds.) The Oxford Handbook of Grammaticalization, –. Oxford:
Oxford University Press.
Giacalone Ramat, Anna and Andrea Sansò. . Venire (ʽcomeʼ) as a passive
auxiliary in Italian. In: Devos, Maud and Jenneke van der Wal. (eds.). .
COME and GO Off the Beaten Grammaticalization Path, –. Berlin:
Mouton de Gruyter.
Gisborne, Nikolas. . Constructions, Word Grammar, and grammatical-
ization. Cognitive Linguistics /, –.
Gisborne, Nikolas and Amanda Patten. . Construction Grammar and
grammaticalization. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford
Handbook of Grammaticalization, –. Oxford: Oxford University
Press.
Givón, T. . Historical syntax and synchronic morphology: an
archaeologist’s field trip. Chicago Linguistic Society : –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Givón, T. a. Serial verbs and syntactic change: Niger-Congo. In: Li,
Charles N. (ed.) Word Order and Word Order Change, –. Austin:
University of Texas Press.
Givón, T. b. Topic, pronoun and grammatical agreement. In: Li, Charles
N. (ed.) Word Order and Word Order Change, –. Austin: University
of Texas Press.
Givón, T. c. Focus and the scope of assertion: Some Bantu evidence.
Studies in African Linguistics /, –.
Givón, T. . On Understanding Grammar. New York, San Francisco,
London: Academic Press.
Givón, T. . Tense-aspect-modality. The Creole prototype and beyond. In:
Hopper, Paul (ed.) . Tense-Aspect: Between Semantics and Pragmatics,
–. Amsterdam: Benjamins.
Givón, T. . The evolution of dependent clause morpho-syntax in Biblical
Hebrew. In: Traugott, Elizabeth C. and Bernd Heine (eds.) Approaches to
Grammaticalization. Volume , –. Amsterdam: Benjamins.
Givón, T. . Internal reconstruction: As method, as theory. In: Gildea,
Spike (ed.) Reconstructing Grammar: Comparative Linguistics and Gram-
maticalization, –. Amsterdam: Benjamins.
Givón, T. . On the relational properties of passive clauses. In: Fernández,
Zarina Estrada et al. (eds.) Studies in Voice and Transitivity, –.
München: LINCOM.
Givón, T. . The Genesis of Syntactic Complexity: Diachrony, Ontogeny,
Neuro-Cognition, Evolution. Amsterdam: Benjamins.
Givón, T. . The Diachrony of Grammar. Two volumes. Amsterdam:
Benjamins.
Givón, T. and Lynne Yang. . The rise of the English get-passive. In: Fox,
Barbara and Paul J. Hopper (eds.) . Voice: Form and Function, –.
Amsterdam: Benjamins.
Goossens, Louis. . Metonymic bridges in modal shifts. In: Panter, Klaus-
Uwe and Günter Radden (eds.) Metonymy in Language and Thought,
–. Amsterdam: Benjamins.
Goossens, Louis. . Patterns of meaning extension, “parallel chaining”,
subjectification, and modal shifts. In: Barcelona, Antonio (ed.) Metaphor
and Metonymy at the Crossroads. A Cognitive Perspective, –. Berlin:
Mouton de Gruyter.
Gould, Stephen Jay and Elisabeth S. Vrba. . Exaptation: A missing term in
the science of form. Paleobiology /, –.
Greenberg, Joseph H. a. The Languages of Africa. The Hague: Mouton.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Greenberg, Joseph H. b. Some universals of grammar with particular
reference to the order of meaningful elements. In: Greenberg, Joseph
H. (ed.) Universals of Language, –. Cambridge: MA: MIT Press.
Greenberg, Joseph H. . Some iconic relationships among place, time, and
discourse deixis. In: Haiman, John (ed.) Iconicity in Syntax, –.
Amsterdam: Benjamins.
Greenberg, Joseph H. . The last stages of grammatical elements: Con-
tractive and expansive desemanticization. In: Traugott, Elizabeth Closs and
Bernd Heine (eds.) Approaches to Grammaticalization, vol. , –.
Amsterdam: Benjamins.
Greenberg, Joseph H. . The diachronic typological approach to language.
In: Shibatani, Masayoshi and Theodora Bynon (eds.) Approaches to Language
Typology, –. Oxford: Clarendon Press. Greenberg, Joseph H. . In:
Spike Gildea, Spike (ed.) Reconstructing Grammar: Comparative Linguistics
and Grammaticalization, –. Amsterdam: Benjamins.
Grice, H. Paul. . Logic and conversation. In: Cole, Peter and Jerry L. Morgan
(eds.) Syntax and Semantics, Volume , –. New York: Academic Press.
Grimm, Jacob and Wilhelm Grimm. . Deutsches Wörterbuch, vols.
Leipzig: S. Hirzel. Reprinted by Deutscher Taschenbuchverlag .
Günthner, Susanne. . Entwickelt sich der Konzessivkonnektor obwohl
zum Diskursmarker? Grammatikalisierungstendenzen im gesprochenen
Deutsch. Linguistische Berichte , –.
Günthner, Susanne and Katrin Mutz. . Grammaticalization vs. pragma-
ticalization? The development of pragmatic markers in German and Italian.
In: Bisang, Walter, Nikolaus Himmelmann, and Björn Wiemer (eds.) What
Makes Grammaticalization? A Look from its Fringes and its Components,
–. Berlin: Mouton de Gruyter.
Guthrie, Malcolm. –. Comparative Bantu. Farnborough: Gregg Press.
Haase, Martin. . Sprachkontakt und Sprachwandel im Baskenland: die
Einflüsse des Gaskognischen und Französischen auf das Baskische. Ham-
burg: Buske.
Haase, Martin. . Gascon et basque: bilinguisme et substrat. Sprachtypologie
und Universalienforschung /, –.
Haase, Martin. . Mehrschichtiger Sprachkontakt in Malta. In: Bommes,
Michael (ed.) Sprache als Form: Festschrift für Utz Maas zum . Geburt-
stag, –. Wiesbaden: Harrassowitz.
Habermann, Mechthild. . Grammaticalization in German word-
formation. In: Müller, Peter O., Ingeborg Ohnheiser, Susan Olsen, and
Franz Rainer (eds.) Word Formation: An International Handbook of the
Languages of Europe, Volume , –. Berlin: Mouton de Gruyter.
Haegeman, Liliane. . Introduction to Government and Binding Theory.
Oxford: Blackwell.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Haig, Geoffrey. . Grammaticalization and inflectionalization in Iranian.
In: Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a
Typological Perspective, –. Oxford : Oxford University Press.
Haiman, John. . Ritualization and the development of language. In:
Pagliuca, William (ed.) . Perspectives on Grammaticalization, –.
Amsterdam: Benjamins.
Haiman, John. . Exaptation. In: Ledgeway, Adam and Ian Roberts (eds.)
The Cambridge Handbook of Historical Syntax, –. Cambridge: Cam-
bridge University Press.
Hale, Ken. . Navajo verb stem position and the bipartite structure of the
Navajo conjunct sector. Linguistic Inquiry /: –.
Halliday, M.A.K. . Categories of the Theory of Grammar. Word /,
–.
Hansen, Maj-Britt Mosegaard. a. The Function of Discourse Particles:
A Study with Special Reference to Spoken Standard French. Amsterdam:
Benjamins.
Hansen, Maj-Britt Mosegaard. b. The semantic status of discourse
markers. Lingua (/), –.
Hansen, Maj-Britt Mosegaard. . Particles at the Semantics/Pragmatics
Interface: Synchronic and Diachronic Issues. Oxford: Elsevier.
Hansen, Maj-Britt Mosegaard. . Negative cycles and grammaticalization.
In: Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Gram-
maticalization, –. Oxford: Oxford University Press.
Hansen, Maj-Britt Mosegaard. . Cyclicity in semantic/pragmatic change:
The medieval particle ja between Latin IAM and Modern French déjà. In:
Ghezzi, Chiara and Piera Molinelli (eds.) Discourse and Pragmatic Markers
from Latin to the Romance Languages, –. Oxford: Oxford University
Press.
Hansen, Maj-Britt Mosegaard. . Cyclic phenomena in the evolution of
pragmatic markers: Examples from Romance. In: Bordería, Salvador Pons,
and Òscar Loureda Lamas (eds.) Beyond Grammaticalization and Discourse
Markers: New Issues in the Study of Language Change, –. Leiden,
Boston: Brill.
Hansen, Maj-Britt Mosegaard. . In defense of a pragmatic view of
reanalysis. Typescript, University of Manchester.
Harnisch, Rüdiger. . Verstärkungsprozesse. Zu einer Theorie der
“Sekretion” und des “Re-konstruktionellen Ikonismus”. Zeitschrift für ger-
manistische Linguistik , –.
Harnisch, Rüdiger. . Zu einer Typologie sprachlicher Verstärkungspro-
zesse. In: Harnisch, Rüdiger (ed.) Prozesse sprachlicher Verstärkung. Typen
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
formaler Resegmentierung und semantischer Remotivierung, –. Berlin:
Mouton de Gruyter.
Harnisch, Rüdiger and Manuela Krieger. . Die Suche nach mehr Sinn.
Lexikalischer Wandel durch Remotivierung. Jahrbuch für germanistische
Sprachgeschichte /, –.
Harris, Alice C. . Cross-linguistic perspectives on syntactic change. In:
Joseph, Brian D. and Richard Janda (eds.) The Handbook of Historical
Linguistics, –. Oxford: Blackwell.
Harris, Alice C. and Lyle Campbell. . Historical Syntax in Cross-linguistic
Perspective. Cambridge: Cambridge University Press.
Harrison, Andrew and William J. Ashby. . Remodelling the house: the
grammaticalisation of Latin casa to French chez. Forum for Modern
Language Studies , –.
Haselow, Alexander. . Typological Changes in the Lexicon. Analytic
Tendencies in English Noun Formation. Berlin: Mouton de Gruyter.
Haselow, Alexander. . Arguing for a wide conception of grammar: The
case of final particles in spoken discourse. Folia Linguistica /, –.
Haselow, Alexander. . Sequentiality in dialogue as a trigger for
grammaticalization. In: Hancil, Sylvie and Ekkehard König (eds.)
Grammaticalization—Theory and Data, –. Amsterdam: Benjamins.
Haspelmath, Martin. . From purposive to infinitive—a universal path of
grammaticization. Folia Linguistica Historica /–, –.
Haspelmath, Martin. . The grammaticization of passive morphology.
Studies in Language /, –.
Haspelmath, Martin. a. A Grammar of Lezgian. Berlin: Mouton de
Gruyter.
Haspelmath, Martin. b. The diachronic externalization of inflection.
Linguistics , –.
Haspelmath, Martin. . Functional categories, X-bar theory, and gram-
maticalization theory. Sprachtypologie und Universalienforschung /,
–.
Haspelmath, Martin. . From Space to Time: Temporal Adverbs in the
World’s Languages. München: Lincom Europa.
Haspelmath, Martin. a. Does grammaticalization need reanalysis? Studies
in Language /, –.
Haspelmath, Martin. b. How young is Standard Average European?
Language Sciences /, –.
Haspelmath, Martin. . Why is grammaticalization irreversible? Linguistics
/, –.
Haspelmath, Martin. . The European linguistic area: Standard Average
European. In: Haspelmath, Martin, Ekkehard König, Wulf Oesterreicher,
and Wolfgang Raible (eds.) Language Typology and Language Universals:
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
An International Handbook. Volume , –. Berlin: Walter de
Gruyter.
Haspelmath, Martin. . The geometry of grammatical meaning: semantic
maps and cross-linguistic comparison. In: Tomasello, Michael (ed.) The
New Psychology of Language, vol. , –. New York: Erlbaum.
Haspelmath, Martin. . On directionality in language change with particu-
lar reference to grammaticalization. In: Fischer, Olga, Muriel Norde, and
Harry Perridon (eds.) Up and Down the Cline—the Nature of Grammat-
icalization, –. Amsterdam: Benjamins.
Haspelmath, Martin. . The gradual coalescence into ‘words’ in grammat-
icalization. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford Hand-
book of Grammaticalization, –. Oxford: Oxford University Press.
Haspelmath, Martin. . The serial verb construction: comparative concept
and cross-linguistic generalizations. Language and Linguistics /,
–.
Haspelmath, Martin. . Revisiting the anasynthetic spiral. In: Narrog,
Heiko and Bernd Heine (eds.) Grammaticalization from a Typological
Perspective, –. Oxford: Oxford University Press.
Hawkins, John. . Efficiency and Complexity in Grammars. Oxford: Oxford
University Press.
Hawkins, John. . Processing efficiency and complexity in typological
patterns. In: Song, Jae Jung (ed.) The Oxford Handbook of Linguistic
Typology, –. Oxford: Oxford University Press.
Heath, Jeffrey. . Hermit crabs: Formal renewal of morphology by phono-
logically mediated affix substitution. Language /, –.
Heine, Bernd. . Pidgin-Sprachen im Bantu-Bereich. Berlin: Dietrich
Reimer.
Heine, Bernd. . Adpositions in African languages. Linguistique Africaine
, –.
Heine, Bernd. . On the development of Kenya Pidgin Swahili. In:
Boretzky, Norbert, Werner Enninger, and Thomas Stolz (eds.) .
Beiträge zum . Essener Kolloquium über “Kontakt und Simplifikation”
vom –.. an der Universität Essen, –. Bochum: Brockmeyer.
Heine, Bernd. . Grammaticalization chains. Studies in Language /,
–.
Heine, Bernd. . Auxiliaries: Cognitive Forces and Grammaticalization.
Oxford: Oxford University Press.
Heine, Bernd. . Grammaticalization as an explanatory parameter. In:
Pagliuca, William (ed.) Perspectives on Grammaticalization, –.
Amsterdam: Benjamins.
Heine, Bernd. a. Cognitive Foundations of Grammar. Oxford: Oxford
University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Heine, Bernd. b. Possession: Sources, Forces, and Grammaticalization.
Cambridge: Cambridge University Press.
Heine, Bernd. c. Grammaticalization theory and its relevance for African
linguistics. In: Herbert, Robert K. (ed.) African Linguistics at the Crossroads,
–. Köln: Köppe.
Heine, Bernd. . On explaining grammar. The grammaticalization of have-
constructions. Theoretical Linguistics /, –.
Heine, Bernd. a. Grammaticalization chains across languages: an example
from Khoisan. In: Gildea, Spike (ed.) Reconstructing Grammar: Comparative
Linguistics and Grammaticalization, –. Amsterdam: Benjamins.
Heine, Bernd. b. Polysemy involving reflexive and reciprocal markers in
African languages. In: Frajzyngier, Zygmunt and Traci S. Curl (eds.) Reflex-
ives: Forms and Functions, –. Amsterdam: Benjamins.
Heine, Bernd. . On the role of context in grammaticalization. In: Wischer,
Ilse and Gabriele Diewald (eds.) New Reflections on Grammaticalization,
–. Amsterdam: Benjamins.
Heine, Bernd. a. Grammaticalization. In: Joseph, Brian D. and Richard Df.
Janda (eds.) The Handbook of Historical Linguistics, –. Oxford: Blackwell.
Heine, Bernd. b. On degrammaticalization. In: Blake, Barry, Kate
Burridge, and John Taylor (eds.) . Historical Linguistics . Selected
Papers from the th International Conference on Historical Linguistics,
Melbourne, – August , –. Amsterdam: Benjamins.
Heine, Bernd. . On genetic motivation in grammar. In: Radden, Günter
and Klaus-Uwe Panther (eds.) Studies in Linguistic Motivation, –.
Berlin: Mouton de Gruyter.
Heine, Bernd. . On reflexive forms in creoles. Lingua , –.
Heine, Bernd. . Constraints on contact-induced linguistic change. Journal
of Language Contact—Thema , –.
Heine, Bernd. . Identifying instances of contact-induced grammatical
replication. In: Obeng, Samuel Gyasi (ed.) Topics in Descriptive and African
Linguistics: Essays in Honor of Distinguished Professor Paul Newman,
–. Munich: Lincom Europa.
Heine, Bernd. . On polysemy copying and grammaticalization in lan-
guage contact. In: Chamoreau, Claudine and Isabelle Léglise (eds.) Cross-
linguistic Tendencies in Contact-induced Change: A Typological Approach
Based on Morphosyntactic Studies, –. Berlin: Mouton de Gruyter.
Heine, Bernd. . On discourse markers: Grammaticalization, pragmatica-
lization, or something else? Linguistics /, –.
Heine, Bernd. a. Are there two different ways of approaching grammat-
icalization? In: Hancil, Sylvie, Tine Breban, and José Vicente Lozano (eds.)
New Trends on Grammaticalization and Language Change, –. Amster-
dam: Benjamins.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Heine, Bernd. b. Grammaticalization in Africa: Two contrasting hypoth-
eses. In: Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a
Typological Perspective, –. Oxford: Oxford University Press.
Heine, Bernd. . On formulas of equivalence: Evidence from Molise Slavic.
In: Danylenko, Andrii and Motoki Nomachi (eds.) Slavic on the Language
Map of Europe: Historical and Areal-Typological Dimensions, –. Ber-
lin: De Gruyter Mouton.
Heine, Bernd and Ulrike Claudi. . On the Rise of Grammatical Categories:
Some Examples from Maa. Berlin: Reimer.
Heine, Bernd and Friederike Hünnemeyer. . On the fate of Ewe ví
‘child’—the development of a diminutive marker. AAP (Afrikanistische
Arbeitspapiere, Cologne) , –.
Heine, Bernd, Gunther Kaltenböck, and Sylvie Hancil. . Observations sur
la transcatégorialité. In: Danh-Thành Do-Hurinville, Huy-Linh Dao, and
Annie Rialland (eds.) De la transcatégorialité dans les langues. Description,
modélisation, typologie, –. Paris: Editions de la Société de Linguistique
de Paris.
Heine, Bernd, Gunther Kaltenböck, Tania Kuteva, and Haiping Long. .
The Rise of Discourse Markers. Cambridge: Cambridge University Press.
Heine, Bernd and Tania Kuteva. . World Lexicon of Grammaticalization.
Cambridge: Cambridge University Press.
Heine, Bernd and Tania Kuteva. . Contact-induced grammaticalization.
Studies in Language /, –.
Heine, Bernd and Tania Kuteva. . Language Contact and Grammatical
Change. Cambridge: Cambridge University Press.
Heine, Bernd and Tania Kuteva. . The Changing Languages of Europe.
Oxford: Oxford University Press.
Heine, Bernd and Tania Kuteva. . The Genesis of Grammar: A Reconstruc-
tion. Oxford: Oxford University Press.
Heine, Bernd and Hiroyuki Miyashita. . Accounting for a functional
category: German drohen ‘to threaten’. Language Sciences , –.
Heine, Bernd and Motoki Nomachi. . Contact-induced replication: Some
diagnostics. In: Robbeets, Martine and Hubert Cuyckens (eds.) Shared
Grammaticalization, –. Amsterdam: Benjamins.
Heine, Bernd and Mechthild Reh. . Grammaticalization and Reanalysis in
African Languages. Hamburg: Buske.
Heine, Bernd, Ulrike Claudi, and Friederike Hünnemeyer. . Grammatical-
ization: A Conceptual Framework. Chicago: University of Chicago Press.
Heine, Bernd, Gunther Kaltenböck, Tania Kuteva, and Haiping Long. .
An outline of discourse grammar. In: Bischoff, Shannon and Carmen Jany (eds.)
Functional Approaches to Language, –. Berlin: Mouton de Gruyter.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Heine, Bernd, Heiko Narrog, and Haiping Long. . Constructional change
vs. grammaticalization: From compounding to derivation. Studies in
Language /, –.
Heine, Bernd, Gunther Kaltenböck, Tania Kuteva, and Haiping Long. .
Cooptation as a discourse strategy. Linguistics , –.
Heine, Bernd, Tania Kuteva, and Heiko Narrog. . Back again to the future:
How to account for directionality in grammatical change? In: Bisang,
Walter and Andrej Malchukov (eds.) Unity and Diversity in Grammatical-
ization Scenarios, –. Berlin: Language Science Press.
Heine, Bernd, Tania Kuteva, Haiping Long, Heiko Narrog, and Fang Wu.
. Where do demonstratives come from? Language Typology and
Universals , –.
Helbig, Gerhard. . Entwicklung der Sprachwissenschaft seit . Leipzig:
VEB.
Hengeveld, Kees. . Layers and operators in Functional Grammar. Journal
of Linguistics , –.
Hengeveld, Kees. . The grammaticalization of tense and aspect. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Hengeveld, Kees. . A hierarchical approach to grammaticalization. In:
Hengelveld, Kees, Heiko Narrog, and Hella Olbertz (eds.) The Grammat-
icalization of Tense, Aspect, Modality and Evidentiality: A Functional
Perspective, –. Berlin: Mouton de Gruyter.
Hengeveld, Kees and J. Lachlan Mackenzie. . Functional Discourse Gram-
mar: A Typologically-Based Theory of Language Structure. Oxford: Oxford
University Press.
Hengeveld, Kees and J. Lachlan Mackenzie. . Functional discourse gram-
mar. In: Heine, Bernd and Heiko Narrog (eds.) The Oxford Handbook of
Linguistic Analysis, –. Oxford: Oxford University Press.
Hengelveld, Kees, Heiko Narrog, and Hella Olbertz (eds.). . The Gram-
maticalization of Tense, Aspect, Modality and Evidentiality: A Functional
Perspective. Berlin: Mouton de Gruyter.
Herring, Susan C. . The grammaticalization of rhetorical questions in
Tamil. In: Traugott, Elizabeth C. and Bernd Heine (eds.) Approaches to
Grammaticalization vol. I, –. Amsterdam: Amsterdam.
Higashiizumi, Yuko. . From a Subordinate Clause to an Independent
Clause: A History of English because-clause and Japanese kara-clause.
Tokyo: Hituzi Syobo.
Hilpert, Martin. . Germanic Future Constructions: A Usage-Based
Approach to Language Change. Amsterdam: Benjamins.
Hilpert, Martin. a. Constructional Change in English: Developments in Allo-
morphy, Word-Formation, and Syntax. Cambridge: Cambridge University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Hilpert, Martin. b. Corpus-based approaches to constructional change. In:
Hoffmann, Thomas and Graeme Trousdale (eds.) The Oxford Handbook of
Construction Grammar. DOI: ./oxfordhb/...
Oxford: Oxford University Press.
Hilpert, Martin and Christian Koops. . A quantitative approach to the
development of complex predicates. The case of Swedish pseudo-
coordination with sitta “sit”*. In: Givón, T. and Masayoshi Shibatani
(eds.) Syntactic Complexity. Diachrony, Acquisition, Neuro-cognition, Evo-
lution, –. Amsterdam: Benjamins.
Hilpert, Martin and David Correia Saavedra. . The unidirectionality of
semantic changes in grammaticalization: An experimental approach to the
asymmetric priming hypothesis. English Language and Linguistics /,
–.
Himmelmann, Nikolaus P. . Grammaticalization and Grammar [Arbeitspa-
piere , Neue Folge]. Cologne: University of Cologne: Dept. of Linguistics.
Himmelmann, Nikolaus P. . Deiktikon, Artikel, Nominalphrase: Zur
Emergenz syntaktischer Struktur. Tübingen: Niemeyer.
Himmelmann, Nikolaus P. . Lexicalization and grammaticalization:
Opposite or orthogonal? In: Bisang, Walter et al. (eds.) What Makes Gram-
maticalization? A Look from its Fringes and its Components, –. Berlin:
Mouton de Gruyter.
Himmelmann, Nikolaus P. . Gram, construction, and class formation. In:
Knobloch, Clemens and Burhard Schaeder (eds.) Wortarten und Gramma-
tikalisierung. Perspektiven in System und Erwerb, –. Berlin: Walter de
Gruyter.
Hock, Hans Henrich and Brian D. Joseph. . Language History, Language
Change, and Language Relationship: An Introduction to Historical and
Comparative Linguistics. nd revised edition. Berlin: Mouton de Gruyter.
Hodge, Carleton T. . The linguistic cycle. Language Sciences , –.
Hoefler, Stefan H. and Andrew D. M. Smith. . The pre-linguistic basis of
grammaticalisation: A unified approach to metaphor and reanalysis. Studies
in Language /, –.
Hoffmann, Sebastian. . Are low-frequency complex prepositions gram-
maticalized? On the limits of corpus data—and the importance of intuition.
In: Lindquist, Hans and Christian Mair (eds.) Corpus Approaches to Gram-
maticalization in English, –. Amsterdam: Benjamins.
Holm, John A. . Pidgins and Creoles. Volume I: Theory and Structure.
Cambridge: Cambridge University Press.
Holm, John A. . An Introduction to Pidgins and Creoles. Cambridge:
Cambridge University Press.
Hopper, Paul J. . Where do words come from? In: Croft, William, Keith
Denning, and Suzanne Kemmer (eds.) Studies in Typology and Diachrony.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Papers Presented to Joseph H. Greenberg on his th Birthday, –.
Amsterdam: Benjamins.
Hopper, Paul J. . On some principles of grammaticization. In: Traugott,
Elizabeth C. and Bernd Heine (eds.) Approaches to Grammaticalization.
Volume , –. Amsterdam: Benjamins.
Hopper, Paul J. . Phonogenesis. In: Pagliuca, William (ed.) Perspectives on
Grammaticalization, –. Amsterdam: Benjamins.
Hopper, Paul J. and Sandra A. Thompson (eds.). . Studies in Transitivity.
New York: Academic Press.
Hopper, Paul J. and Sandra A. Thompson. . The discourse basis for lexical
categories in universal grammar. Language / , –.
Hopper, Paul J. and Elizabeth C. Traugott. . Grammaticalization. Cam-
bridge: Cambridge University Press.
Hopper, Paul J. and Elizabeth C. Traugott. . Grammaticalization. nd
revised edition. Cambridge: Cambridge University Press.
Hornstein, Norbert. . Logic as Grammar. Cambridge, MA: MIT Press.
Huang, Yan. . Pragmatics. Oxford: Oxford University Press.
Huddleston, Rodney and Geoffrey K. Pullum. . The Cambridge Grammar
of the English Language. Cambridge: Cambridge University Press.
Hüning, Matthias and Gert Booij. . From compounding to derivation.
The emergence of derivational affixes through “constructionalization”.
Folia Linguistica /, –.
Hurch, Bernhard. . Hispanisierung im Baskischen. In: Boretzky, Norbert,
Werner Enninger, and Thomas Stolz (eds.) Beiträge zum . Essener Kollo-
quium über Grammatikalisierung: Natürlichkeit und Systemökonomie,
Volume , –. Bochum: Brockmeyer.
Hwang, Jya-Lin. . On grammaticalization in serial verb constructions in
Chinese. Ph.D. dissertation. University of Hawaii.
Itkonen, Esa. . Analogy as Structure and Process: Approaches in Linguistics,
Cognitive Psychology and Philosophy of Science. Amsterdam: Benjamins.
Jäger, Gerhard and Anette Rosenbach. . Priming and unidirectional
language change. Theoretical Linguistics /, –.
Jakobson, Roman. . Linguistics and poetics. In: Sebeok, T. A. (ed.) Style in
Language, –. Cambridge, MA: MIT Press.
Jamison, Stephanie. . Functional ambiguity and syntactic change. Papers
from the parasession on diachronic syntax April , –. Chicago:
Chicago Linguistic Society.
Jamison, Stephanie. . Function and Form in the -áya-Formations of the
Rig Veda and Atharva Veda. Göttingen: Vandenhoeck & Ruprecht.
Janda, Laura A. . Back From the Brink. München: Lincom.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Janda, Richard D. . Beyond “pathways” and “unidirectionality”: on the
discontinuity of language transmission and the counterability of grammat-
icalization. Language Sciences /–, –.
Jendraschek, Gerd. . Basque in contact with Romance languages. In:
Aikhenvald, Alexandra Y. and R.M.W. Dixon (eds.) Grammars in Contact:
A Cross-Linguistic Typology, –. Oxford: Oxford University Press.
Jenkins, Rebecca Sue. . Language Contact and Composite Structures
in New Ireland, Papua New Guinea. Ann Arbor: UMI Dissertation
Services.
Jespersen, Otto. . Negation in English and Other Languages. Copenhagen:
Høst og Søn.
Jespersen, Otto. . Language: Its Nature, Origin and Development. London:
George Allen and Unwin Ltd.
Johanson, Lars. . Strukturelle Faktoren in türkischen Sprachkontakten.
Stuttgart: Franz Steiner.
Johanson, Lars. . Structural Factors in Turkic Language Contacts. London:
Curzon.
Johanson, Lars. . Remodeling grammar: Copying, conventionalization,
grammaticalization. In: Siemund, Peter and Kintana, Noemi (eds.)
Language Contact and Contact Languages, –. Amsterdam: Benjamins.
Johanson, Lars and Éva Csató. . Grammaticalization in Turkic. In:
Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a Typo-
logical Perspective. –. Oxford: Oxford University Press.
Jones, Mari C. . Mette a haout dauve la grippe des Angllais: convergence
on the Island of Guernsey. In: Jones, Mari C. and Edith Esch (eds.)
Language Change: The Interplay of Internal, External and Extra-Linguistic
Factors, –. Berlin: Mouton de Gruyter.
Joseph, Brian D. . Is there such a thing as “grammaticalization”? Lan-
guage Sciences /–, –.
Joseph, Brian D. . Morphologization from syntax. In: Joseph, Brian D.
and Richard D. Janda (eds.) The Handbook of Historical Linguistics,
–. Oxford: Blackwell.
Joseph, Brian D. . How accommodating of change is grammaticalization?
The case of “lateral shifts”. Logos and Language /, –.
Joseph, Brian D. and Richard Janda. . The how and why of diachronic
morphologization and demorphologization. In: Hammond, Michael and
Michael Noonan (eds.) Theoretical Morphology: Approaches in Modern
Linguistics, –. New York: Academic Press.
Jucker, Andreas H. . The discourse marker well: a relevance-theoretical
account. Journal of Pragmatics /, –.
Jucker, Andreas H. . The discourse marker well in the history of English.
English Language and Linguistics /, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Jucker, Andreas H. and Yael Ziv (eds.) . Discourse Markers: Description
and Theory. Amsterdam: Benjamins.
Juge, Matthew L. . Metaphor and teleology do not drive grammatical-
ization. In: Salmons, Joseph C. and Shannon Dubenion-Smith (eds.) His-
torical Linguistics : Selected Papers From the th International
Conference On Historical Linguistics, –. Amsterdam: Benjamins.
Kailuweit, Rolf. . Exaptation, refunctionalization, decapitalization—BE +
past participle with intransitive verbs in Medieval and Early Modern
Spanish. Languages, , .
Kaltenböck, Gunther. . Spoken parenthetical clauses in English. In: Dehé,
Nicole and Yordanka Kavalova (eds.) Parentheticals, –. Amsterdam:
Benjamins.
Kaltenböck, Gunther. . Pragmatic functions of parenthetial I think. In:
Kaltenböck, Gunther, Gudrun Mihatsch, and Stefan Schneider (eds.) New
Approaches to Hedging, –. Bingley: Emerald Publishers.
Kaltenböck, Gunther, Bernd Heine, and Tania Kuteva. . On thetical
grammar. Studies in Language /, –.
Kastovsky, Dieter. . Historical morphology from a typological point of
view. Examples from English. In: Nevalainen, Terttu, Juhani Klemola, and
Mikko Laitinen (eds.) Types of Variation: Diachronic, Dialectal and Typo-
logical Interfaces, –. Amsterdam: Benjamins.
Kearns, Kate. . lmplicature and language change. In: Fried, Mirjam et al.
(eds.) Variation and Change: Pragmatic Perspectives, –. Amsterdam:
Benjamins.
Keesing, Roger M. . Substrates, calquing and grammaticalization in Mela-
nesian Pidgin. In: Traugott, Elizabeth C. and Bernd Heine (eds.) Approaches
to Grammaticalization, Volume , –. Amsterdam: Benjamins.
Keizer, Evelien. . The lexical-grammatical dichotomy in Functional
Discourse Grammar. In: Marize Mattos Dall’Aglio Hattnher and Kees
Hengeveld (eds.) Advances in Functional Discourse Grammar. Special
issue of Alfa—Revista de Lingüística /, –.
Keller, Rudi. . Language Change: The Invisible Hand in Language. London:
Routledge.
Keller, Rudi. . In what sense can explanations of language change be
functional? In: Gvozdanovic, Jadranka (ed.) Language Change and Func-
tional Explanations, –. Berlin: Mouton de Gruyter.
Kemmer, Suzanne and Arie Verhagen. . The grammar of causatives and
the conceptual structure of events. Cognitive Linguistics /, –.
Kinuhata, Tomohide. . Historical development from subjective to object-
ive meaning: evidence from the Japanese question particle ka. Journal of
Pragmatics , –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Kiparsky, Paul. . Grammaticalization as optimization. In: Jonas, Dianne,
John Whitman, and Andrew Garrett (eds.) Grammatical Change: Origins,
Nature, Outcomes, –. Oxford: Oxford University Press.
Kissine, Mikhail. . Metaphorical projection, subjectification and English
speech act verbs. Folia Linguistica /, –.
Klamer, Marian. . How report verbs become quote markers and comple-
mentisers. Lingua , –.
Klamer, Marian. . Typology and grammaticalization in the Papuan lan-
guages of Timor, Alor, and Pantar. In: Narrog, Heiko and Bernd Heine
(eds.) Grammaticalization from a Typological Perspective. –. Oxford:
Oxford University Press.
Klausenburger, Jürgen. . Grammaticalization within a theory of morpho-
centricity. In: Wischer, Ilse and Gabriele Diewald (eds.) New Reflections on
Grammaticalization, –. Amsterdam: Benjamins.
König, Ekkehard. . Manner deixis as source of grammatical markers in
Indo-European languages. In: Viti, Carlotta (ed.) Perspectives on Historical
Syntax, –. Amsterdam: Benjamins.
König, Ekkehard and Bernd Kortmann. . On the reanalysis of verbs as
prepositions. In: Rauh, Gisa (ed.) Approaches to Prepositions, –.
Tübingen: Narr.
König, Ekkehard and Peter Siemund. . Intensifiers and reflexives: A
typological perspective. In: Frajzyngier, Zygmunt and Traci S. Curl (eds.)
Reflexives: Forms and Functions (Typological Studies in Language, .),
–. Amsterdam, Philadelphia: Benjamins.
Kortmann, Bernd and Ekkehard König. . Categorial reanalysis: the case of
deverbal prepositions. Linguistics , –.
Kortmann, Bernd and Agnes Schneider. . Grammaticalization in non-
standard varieties of English. In: Narrog, Heiko and Bernd Heine (eds.) The
Oxford Handbook of Grammaticalization, –. Oxford: Oxford Univer-
sity Press.
Koyanagi, Tomokazu. . Bunpō henka no kenkyū [The study of grammat-
ical change]. Tokyo: Kuroshi Shuppan.
Kranich, Svenja. a. The Progressive in Modern English: A Corpus-Based
Study of Grammaticalization and Related Changes. Amsterdam: Rodopi.
Kranich, Svenja. b. Grammaticalization, subjectification and objectifica-
tion. In: Stathi, Katerina et al. (eds.) Grammaticalization. Current Views
and Issues, –. Amsterdam: Benjamins.
Kranich, Svenja. . The impact of input and output domains: towards a
function-based categorization of types of grammaticalization. Language
Sciences , –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Krug, Manfred G. . Emerging English Modals: A Corpus-based Study of
Grammaticalization. Berlin: Mouton de Gruyter.
Krug, Manfred. . Frequency as a determinant in grammatical variation
and change. In: Rohdenburg, Günter and Britta Mondorf (eds.) Determin-
ants of Grammatical Variation in English, –. Berlin: Mouton de Gruyter.
Kuryɬowicz, Jerzy. []. The evolution of grammatical categories.
Reprinted in Kuryɬowicz, Jerzy. Esquisses linguistiques. Volume , –.
Munich: Fink.
Kuteva, Tania. a. On identifying an evasive gram: action narrowly
averted. Studies in Language /, –.
Kuteva, Tania. b. Large linguistic areas in grammaticalization: auxiliation
in Europe. Language Sciences /, –.
Kuteva, Tania. . On ‘sit’/’stand’/’lie’ auxiliation. Linguistics /, –.
Kuteva, Tania. . Areal grammaticalization: The case of the Bantu-Nilotic
borderland. Folia Linguistica /–, –.
Kuteva, Tania. . Auxiliation: An Enquiry into the Nature of Grammat-
icalization. Oxford: Oxford University Press.
Kuteva, Tania. . On the “frills” of grammaticalization. In: López-Couso,
María José and Elena Seoane (eds.) Rethinking Grammaticalization: New
Perspectives for the Twenty-first Century, –. Amsterdam: Benjamins.
Kuteva, Tania and Bernd Heine. . On the explanatory value of grammat-
icalization. In: Good, Jeff (ed.) Linguistic Universals and Language Change,
–. Oxford: Oxford University Press.
Kuteva, Tania, Bernd Heine, Bo Hong, Haiping Long, Heiko Narrog, and
Seongha Rhee. . World Lexicon of Grammaticalization. Second, exten-
sively revised and updated edition. Cambridge: Cambridge University Press.
Labov, William. . Principles of Linguistic Change, Volume I: Internal
Factors. Oxford: Blackwell.
Labov, William. . Principles of Linguistic Change, Volume II: Social
Factors. Oxford: Blackwell.
Lai, Vicky Tzuyin, Gabriela Garrido Rodriguez, and Bhuvana Narasimhan.
. Thinking-for-speaking in early and late bilinguals. Bilingualism:
Language and Cognition /, –.
Lamiroy, Béatrice and Walter de Mulder. . Degrees of grammaticalization
across languages. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford
Handbook of Grammaticalization, –. Oxford: Oxford University
Press.
Langacker, Ronald W. . Syntactic reanalysis. In: Li, Charles N. (ed.)
Mechanisms of Syntactic Change, –. Austin: University of Texas Press.
Langacker, Ronald W. . Observations and speculations on subjectivity. In:
Haiman, John (ed.) Iconicity in Syntax, –. Amsterdam: Benjamins.
Langacker, Ronald W. . Subjectification. Cognitive Linguistics /, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Langacker, Ronald W. . On Subjectification and grammaticization. In:
Koenig, Jean-Pierre (ed.) Discourse and Cognition: Bridging the Gap, –.
Stanford: CSLI Publications.
Langacker, Ronald W. . Losing control: grammaticization, subjectifica-
tion, and transparency. In: Blank, Andreas and Peter Koch (eds.) Historical
Semantics and Cognition, –. Berlin: Mouton de Gruyter.
Langacker, Ronald W. . Deixis and subjectivity. In: Brisard, Frank (ed.)
Grounding: The Epistemic Footing of Deixis and Reference, –. Berlin:
Mouton de Gruyter.
Langacker, Ronald W. . Extreme subjectification: English tense and
modals. In: Cuyckens, Hubert et al. (eds.) Motivation in Language: Studies
in Honor of Günter Radden, –. Amsterdam: Benjamins.
Langacker, Ronald W. . Subjectification, grammaticization, and concep-
tual archetypes. In: Athanasiadou, Angeliki, Costas Canakis, and Bert
Cornillie (eds.) Subjectification: Various Paths to Subjectivity, –. Berlin:
Mouton de Gruyter.
Langacker, Ronald W. . Grammaticalization and Cognitive Grammar. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Lass, Roger. . How to do things with junk: exaptation in language
evolution. Journal of Linguistics , –.
Lass, Roger. . Historical Linguistics and Language Change. Cambridge:
Cambridge University Press.
Ledgeway, Adam. . Grammaticalization from Latin to Romance. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Lehmann, Christian. . Thoughts on Grammaticalization: A programmatic
Sketch. Volume . AKUP (Arbeiten des Kölner Universalien-Projekts).
Cologne: Universität zu Köln, Institut für Sprachwissenschaft.
Lehmann, Christian. . Grammaticalization: synchronic variation and
diachronic change. Lingua e Stile /, –.
Lehmann, Christian. . Grammaticalization and linguistic typology. Gen-
eral Linguistics /, –.
Lehmann, Christian. . Grammatikalisierung und Lexikalisierung. Zeitschrift
für Phonetik, Sprachwissenschaft und Kommunikationsforschung /, –.
Lehmann, Christian. . New reflections on grammaticalization and lexi-
calization. In: Wischer, Ilse and Gabriele Diewald (eds.) New Reflections on
Grammaticalization, –. Amsterdam: Benjamins.
Lehmann, Christian. . Theory and method in grammaticalization.
Zeitschrift für Germanistische Linguistik , –.
Lehmann, Christian. []. Thoughts on Grammaticalization. rd edi-
tion. Berlin: Language Science Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Lenker, Ursula. . Soþlice and witodlice: Discourse markers in Old
English. In: Olga Fischer, Anette Rosenbach and Dieter Stein (eds.) Path-
ways of Change: Grammaticalization in English, –. Amsterdam:
Benjamins.
Lepore, Ernie and Matthew Stone. . Imagination and Convention: Distin-
guishing Grammar and Inference in Language. Oxford: Oxford University
Press.
Lessau, Donald Andreas. . A Dictionary of Grammaticalization. Three
volumes. Bochum: Universitätsverlag Dr. N. Brockmeyer.
Levinson, Stephen C. . Pragmatics. Cambridge: Cambridge University Press.
Lichtenberk, Frantisek.. Semantic change and heterosemy in grammat-
icalization. Language /, –.
Lightfoot, David. . Principles of Diachronic Syntax. Cambridge:
Cambridge University Press.
Lightfoot, David. . How to Set Parameters: Arguments from Language
Change. Cambridge, MA: MIT Press.
Lightfoot, Douglas J. . On the grammaticalization of a nominalizing suffix
in German: (-)tum. Ph.D. thesis. University of California, Los Angeles.
Lightfoot, Douglas J. . Can the lexicalization/grammaticalization distinc-
tion be reconciled? Studies in Language , –.
Lightfoot, Douglas J. . Grammaticalization and lexicalization. In: Narrog,
Heiko and Bernd Heine (eds.) Handbook of Grammaticalization, –.
Oxford: Oxford University Press.
Lima, José Pinto de. . Zur Grammatikalisierung von dt. drohen und pg.
ameaçar. In: Schmidt-Radefeldt, Jürgen (ed.) Portugiesisch kontrastiv gese-
hen und Anglizismen weltweit, –. Frankfurt: Peter Lang.
López-Couso, María José. . Subjectification and intersubjectification. In:
Andreas Jucker and Irma Taavitsainen (eds.) Historical Pragmatics,
–. Berlin: Mouton de Gruyter.
López-Couso, María José. a. Continuing the dialogue between corpus
linguistics and grammaticalization theory: three case studies. Corpus Lin-
guistics and Linguistic Theory /, –.
López-Couso, María José. b. Corpora and online resources in English
historical linguistics. In: Kytö, Merja, and Päivi Pahta (eds.) The Cambridge
Handbook of English Historical Linguistics, –. Cambridge: Cambridge
University Press.
Lord, Carol. . Evidence for syntactic reanalysis: from verb to comple-
mentizer in Kwa. In: Steever, Sanford B., Carol A. Walker, and Salikoko
Mufwene (eds.) Papers from the Parasession on Diachronic Syntax: April ,
, –. Chicago: Chicago Linguistic Society.
Lord, Carol. . Syntactic Reanalysis in the Historical Development of Serial
Verb Constructions in Languages of West Africa. Ph.D. Dissertation,
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
University of California, Los Angeles. Ann Arbor, Michigan: University
Microfilms International.
Lord, Carol Diane. . Historical Change in Serial Verb Constructions.
Amsterdam: Benjamins.
Luraghi, Silvia. . Some remarks on instrument, comitative, and agent in
Indo-European. Sprachtypologie und Universalienforschung /, –.
Luraghi, Silvia. . Causes of language change. In: Luraghi, Silvia and Vit
Bubenik (eds.) Continuum Companion to Historical Linguistics, –.
New York: Continuum.
Luraghi, Silvia and Vit Bubenik (eds.). . Continuum Companion to
Historical Linguistics. New York: Continuum.
Lyons, John. . Glossary of technical terms. In: Lyons, John (ed.) New
Horizons in Linguistics, –. Hammondsworth: Penguin.
Lyons, John. . Semantics. Two volumes. Cambridge: Cambridge Univer-
sity Press.
Lyons, John. . Deixis and subjectivity: Loquor, ergo sum? In: Jarvella,
Robert J. and Wolfgang Klein (eds.) Speech, Place, and Action: Studies in
Deixis and Related Topics, –. New York: Wiley.
Maat, Henk Pander. . Subjectification in gradable adjectives. In:
Athanasiadou, Angeliki, Costas Canakis, and Bert Cornillie (eds.) Subjecti-
fication: Various Paths to Subjectivity, –. Berlin: Mouton de Gruyter.
Mair, Christian. . Corpus linguistics and grammaticalization theory: Statistics,
frequency, and beyond. In: Lindquist, H. and C. Mair (eds.) Corpus Approaches
to Grammaticalization in English, –. Amsterdam: Benjamins.
Mair, Christian. . Grammaticalization and corpus linguistics. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Maslova, Elena. . Unidirectionality of grammaticalization in an evolu-
tionary perspective. In: Verhoeven, Elisabeth, Stavros Skopeteas, Yong-Min
Shin, Yoko Nishina, and Johannes Helmbrecht (eds.) Studies on Grammat-
icalization, –. Mouton de Gruyter.
Matasović, Rolf. . Patterns of grammaticalization and the layered struc-
ture of the clause. In: Kailuweit, Rolf, Björn Wiemer, E. Staudinger, and
Rolf Matasović (eds.) New Applications of Role and Reference Grammar:
Diachrony, Grammaticalization, Romance Languages, –. Newcastle:
Cambridge Scholars Publishing.
Matisoff, James A. . Lahu causative constructions: case hierarchies and
the morphology/syntax cycle in a Tibeto-Burman perspective. In: Shiba-
tani, Masayoshi (ed.) Syntax and Semantics Volume , –. New York:
Academic Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Matisoff, James A. . Variational Semantics in Tibeto-Burman: The
“Organic” Approach to Linguistic Comparison. Philadelphia: Institute for
the Study of Human Issues.
Matras, Yaron. . Grammaticalization and language contact. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Matsumoto, Yo. . From bound grammatical markers to free discourse
markers: History of some Japanese connectives. Proceedings of the
Fourteenth Annual Meeting of the Berkeley Linguistics Society, –.
McEnery, Tony and Andrew Hardie. . Corpus Linguistics: Method, Theory
and Practice. Cambridge: Cambridge University Press.
McNeill, David and Susan Duncan. . Growth points in thinking-for-
speaking. In: McNeill, David (ed.) Language and Gesture, –.
Cambridge: Cambridge University Press.
McWhorter, John. . Is grammaticalization in creoles different? In:
Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a Typo-
logical Perspective. –. Oxford: Oxford University Press.
Meillet, Antoine. [] . L’évolution des formes grammaticales. Scientia
(Rivista di Scienza vol. , Nr. -). Reprinted in Meillet, Antoine .
Linguistique historique et linguistique générale, –. Paris: Champion.
Miller, George A. and Philip N. Johnson-Laird. . Language and Perception.
Cambridge, MA: Harvard University Press.
Minami, Fujio. . Gendai nihongo no kōzō [The Structure of Modern
Japanese]. Tōkyō: Taishūkan Shoten.
Mirčev, Kiril. . Istoričeska gramatika na bǎlgarskija ezik. Sofia: Nauka i Izkustvo.
Mitchell, Bruce. . Old English Syntax, Volume II: Subordination, Inde-
pendent Elements, and Element Order. Oxford: Clarendon Press.
Mithun, Marianne. . The reordering of morphemes. In: Gildea, Spike
(ed.) Reconstructing Grammar: Comparative Linguistics and Grammatical-
ization, –. Amsterdam: Benjamins.
Mithun, Marianne. . An invisible hand at the root of causation: The role
of lexicalization in the grammaticalization of causatives. In: Wischer, Ilse
and Gabriele Diewald (eds.) New Reflections on Grammaticalization,
–. Amsterdam: Benjamins.
Mithun, Marianne. . Grammaticalization and explanation. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Mithun, Marianne. . Shaping typology through grammaticalization: North
America. In: Narrog, Heiko and Bernd Heine (eds.) Grammaticalization
from a Typological Perspective. –. Oxford: Oxford University Press.
Moravcsik, Edith A. . Introducing Language Typology. Cambridge:
Cambridge University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Morfill, W. R. [] . Review of Schuchardt . The Academy,
Edinburgh, . IV, , –. Reprinted in Gerhardt, Dietrich (ed.)
. Hugo Schuchardt, Slawo-deutsches und Slawo-italienisches. Mit Schu-
chardts übrigen Arbeiten zur Slavistik und mit neuen Registern, –.
Munich: Wilhelm Fink.
Moser, Rosemarie. . Grammaticalization chains of the verb kàre ‘to give’
in Kabba. In: Voeltz, F.K. Erhard (ed.) Studies in African Linguistic Typ-
ology, –. Amsterdam: Benjamins.
Moyse-Faurie, Claire. . Grammaticalization in Oceanic languages. In:
Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a Typo-
logical Perspective. –. Oxford: Oxford University Press.
Mufwene, Salikoko S. . The Ecology of Language Evolution. Cambridge:
Cambridge University Press.
Mühlhäusler, Peter. . Pidgin and Creole Linguistics. Oxford: Blackwell.
Mushin, Ilana. . Grammaticalization and typology in Australian Abori-
ginal languages: Evidence from second position clitic constructions. In:
Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a Typo-
logical Perspective, –. Oxford: Oxford University Press.
Nakayama, Toshihide and Kumiko Ichihashi-Nakayama. . Japanese kedo:
Discourse genre and grammaticization. In: Sohn, Ho-min and John Haig
(eds.) Japanese/Korean Linguistics , –. Stanford: CSLI Publications.
Narrog, Heiko. . From Transitive to Causative in Japanese: Morphologi-
zation through exaptation. Diachronica /, –.
Narrog, Heiko. . Exaptation, Grammaticalization, and Reanalysis.
California Linguistic Notes /.
Narrog, Heiko. a. A diachronic dimension in maps of case functions.
Linguistic Discovery /, –.
Narrog, Heiko. b. (Inter)subjectification in the domain of modality and
mood—Concepts and cross-linguistic realities. In: Davidse, Kristin, Lieven
Vandelanotte, and Hubert Cuyckens (eds.) Subjectification, Intersubjectifi-
cation and Grammaticalization, –. Berlin: De Gruyter.
Narrog, Heiko. c. Voice and non-canonical case marking in the expres-
sion of event-oriented modality. Linguistic Typology , –.
Narrog, Heiko. a. Modality, Subjectivity, and Semantic Change: A Cross-
Linguistic Perspective. Oxford: Oxford University Press.
Narrog, Heiko. b. Beyond intersubjectification. Textual uses of modality
and mood in subordinate clauses as part of speech act orientation. English
Text Construction /, –.
Narrog, Heiko. . The grammaticalization chain of case functions.
Extension and reanalysis of case marking vs. universals of grammatical-
ization. In: Luraghi, Silvia and Heiko Narrog (eds.) Perspectives on Semantic
Roles, –. Amsterdam: Benjamins.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Narrog, Heiko. . (Inter)subjectification and its limits in secondary gram-
maticalization. Language Sciences , –.
Narrog, Heiko. . Exaptation in Japanese and beyond. In: Norde, Muriel
and Freek van de Velde (eds.) Exaptation and Language Change, –.
Amsterdam: Benjamins.
Narrog, Heiko. a. Typology and grammaticalization. In: Aikhenvald,
Alexandra Y. and R.M.W. Dixon (eds.) The Cambridge Handbook of
Linguistic Typology, –. Cambridge: Cambridge University Press.
Narrog, Heiko. b. Three types of subjectivity, three types of intersubject-
ivity, their dynamicization and a synthesis. In: van Olmen, Daniel, Hubert
Cuyckens, and Lobke Ghesquière (eds.) Aspects of Grammaticalization.
(Inter)Subjectification and Directionality, –. Berlin: Mouton de Gruyter.
Narrog, Heiko. c. Relationship of form and function in grammatical-
ization – the case of modality. In: Hengeveld, Kees, Heiko Narrog, and
Hella Olbertz (eds.) The Grammaticalization of Tense, Aspect, Modality and
Evidentiality, –. Berlin: Mouton de Gruyter.
Narrog, Heiko. d. The morphosyntax of grammaticalization in Japanese.
In: Shibatani, Masyoshi, Shigeru Miyagawa, and Hisashi Noda (eds.) Hand-
book of Japanese Syntax, –. Berlin: Mouton de Gruyter.
Narrog, Heiko. . Historical change in the Japanese tense-aspect system.
In: Journal of Historical Linguistics /, –.
Narrog, Heiko. . Scope and unidirectionality in grammaticalization—the
anatomy and history of a misunderstanding. Manuscript under review.
Narrog, Heiko and Bernd Heine (eds.) . The Oxford Handbook of Gram-
maticalization. Oxford: Oxford University Press.
Narrog, Heiko and Bernd Heine . Grammaticalization. In: Ledgeway,
Adam and Ian Roberts (eds.) The Cambridge Handbook of Historical
Syntax, –. Cambridge: Cambridge University Press.
Narrog, Heiko and Bernd Heine. . Typology and grammaticalization. In:
Narrog, Heiko and Bernd Heine (eds.) Grammaticalization from a Typo-
logical Perspective, –. Oxford: Oxford University Press.
Narrog, Heiko and Bernd Heine (eds.). . Grammaticalization from a
Typological Perspective. Oxford: Oxford University Press.
Narrog, Heiko and Shinya Ito. . Reconstructing semantic maps. The
comitative-instrumental area. Sprachtypologie und Universalienforschung
/, –.
Narrog, Heiko and Toshio Ohori. . Grammaticalization in Japanese. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Narrog, Heiko and Johan van der Auwera. . Grammaticalization and
semantic maps. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford Hand-
book of Grammaticalization, –. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Narrog, Heiko, Seongha Rhee, and John Whitman. . Grammaticalization
in Japanese and Korean. In: Narrog, Heiko and Bernd Heine (eds.) Gram-
maticalization from a Typological Perspective. –. Oxford: Oxford
University Press.
Nevalainen, Terttu. . Social variation in intensifier use: constraint on -ly
adverbialization?’ English Language and Linguistics /, –.
Nevalainen, Terttu and Minna Palander-Collin. . Grammaticalization
and sociolinguistics. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford
Handbook of Grammaticalization, –. Oxford: Oxford University
Press.
Newmeyer, Frederick J. . Language Form and Language Function.
Cambridge, MA: MIT Press.
Nichols, Johanna, David A. Peterson, and Jonathan Barnes. . Transitiviz-
ing and detransitivizing languages. Linguistic Typology , –.
Nicolle, Steve. . A relevance theory perspective on grammaticalization.
Cognitive Linguistics /, –.
Nicolle, Steve. . Pragmatic Aspects of Grammatialization. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Nicolle, Steve. . Diachrony and grammaticalization. In: Binnick, Robert
(ed.) The Oxford Handbook of Tense and Aspect, –. Oxford: Oxford
University Press.
NKD , = Nihon Kokugo Daijiten [Great Dictionary of the Japanese Lan-
guage], Volumes and . nd edition. . Tokyo: Shogakukan.
Noël, Dirk. . Diachronic construction grammar and grammaticalization
theory. Functions of Language /, –.
Nomachi, Motoki. . Placing Kashubian on the language map of Europe.
In: Danylenko, Andrii and Motoki Nomachi (eds.) Slavic on the Language
Map of Europe: Historical and Areal-Typological Dimensions, –.
Berlin: De Gruyter Mouton.
Norde, Muriel. . Deflexion as a counterdirectional factor in grammatical
change. Language Sciences /–, –.
Norde, Muriel. . The final stages of grammaticalization: affixhood and
beyond. In: Wischer, Ilse and Gabriele Diewald (eds.) New Reflections on
Grammaticalization, –. Amsterdam: Benjamins.
Norde, Muriel. . Degrammaticalization. Oxford: Oxford University Press.
Norde, Muriel. . Degrammaticalization: Three common controversies. In:
Stathi, Katerina, Elke Gehweiler, and Ekkehard König (eds.) Grammatical-
ization: Current Views and Issues, –. Amsterdam: Benjamins.
Norde, Muriel. . Lehmann’s parameters revisited. In: Davidse, Kristin,
Tine Breban, Liselotte Brems, L. and Tania Mortelmans (eds.) Grammat-
icalization and Language Change: New Reflections, –. Amsterdam:
Benjamins.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Norde, Muriel and Karin Beijering. . Facing interfaces: A clustering
approach to grammaticalization and related changes. Folia Linguistica
/, –.
Norde, Muriel and Freek van de Velde. (eds.) . Exaptation and Language
Change. Amsterdam: Benjamins.
Nørgård-Sørensen, Jens and Lars Heltoft. . Grammaticalisation as para-
digmatisation. In: Smith, Andrew D.M. et al. (eds.) New Directions in
Grammaticalization Research, –. Amsterdam: Benjamins.
Nørgård-Sørensen, Jens, Lars Heltoft, and Lene Schøsler. . Connecting
Grammaticalisation. Amsterdam: Benjamins.
Nuyts, Jan. a. Epistemic Modality, Language, and Conceptualization.
Amsterdam: Benjamins.
Nuyts, Jan. b. Subjectivity as an evidential dimension in epistemic modal
expressions. Journal of Pragmatics , –.
Ocampo, Francisco. . Movement towards discourse is not grammatical-
ization: The evolution of /claro/ from adjective to discourse particle in
spoken Spanish. In: Sagarra, Nura and Almeida Jacqueline Toribio (eds.)
Selected Proceedings of the th Hispanic Linguistics Symposium, –.
Somerville, MA: Cascadilla Proceedings Project.
Onodera, Noriko. . Interplay of (inter)subjectivity and social norm.
Journal of Historical Pragmatics /, –.
Onodera, Noriko. . The grammaticalization of discourse markers. In:
Narrog, Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammat-
icalization, –. Oxford: Oxford University Press.
Onodera, Noriko and Ryoko Suzuki (eds.) . Historical changes in Japan-
ese: With special focus on subjectivity and intersubjectivity. Special issue of
the Journal of Historical Pragmatics /.
Osawa, Fuyo. . Syntactic parallels between ontogeny and phylogeny.
Lingua , –.
Osawa, Fuyo. . The emergence of DP in the history of English. The role of
the mysterious genitive. In: Dufresne, Monique, Fernande Dupuis, and
Etleva Vocaj (eds.) Historical Linguistics . Selected Papers From the
th International Conference On Historical Linguistics, Montreal, -
August , –. Amsterdam: Benjamins.
Park, Insun. . Grammaticalization of verbs in three Tibeto-Burman
languages. Ph.D. Dissertation. University of Oregon, Eugene.
Patten, Amanda L. . The English It-Cleft: A Constructional Account and a
Diachronic Investigation. Berlin: De Gruyter Mouton.
Paul, Herrmann. [] . Principien der Sprachgeschichte. th edition.
Tübingen: Niemeyer.
Paul, Hermann. . Principles of the History of Language. Translated of the
Second Edition of the Original By H.A. Strong, M.A., LL.D. London: Longmans.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Pedersen, Eric. . The Ecology of Semantic Space. Proceedings of the
Seventeenth Annual Meeting of the Berkeley Linguistics Society, –.
Peng, Rui. . Critical frequency as an independent variable in grammat-
icalization. Studies in Language /, –.
Persson, Gunnar. . Homonymy, polysemy and heterosemy: The types of
lexical ambiguity in English. In: Hyldgaard-Jensen, Karl and Arne Zetter-
sten (eds.) Symposium on Lexicography III: Proceedings of the Third Inter-
national Symposium on Lexicography, Copenhagen, – May, –.
Tübingen: Niemeyer.
Peters, Hans. . English boosters: Some synchronic and diachronic aspects.
In: Kellermann, Gunter and Michael D. Morrissey (eds.) Diachrony within
Synchrony: Language History and Cognition, –. Frankfurt: Peter Lang.
Poplack, Shana. . Grammaticalization and linguistic variation. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Poplack, Shana and Elisabete Malvar. . Elucidating the transition period
in linguistic change. Probus /, –.
Poplack, Shana and Rena Torres Cacoullos. . Linguistic emergence on the
ground: A variationist paradigm. In: MacWhinney, Brian and William
O’Grady (eds.) The Handbook of Language Emergence, –. Wiley-
Blackwell.
Poplack, Shana, Rena Torres Cacoullos, Nathalie Dion, Rosane de Andrade
Berlinck, Salvatore Digesto, Dora Lacasse, and Jonathan Steuck. .
Variation and grammaticalization in Romance: a cross-linguistic study of
the subjunctive. In: Ayres-Bennett, Wendy and Janice Carruthers (eds.)
Manual of Romance Sociolinguistics, –. Berlin: Mouton de Gruyter.
Portner, Paul. . Modality. Oxford: Oxford University Press.
Prévost, Sophie. . A propos from verbal complement to discourse marker:
a case of grammaticalization? Linguistics /, –.
Prince, Ellen F. . The borrowing of meaning as a cause of internal
syntactic change. In: Schmid, Monika S., Jennifer R. Austin, and Dieter
Stein (eds.) . Historical Linguistics : Selected Papers from the th
International Conference on Historical Linguistics, Düsseldorf, – August
, –. Amsterdam: Benjamins.
Proudfoot, Michael and Alan Robert Lacey. . The Routledge Dictionary of
Philosophy. th edition. London: Routledge.
Pulvermüller, Friedemann. . The Neuroscience of Language: On Brain
Circuits of Words and Serial Order. Cambridge: Cambridge University Press.
Pustet, Regina. . On discourse frequency, grammar, and grammatical-
ization. In: Frajzyngier, Zygmunt et al. (eds.) Linguistic Diversity and Lan-
guage Theories, –. Amsterdam: Benjamins.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. .
A Comprehensive Grammar of the English Language. London, New York:
Longman.
Ramat, Paolo. . Ein Beispiel von ‘reanalysis’ typologisch betrachtet. Folia
Linguistica , –.
Ramat, Paolo. . Thoughts on degrammaticalization. Linguistics , –.
Ramat, Paolo. . Typological comparison and linguistic areas: some intro-
ductory remarks. Language Sciences /, –.
Ramat, Paolo. . The (early) history of linguistic typology. In: Song, Jae
Jung (ed.) The Oxford Handbook of Linguistic Typology, –. Oxford:
Oxford University Press.
Ramisch, Heinrich. . The Variation of English in Guernsey/Channel
Islands. Frankfurt: Peter Lang.
Rayfield, Joan Rachel. . The Language of a Bilingual Community. The
Hague: Mouton.
Reinöhl, Uta and Nikolaus P. Himmelmann. . Renewal: A figure of speech
or a process sui generis? Language /, –.
Rhee, Seongha. . From discourse to grammar: grammaticalization and
lexicalization of rhetorical questions in Korean. In: Fulton, Gordon,
William J. Sullivan, and Arle R. Lommel (eds.) LACUS: Forum XXX:
Language, Thought and Reality, –. Houston: Lacus.
Rhee, Seongha. . Context-induced reinterpretation and (inter)subjectifi-
cation: the case of grammaticalization of sentence-final particles. Language
Sciences , –.
Rhee, Seongha. . “I know I’m shameless to say this”: Grammaticalization
of the mitigating discourse marker Makilay in Korean. Procedia—Social
and Behavioral Sciences , –.
Rhee, Seongha. . From quoting to reporting to stance-marking: rhetorical
strategies and intersubjectification of reportative. Language Sciences , –.
Rice, Keren. . Morpheme Order and Semantic Scope: Word Formation in
the Athapaskan Verb. Cambridge: Cambridge University Press.
Ringe, Donald A. . Internal reconstruction. In: Joseph, Brian D. and
Richard D. Janda (eds.) The Handbook of Historical Linguistics, –.
Oxford: Blackwell.
Robbeets, Martine. . Genealogically motivated grammaticalization. In:
Robbeets, Martine and Hubert Cuyckens (eds.) Shared Grammatical-
ization: With Special Focus on the Transeurasian Languages, –.
Amsterdam: Benjamins.
Robert, Stéphane (ed.). a. Perspectives synchroniques sur la grammatica-
lisation. Louvain-Paris: Peeters.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Robert, Stéphane. b. Polygrammaticalisation, grammaire fractale et pro-
priétés d’échelle. In: Robert, Stéphane (ed.). Perspectives synchroniques sur
la grammaticalisation, –. Louvain-Paris: Peeters.
Robert, Stéphane. c. Vers une typologie de la transcatégorialité. In:
Robert, Stéphane (ed.). Perspectives synchroniques sur la grammaticalisa-
tion, –. Louvain-Paris: Peeters.
Robert, Stéphane. . The challenge of polygrammaticalization for linguistic
theory: fractal grammar and transcategorial functioning. In: Frajzyngier,
Zygmunt, Adam Hodges, and David S. Rood (eds.) Linguistic Diversity and
Language Theories, –. Amsterdam: Benjamins.
Roberts, Ian. . A formal account of grammaticalisation in the history of
Romance futures. Folia Linguistica Historica , –.
Roberts, Ian. . Diachronic Syntax. Oxford: Oxford University Press.
Roberts, Ian. . Grammaticalization, the clausal hierarchy and semantic
bleaching. In: Traugott, Elizabeth C. and Graeme Trousdale (eds.) Gradience,
Gradualness and Grammaticalization, –. Amsterdam: Benjamins.
Roberts, Ian and Anna Roussou. . A formal approach to grammatical-
ization. Linguistics /, –.
Roberts, Ian and Anna Roussou. . Syntactic Change: A Minimalist
Approach to Grammaticalization. Cambridge: Cambridge University Press.
Romaine, Suzanne. . Pidgin and Creole Languages. London, New York:
Longman.
Romaine, Suzanne. . The grammaticalization of the proximative in Tok
Pisin. Language /, –.
Rosemeyer, Malte. . Refunctionalization and usage frequency: an explora-
tory questionnaire study. Languages , , .
Rosenbach, Annette. . The English s-genitive. A case of degrammaticali-
zation? In: Fischer, Olga, Muriel Norde, and Harry Perridon (eds.) Up and
Down the Cline: The Nature of Grammaticalization, –. Amsterdam:
Benjamins.
Rostila, Jouni. . Storage as a way to grammaticalization. Constructions
/. www.constructions-online.de, urn:nbn:de:--.
Saavedra, David Correia. . Measurements of grammaticalization: Developing
a quantitative index for the study of grammatical change. Ph.D. dissertation,
Université de Neuchâtel and Universiteit Antwerpen.
Saigo, Hideki. . The Japanese Sentence-Final Particles in Talk-in-
Interaction. Amsterdam: Benjamins.
Sankoff, Gillian. . Linguistic outcomes of language contact. In: Chambers,
J.K., P. Trudgill, and Natalie Schilling-Estes (eds.) Handbook of Sociolin-
guistics, –. Oxford: Blackwell.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Sauer, Hans. . Lexicalization and demotivation. In: Booij, Geert et al.
(eds.) Morphologie/Morphology: An International Handbook on Inflection
and Word-Formation, Volume , –. Berlin: Mouton de Gruyter.
Schiffrin, Deborah. . Discourse Markers. Cambridge: Cambridge Univer-
sity Press.
Schiffrin, Deborah. . Discourse markers, meaning, and context. In: Schif-
frin, Deborah, Deborah Tannen, and Heidi E. Hamilton (eds.) The Hand-
book of Discourse Analysis. (Blackwell Handbooks in Linguistics), –.
Oxford, Maldon, MA: Blackwell.
Schladt, Mathias. . The typology and grammaticalization of reflexives. In:
Frajzyngier, Zygmunt and Traci S. Curl (eds.) Reflexives: Forms and Func-
tions, –. Amsterdam: Benjamins.
Schlüter, Julia. . To dare or not to. Is auxiliarization reversible? In: Van
linden, An, Jean-Christophe Verstraete, and Kristin Davidse (eds.) Formal
Evidence in Grammaticalization Research, –. Amsterdam: Benjamins.
Schneider, Stefan, Julie Glikman, and Mathieu Avanzi (eds.). . Parenthet-
ical Verbs. Berlin: De Gruyter.
Schneider, Wolf. . Wörter machen Leute: Magie und Macht der Sprache.
Reinbeck: Rowohlt.
Schourup, Lawrence. . Discourse markers. Lingua , –.
Schulte, Kim and José Luis Blas Arroyo. . Entrenchment and frequency
effects in the diffusion and replacement of modal periphrases in Spanish: a
diachronic variationist analysis. In: Kabatek, Johannes (ed.) Lingüística de
corpus y lingüística histórica iberorrománica, –. Berlin: Mouton de
Gruyter.
Shindo, Mika. . Semantic Extension, Subjectification and Verbalization.
Lanham: University Press of America.
Shinzato, Rumiko. . (Inter)subjectification, Japanese syntax and syntactic
scope increase. Journal of Historical Pragmatics /, –.
Silva-Corvalán, Carmen. . The limits of convergence in language contact.
Paper presented at the symposium “Language Contact and the Dynamics of
Language: Theory and Implications”, Max Planck Institute for Evolution-
ary Anthropology, Leipzig, – May, .
Smith, John Charles. . How to do things without junk: The refunction-
alization of a pronominal subsystem between Latin and Romance. In: Mon-
treuil, J.-P. Y. (ed.) New Perspectives on Romance Linguistics. Volume II:
Phonetics, Phonology and Dialectology, –. Amsterdam: Benjamins.
Smith, John Charles. . Towards a typology of refunctionalization. Pres-
entation at the th International Conference of Historical Linguistics
(ICHL XX), Osaka, Japan. July – .
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Sohn, Sung-Ock S. . On the emergence of intersubjectivity: An analysis of
the sentence-final nikka in Korean. Japanese/Korean Linguistics , –.
Song, Jae Jung. . Linguistic Typology: Morphology and Syntax. Harlow:
Longman.
Song, Jae Jung. . Grammaticalization and structural scope increase:
possessive-classifier-based benefactive marking in Oceanic languages. Lin-
guistics /, –.
Stein, Dieter and Susan Wright (eds.). Subjectivity and Subjectivisation: Lin-
guistic Perspectives. Cambridge: Cambridge University Press.
Stolz, Thomas. a. Forschungen zu den Interrelationen von Grammatika-
lisierung und Metaphorisierung: Von der Grammatikalisierbarkeit des Kör-
pers. I: Vorbereitung (= ProPrinS ). Essen: University of Essen.
Stolz, Thomas. b. Sprachbund im Baltikum? Estnisch und Lettisch im
Zentrum einer sprachlichen Konvergenzlandschaft. Bochum: Brockmeyer.
Stolz, Thomas. . Grammatikalisierung und Metaphorisierung. Bochum:
Universitätsverlag Dr. N. Brockmeyer.
Swan, Toril. . A note on the scope(s) of sadly. Studia Linguistica /,
–.
Swan, Toril. . The development of sentence adverbs in English. Studia
Linguistica /, –.
Sweetser, Eve E. . Grammaticalization and semantic bleaching. Berkeley
Linguistics Society , –.
Sweetser, Eve E. . From Etymology to Pragmatics: Metaphorical and
Cultural Aspects of Semantic Structure. Cambridge: Cambridge University
Press.
Szmrecsanyi, Benedikt. . About text frequencies in historical linguistics:
Disentangling environmental and grammatical change. Corpus Linguistics
and Linguistic Theory /, –.
Tabor, Whitney and Elizabeth C. Traugott. . Structural scope expansion
and grammaticalization. In: Giacalone Ramat, Anna and Paul J. Hopper
(eds.). The Limits of Grammaticalization, –. Amsterdam: Benjamins.
Tantucci, Vittorio. . From immediate to extended intersubjectification: a
gradient approach to intersubjective awareness and semasiological change.
Language and Cognition , –.
Tantucci, Vittorio. . From co-actionality to extended intersubjectivity:
Drawing on language change and ontogenetic development. Applied Lin-
guistics. DOI: ./applin/amy.
Thieroff, Rolf. . On the areal distribution of tense-aspect categories in
Europe. In: Dahl, Östen (ed.) Tense and Aspect in the Languages of Europe,
–. Berlin: Mouton de Gruyter.
Thomason, Sarah Grey. . Language Contact: An Introduction. Edinburgh:
Edinburgh University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Thomason, Sarah Grey and Terrence Kaufman. . Language Contact,
Creolization, and Genetic Linguistics. Berkeley, Los Angeles, London:
University of California Press.
Thompson, Sandra A. and Anthony Mulac. . A quantitative perspective
on the grammaticization of epistemic parentheticals in English. In:
Traugott, Elizabeth C. and Bernd Heine Approaches to Grammatical-
ization, Volume , –. Amsterdam: Benjamins.
Timberlake, Alan. . Reanalysis and actualization in syntactic change. In:
Li, Charles N. (ed.) Mechanisms of Syntactic Change, –. Austin:
University of Texas Press.
Tomasello, Michael. . Origins of Human Communication. Cambridge,
MA: MIT Press.
Torres Cacoullos, Rena. . Grammaticalization through inherent variabil-
ity. The development of a progressive in Spanish. Studies in Language /,
–.
Torres Cacoullos, Rena and James A. Walker. . The present of the English
future: grammatical variation and collocations in discourse. Language ,
–.
Torres Cacoullos, Rena and James A. Walker. . Collocations in grammat-
icalization and variation. In: Narrog, Heiko and Bernd Heine (eds.) The
Oxford Handbook of Grammaticalization, –. Oxford: Oxford Univer-
sity Press.
Trask, Robert L. . A Dictionary of Grammatical Terms in Linguistics.
London: Routledge.
Trask, Robert L. . The typological position of Basque: then and now.
Language Sciences /, –.
Traugott, Elizabeth C. . From polysemy to internal semantic reconstruc-
tion. Berkeley Linguistics Society , –.
Traugott, Elizabeth C. . Pragmatic strengthening and grammaticalization.
Berkeley Linguistics Society , –.
Traugott, Elizabeth C. . On the rise of epistemic meanings in English: an
example of subjectification in semantic change. Language /, –.
Traugott, Elizabeth C. . Syntax. In: Hogg, Richard (ed.) The Cambridge
History of the English Language. Vol. I. The Beginnings to , –.
Cambridge: Cambridge University Press.
Traugott, Elizabeth C. a. The role of the development of discourse
markers in a theory of grammaticalization. Paper presented at the Inter-
national Conference of Historical Linguistics XII, Manchester.
Traugott, Elizabeth C. b. Subjectification in grammaticalisation. In: Stein,
Dieter and Susan Wright (eds.) Subjectivity and Subjectivisation: Linguistic
Perspectives, –. Cambridge: Cambridge University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Traugott, Elizabeth C. . The rhetoric of counter-expectation in semantic
change: a study in subjectification. In: Blank, Andreas and Peter Koch (eds.)
Historical Semantics and Cognition, –. Berlin: De Gruyter.
Traugott, Elizabeth C. . From etymology to historical pragmatics. In:
Minkova, Donka and Robert Stockwell (eds.) Studies in the History of the
English Language, –. Berlin: Mouton de Gruyter.
Traugott, Elizabeth C. a. Constructions in grammaticalization. In: Joseph,
Brian D. and Richard D. Janda (eds.) The Handbook of Historical Linguis-
tics, –. Oxford: Blackwell.
Traugott, Elizabeth C. b. From subjectification to intersubjectification. In:
Hickey, Raymond (ed.) Motives for Language Change, –. Cambridge:
Cambridge University Press.
Traugott, Elizabeth.C. . Exaptation and grammaticalization. In: Akimoto,
Minoru (ed.) Linguistic Studies based on Corpora, –. Tokyo: Hituzi Syobo.
Traugott, Elizabeth C. . Discussion article: Discourse markers, modal
particles, and contrastive analysis, synchronic and diachronic. Catalan
Journal of Linguistics , –.
Traugott, Elizabeth C. a. Grammaticalization. In: Luraghi, Silvia and Vit
Bubenik (eds.) Continuum Companion to Historical Linguistics, –.
London: Continuum.
Traugott, Elizabeth C. b. Dialogic contexts as motivations for syntactic
change. In: Cloutier, Robert A., Anne Marie Hamilton-Brehm, and William
A. Kretzschmar (eds.) Studies in the History of the English Language V:
Variation and Change in English Grammar and Lexicon, –. Berlin:
Mouton de Gruyter.
Traugott, Elizabeth C. c. (Inter)subjectivity and (inter)subjectification:
A reassessment. In: Davidse, Kristin, Lieven Vandelanotte, and Hubert
Cuyckens (eds.) Subjectification, Intersubjectification and Grammatical-
ization, –. Berlin: Mouton de Gruyter.
Traugott, Elizabeth C. d. Grammaticalization. In Andreas H. Jucker and
Irma Taavitsainen (eds.) Historical Pragmatics, –. Berlin: De Gruyter
Mouton.
Traugott, Elizabeth C. . Grammatical constructionalization: rethinking
grammaticalization in the light of constructionalization. Paper presented at
the University of Santiago de Compostela, October , .
Traugott, Elizabeth Closs. . Toward a constructional framework for
research on language change. In: Hancil, Silvia and Ekkehard König
(eds.) Grammaticalization—Theory and Data, –. Amsterdam:
Benjamins.
Traugott, Elizabeth C. a. Rethinking the Role of Invited Inferencing in
Change from the Perspective of Interactional Texts. Open Linguistics , –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Traugott, Elizabeth C. b. Modeling language change with constructional
networks. In: Bordería, Salvador Pons and Óscar Loureda Lamas (eds.)
Beyond Grammaticalization and Discourse Markers: New Issues in the
Study of Language Change, –. Leiden, Boston: Brill.
Traugott, Elizabeth C. and Richard B. Dasher. . Regularity in Semantic
Change. Cambridge: Cambridge University Press.
Traugott, Elizabeth C. and Ekkehard König. . The semantics-pragmatics
of grammaticalization revisited. In: Traugott, Elizabeth C. and Bernd Heine
(eds.) Approaches to Grammaticalization, Volume , –. Amsterdam:
Benjamins.
Traugott, Elizabeth C. and Graeme Trousdale. . Gradience, gradualness
and grammaticalization. How do they intersect? In: Traugott, Elizabeth
C. and Graeme Trousdale (eds.) Gradience, Gradualness and Grammat-
icalization, –. Amsterdam: Benjamins.
Traugott, Elizabeth C. and Graeme Trousdale. . Constructionalization
and Constructional Changes. Oxford: Oxford University Press.
Traugott, Elizabeth C. and Graeme Trousdale. . Contentful constructio-
nalization. Journal of Historical Linguistics /, –.
Trips, Carola. . Lexical Semantics and Diachronic Morphology: The Devel-
opment of -hood, -dom and -ship in the History of English. Tübingen: Max
Niemeyer Verlag.
Trousdale, Graeme. . Constructions in grammaticalization and lexicaliza-
tion: Evidence from the history of a composite predicate in English. In:
Trousdale, Graeme and Nikolas Gisborne (eds.) Constructional Approaches
to English Grammar, –. Berlin: Mouton de Gruyter.
Trousdale, Graeme. a. Grammaticalization, constructions and the gram-
maticalization of constructions. In: Davidse, Kristin et al. (eds.) Grammat-
icalization and Language Change: New Reflections, –. Amsterdam:
Benjamins.
Trousdale, Graeme. b. Grammaticalization, lexicalization and construc-
tionalization from a cognitive-pragmatic perspective. In: Schmid,
Hans-Jörg (ed.) Cognitive Pragmatics, –. Berlin: Mouton de Gruyter.
Trousdale, Graeme. . Multiple inheritance and constructional change.
Studies in Language /, –.
Trousdale, Graeme. . On the relationship between grammaticalization
and constructionalization. Folia Linguistica /, –.
Trousdale Graeme and Muriel Norde. . Degrammaticalization and con-
structionalization: two case studies. Language Sciences , –.
Van Bogaert, Julie . I think and other complement-taking mental predi-
cates: A case of and for constructional grammaticalization. Linguistics /,
–.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Van de Velde, Freek and Muriel Norde. . Exaptation. Taking stock of a
controversial notion in linguistics. In: Norde, Muriel and Freek van de Velde
(eds.) Exaptation and Language Change, –. Amsterdam: Benjamins.
Van der Auwera, Johan and Jan Nuyts (eds.). . Grammaticalization and
(Inter-)subjectification. Brussel: KVAB.
van Gelderen, Elly. . The Rise of Functional Categories. Amsterdam:
Benjamins.
van Gelderen, Elly. . Grammaticalization as Economy. Amsterdam:
Benjamins.
van Gelderen, Elly (ed.). . Cyclical Change. Amsterdam: Benjamins.
van Gelderen, Elly. . Features in reanalysis and grammaticalization. In:
Traugott, Elizabeth C. and Graeme Trousdale (eds.) Gradience, Gradual-
ness and Grammaticalization, –. Amsterdam: Benjamins.
van Gelderen, Elly. a. The Linguistic Cycle. Oxford: Oxford University
Press.
van Gelderen, Elly. b. The grammaticalization of agreement. In: Heiko
Narrog and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
van Gelderen, Elly. c. Grammaticalization and generative grammar:
A difficult liaison. In: Narrog, Heiko and Bernd Heine (eds.) The Oxford
Handbook of Grammaticalization, –. Oxford: Oxford University Press.
van Gelderen, Elly. . Clause Structure. Cambridge: Cambridge Unversity
Press.
van Gelderen, Elly (ed.). . Cyclical Change Continued. Amsterdam:
Benjamins.
van Olmen, Daniel, Hubert Cuyckens, and Lobke Ghesquière (eds.). .
Aspects of Grammaticalization: (Inter)Subjectification and Directionality.
Berlin: Mouton de Gruyter.
Van Rompaey, Tinne. . Discourse conditions in language change: from
binominal to aspectualizer be in the middle or midst of V-ing. Language
Sciences , –.
Velupillai, Viveka. . An Introduction to Linguistic Typology. Amsterdam:
Benjamins.
Vennemann, Theo. . Language change as language improvement. In:
Jones, Charles (ed.) Historical Linguistics: Problems and Perspectives,
–. London: Longman.
Verhagen, Arie. . Constructions of Intersubjectivity. Discourse, Syntax and
Cognition. Oxford: Oxford University Press.
Verhagen, Arie. . Construal and perspectivization. In: Geeraerts, Dirk and
Hubert Cuyckens (eds.) The Oxford Handbook of Cognitive Linguistics,
–. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Verstraete, Jean-Christophe. . Subjective and objective modality:
Interpersonal and ideational functions in the English modal auxiliary
system. Journal of Pragmatics , –.
Vincent, Nigel. . Iconic and symbolic aspects of syntax: Prospects for
reconstruction. In: Ramat, Paolo (ed.) Linguistic Reconstruction and Indo-
European Syntax, –. Amsterdam: Benjamins.
Vindenes, Urd. . Cyclic renewal of demonstratives. Studies in Language
/, –.
Visconti, Jacqueline. . Conditionals and subjectification: implications for
a theory of semantic change. In: Fischer, Olga, Muriel Norde, and Harry
Perridon (eds.) Up and Down the Cline: The Nature of Grammaticalization,
–. Amsterdam: Benjamins.
Visconti, Jacqueline. . Facets of subjectification. Language Sciences , –.
Visser, Frederikus T. . An Historical Syntax of the English Language, Part
Three, First Half: Syntactical Units with Two Verbs. Leiden: Brill.
Viti, Carlotta. . On degrammaticalization: Controversial points and pos-
sible explanations. Folia Linguistica /, –.
Von Mengden, Ferdinand. . Functional changes and (meta-)linguistic
evolution. In: Norde, Muriel and Freek van de Velde (eds.) Exaptation
and Language Change, –. Amsterdam: Benjamins.
Waltereit, Richard. . Grammaticalization and discourse. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Waltereit, Richard. . On the origins of grammaticalization and other types
of language change in discourse strategies. In: Davidse, Kristin, Tine Breban,
Lieselotte Brems, and Tanja Mortelmans (eds.) Grammaticalization and
Language Change: New Reflections, –. Amsterdam: Benjamins.
Warner, Anthony. . Predicting the Progressive Passive: Parametric
Change within a Lexicalist Framework. Language , –.
Weinreich, Uriel. [] . Languages in Contact. London, The Hague,
Paris: Mouton.
Werner, Heinz and Bernard Kaplan. . Symbol-formation: An Organismic
Developmental Approach to Language and the Expression of Thought. New
York, London, Sidney: Wiley.
Whaley, Lindsay J. . Introduction to Typology: The Unity and Diversity of
Language. London: Sage.
Whitehead, A. N. . Process and Reality (corrected edition). New York:
The Free Press [original British edition : Cambridge: Cambridge
University Press].
Wichmann, Anne. . Grammaticalization and prosody. In: Narrog, Heiko
and Bernd Heine (eds.) The Oxford Handbook of Grammaticalization,
–. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Wichmann, Anne, Anne-Marie Simon-Vandenbergen, and Karin Aijmer.
. Of course: How prosody reflects semantic change: a synchronic case
study of of course. In: Cuyckens, Hubert, Kristin Davidse, and Lieven
Vandelanotte (eds.) Subjectification, Intersubjectification and Grammat-
icalization, –. Berlin: De Gruyter Mouton.
Wiemer, Björn. . The grammaticalization of passives. In: Narrog, Heiko
and Bernd Heine (eds.) Handbook of Grammaticalization, –. Oxford:
Oxford University Press.
Wiemer, Björn. . Quo vadis grammaticalization theory? Why complex
language change is like words. Folia Linguistica /, –.
Willis, David. . Syntactic lexicalization as a new type of degrammaticali-
zation. Linguistics /, –.
Willis, David. . Degrammaticalization and obsolescent morphology. Evi-
dence from Slavonic. In: Stathi, Katerina, Elke Gehweiler, and Ekkehard
König (eds.) Grammaticalization: Current Views and Issues, –.
Amsterdam: Benjamins.
Willis, David. . Exaptation and degrammaticalization within an
acquisition-based model of abductive reanalysis. In: Norde, Muriel and
Freek van de Velde (eds.) Exaptation and Language Change, –.
Amsterdam: Benjamins.
Wischer, Ilse. . Lexikalisierung versus Grammatikalislerung. Gemeinsam-
keiten und Unterschiede. Papiere zur Linguistik /, –.
Wischer, Ilse. . Grammaticalization versus lexicalization: ‘Methinks’ there
is some confusion. In: Fischer, Olga, Anette Rosenbach, and Dieter Stein
(eds.) Pathways of Change: Grammaticalization in English, –.
Amsterdam: Benjamins.
Wischer, Ilse. . Sekretion und Exaptation als Mechanismen in der
Wortbildung und Grammatik. In: Harnisch, Rüdiger (ed.) Prozesse spra-
chlicher Verstärkung. Typen formaler Resegmentierung und semantischer
Remotivierung, –. Berlin: De Gruyter.
Wischer, Ilse. . Grammaticalization and word formation. In: Narrog,
Heiko and Bernd Heine (eds.) The Oxford Handbook of Grammatical-
ization, –. Oxford: Oxford University Press.
Wolff, Ekkehard. . A Grammar of the Lamang Language (Gwàd LàmàN).
Glückstadt: J. J. Augustin.
Wright, Susan. . Subjectivity and experiential syntax. In: Stein, Dieter and
Susan Wright (eds.) Subjectivity and Subjectivisation, –. Cambridge:
Cambridge University Press.
Wu, Xiu-Zhi Zoe. . Grammaticalization and Language Change in Chin-
ese: A Formal View. New York: Routledge.
Xing, Janet Zhiqun. . Mechanisms of semantic change in Chinese. Studies
in Language /, –.
OUP CORRECTED PROOF – FINAL, 24/1/2021, SPi
REFERENCES
Xing, Janet Zhiqun. . Semantic change in Chinese. In: Chan, Sin-wai (ed.) The
Routledge Encyclopedia of the Chinese Language, –. London: Routledge.
Xing, Janet Zhiqun (ed.). . Introduction. In: Xing, Janet Zhiqun (ed.) A
Typological Approach to Grammaticalization and Lexicalization: East Meets
West, –. Berlin: De Gruyter.
Young, Robert. . The Navajo Verb System: An Overview. Albuquerque:
University of New Mexico Press.
Young, Robert and William Morgan. . The Navajo Language: A Grammar
and Colloquial Dictionary. nd edition. Albuquerque: University of New
Mexico.
Yuzawa, Kōkichirō. . Zōtei Edo Kotoba no Kenkyū [Research on the Edo
language] (expanded and revised version). Tōkyō: Meiji Shoin.
Zariquiey, Roberto. . Diachronic stories of body-part nouns in some
language families of South America. In: Narrog, Heiko and Bernd Heine
(eds.) Grammaticalization from a Typological Perspective. –. Oxford:
Oxford University Press.
Zavala, Roberto. . Calcos sintácticos en algunos complejos verbales Mayas
y Mixe-Zoques. Pueblos y Fronteras : –.
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
Author index
Aaron, Jessi Elena Blas Arroyo, José Luis
Abraham, Werner , , , , Boland, Johanna H.G.
, Boogaart, Ronny
Adamson, Sylvia , Booij, Gert –
Adone, Dany Boretzky, Norbert –,
Ahn, Mikyung – Börjars, Kersti , , , –, –
Aijmer, Karin , Boye, Kasper , , , , ,
Aikhenvald, Alexandra Y. , , , Bradley, Peter T. ,
–, , Breban, Tine , , ,
Ambrazas, Vytautas Brems, Lieselotte
Andersen, Henning , , – Brinton, Laurel J. –, –, –,
Anderson, Lloyd B. , , , , , , ,
Ansaldo, Umberto , , –, ,
, Bruyn, Adrienne , , ,
Arcodia, Giorgio Francesco Budts, Sara
Arends, Jacques Burridge, Kate –
Ariel, Mira , , –, Bybee, Joan L. , , , , , –, ,
Arkadiev, Peter , , , , , –, –, , ,
Arnovick, Leslie , , –, , –, ,
Arroyo, José Luis Blas , –, ,
Ashby, William J. ,
Askedal, John Ole Campbell, Lyle , , –, , –, ,
Athanasiadou, Angeliki , –, –, , , ,
Auer, Peter Carlier, Anne ,
Awolaye, Yiwola Cennamo, Michela ,
Chafe, Wallace ,
Baker, Mark Chomsky, Noam A.
Baker, Philip Chor, Winnie
Barðdal, Jóhanna Claridge, Claudia
Barth-Weingarten, Dagmar , Claudi, Ulrike , , –
Beijering, Karin , , , , Coates, Jennifer ,
, Collins, Peter ,
Bergqvist, Henrik Comrie, Bernard –, , , ,
Bergs, Alexander ,
Bhat, D.N.S. , – Cornillie, Bert ,
Biber, Douglas , Couper-Kuhlen, Elizabeth –,
Bisang, Walter , , , , –, , ,
, Craig, Colette G. , ,
Blakemore, Diane Croft, William ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
AUTHOR INDEX
Csató, Éva Fortuin, Egbert
Cysouw, Michael – Frajzyngier, Zygmunt
François, Alexandre
Dahl, Östen , , , , –, Frank-Job, Barbara –, –,
, Fraser, Bruce –,
Dao, Huy Linh , Fried, Miriam ,
Dávalos, Jimena Tena –
Davidse, Kristin –, , Gabelentz, Georg von der , ,
De Mulder, Walter , ,
De Smet, Hendrik , , , , Gaeta, Livio –,
, García García, Luisa
De Wolf, Simon Gast, Volker
Degand, Liesbeth , –, , Geis, Michael L.
Dehé, Nicole , Georgakopoulos, Thanasis
DeLancey, Scott Ghesquière, Lobke
Delbeque, Nicole Ghezzi, Chiara –
Denison, David Giacalone Ramat, Anna , –,
Dér, Csilla Ilona ,
Detges, Ulrich , , , Gisborne, Nikolas –,
Devos, Maud Givón, T. , –, –, , , , ,
Diewald, Gabriele , –, –, , ,
, , Goossens, Louis
Dik, Simon C. Gould, Stephen Jay
Dixon, Robert M. W. Gray, Bethany
Do-Hurinville, Thành , Greenberg, Joseph H. , , ,
Dong, Xiufang –,
Dostie, Gaetane. , , Grice, H. Paul ,
Dryer, Matthew S. Grimm, Jacob
Du Feu, Veronica Grimm, Wilhelm
Günthner, Susanne
Eckardt, Regine Guthrie, Malcolm
Enfield, N. J.
Erman, Britt , Haase, Martin –, ,
Espinal, M. Teresa Habermann, Mechthild ,
Esseesy, Mohssen , Haegeman, Liliane
Evans, Nicholas Haig, Geoffrey , , –
Evers-Vermeul, Jacqueline , , Haiman, John , ,
, Hale, Ken
Halliday, M. A. K.
Fagard, Benjamin Hancil, Sylvie ,
Fischer, Kerstin Hansen, Maj-Britt Mosegaard ,
Fischer, Olga , , –, , , –, , , , ,
, , Harnisch, Rüdiger ,
Fitzmaurice, Susan , , Harder, Peter , , , , ,
Fortescue, Michael Hardie, Andrew
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
AUTHOR INDEX
Harris, Alice C. , –, , –, , Ichihashi-Nakayama, Kumiko
–, –, Igla, Birgit –
Harrison, Andrew – Itkonen, Esa
Haselow, Alexander , , Ito, Shinya
Haspelmath, Martin –, ,
–, , –, , , , Jäger, Gerhard
–, , –, , , , Jakobson, Roman
, Jamison, Stephanie
Hawkins, John , , Janda, Laura A.
Heath, Jeffrey Janda, Richard D. , , , ,
Heine, Bernd , –, , , –, , Jendraschek, Gerd –, –
–, , , –, –, , –, Jenkins, Rebecca Sue –
–, , , –, , –, , , Jespersen, Otto –,
–, , , –, , , Johanson, Lars , , ,
–, , –, , –, , Johnson-Laird, Philip N.
–, –, , , , , Jones, Mari C.
–, , –, , –, Joseph, Brian D. , , , ,
–, , , –, , , Jucker, Andreas H. –,
, , , , –, , , Juge, Matthew L.
–, –, , –
Helbig, Gerhard Kaltenböck, Gunther , , , ,
Heltoft, Lars –,
Hengeveld, Kees , – Kailuweit, Rolf
Herring, Susan C. , –, Kaplan, Bernard ,
Higashiizumi, Yuko Kastovsky, Dieter
Hilpert, Martin , , –, , Kaufman, Terrence , , ,
–, Kavalova, Yordanka ,
Himmelmann, Nikolaus P. , , Kearns, Kate
–, , , , Keesing, Roger M. ,
Hock, Hans Henrich Keizer, Evelien
Hodge, Carleton T. , Keller, Rudi , –, , ,
Hoefler, Stefan H. Kemmer, Suzanne
Hoffmann, Sebastian Kinuhata, Tomohide , –
Holm, John A. Kiparsky, Paul , , –,
Hopper, Paul J. , , , –, , , Kissine, Mikhail
–, , , , , , , –, , Klamer, Marian
, –, –, , , –, Klausenburger, Jürgen
–, –, –, –, , , König, Ekkehard –, –, , ,
–, –, ,
Hornstein, Norbert Koops, Christian –
Huang, Yan Kortmann, Bernd –,
Huddleston, Rodney , Kotsinas, Ulla-Britt ,
Hüning, Matthias – Koyanagi, Tomokazu
Hurch, Bernhard –, , , Kranich, Svenja , , –, , ,
Hwang, Jya-Lin –,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
AUTHOR INDEX
Krieger, Manuela. Matisoff, James A. ,
Krug, Manfred G. , , Matras, Yaron
Kuryɬowicz, Jerzy Matsumoto, Yo
Kuteva, Tania , , , , , , , Mauri, Caterina
–, –, , –, –, , , , McEnery, Tony
–, , –, –, , –, McNeill, David
, , , , , –, –, McWhorter, John
, , , –, –, –, Meillet, Antoine , –, ,
, , Miller, George A.
Minami, Fujio
Labov, William Mirčev, Kiril
Lacey, Alan Robert Mitchell, Bruce
Lai, Vicky Tzuyin Mithun, Marianne –, ,
Lamiroy, Béatrice , , , , –,
Langacker, Ronald W. –, , , Miyashita, Hiroyuki –
, , – Molinelli, Piera –
Lass, Roger , –, Moravcsik, Edith A.
Ledgeway, Adam – Morfill, W. R.
Lehmann, Christian , –, , –, Morgan, William
–, , –, –, , –, Moser, Rosemarie –
, –, , , , –, , Moyse-Faurie, Claire ,
, , Mufwene, Salikoko S.
Lenker, Ursula Mühlhäusler, Peter
Lepore, Ernie Mulac, Anthony
Lessau, Donald Andreas Mushin, Ilana
Levinson, Stephen C. Mutz, Katrin
Lichtenberk, Frantisek –
Lightfoot, David , Nakayama, Toshihide
Lightfoot, Douglas J. , , Narrog, Heiko , , , , , , ,
Lim, Lisa , , , , , , –, , –, –,
Lima, José Pinto de , –, , –, , , ,
López-Couso, María José , , , , , , –, , ,
Lord, Carol Diane , , , ,
Luraghi, Silvia , , , Nevalainen, Terttu
Lyons, John , , – Newmeyer, Frederick J. , ,
,
Maat, Henk Pander Nichols, Johanna
Mackenzie, Ian , Nicolle, Steve , ,
Mackenzie, J. Lachlan Noël, Dirk , –,
Mair, Christian , , , , Nomachi, Motoki , ,
Maisak, Timur , Norde, Muriel , , , , , , ,
Malvar, Elisabete –, –, , , , –,
Mardale, Alexandru , ,
Maslova, Elena Nørgård-Sørensen, Jens –
Matasović, Rolf Nuyts, Jan –, ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
AUTHOR INDEX
Ocampo, Francisco , , Rosenbach, Annette , , ,
Ohori, Toshio Rostila, Jouni
Onodera, Noriko , Roussou, Anna ,
Osawa, Fuyo
Saigo, Hideki
Pagliuca, William , Sankoff, Gillian
Palander-Collin, Minna Sansò, Andrea –
Park, Insun , Sauer, Hans
Patten, Amanda –, Schiffrin, Deborah –
Paul, Herrmann Schladt, Mathias
Pedersen, Eric Schlüter, Julia
Peng, Rui Schneider, Stefan
Persson, Gunnar , Schneider, Wolf
Peters, Hans Schourup, Lawrence
Petré, Peter Schulte, Kim
Plag, Ingo Shindo, Mika
Polis, Stephane Shinzato, Rumiko
Poplack, Shana – Silva-Corvalán, Carmen
Portner, Paul Simon-Vandenbergen,
Prévost, Sophie Anne-Marie ,
Prince, Ellen F. Smirnova, Elena
Proudfoot, Michael Smith, John Charles , –,
Pullum, Geoffrey K. , Sohn, Sung-Ock S.
Pulvermüller, Friedemann Song, Jae Jung , ,
Pustet, Regina Stein, Dieter ,
Stolz, Thomas , , –,
Quirk, Randolph , Stone, Matthew
Suzuki, Ryoko
Ramat, Paolo , , –, , , Swan, Toril ,
, Sweetser, Eve E. , –,
Ramisch, Heinrich Syea, Anand
Rayfield, Joan Rachel – Szmrecsanyi, Benedikt
Reh, Mechthild , , , , ,
, –, , , Tabor, Whitney ,
Reinöhl, Uta –, Tantucci, Vittorio
Rhee, Seongha , Thieroff, Rolf
Rice, Keren Thomason, Sarah Grey , ,
Ringe, Donald A. ,
Robbeets, Martine Thompson, Sandra A. , ,
Robert, Stéphane , , , ,
Roberts, Ian , , , , , , , Timberlake, Alan
, , –, Tomasello, Michael
Romaine, Suzanne –, , Torres Cacoullos, Rena ,
Rosemeyer, Malte ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
AUTHOR INDEX
Trask, Robert L. , , , Von Mengden, Ferdinand
Traugott, Elizabeth C. , , , , , , Vrba, Elisabeth S.
, , –, , –, , , , –, ,
, –, , –, , , –, , Walker, James A.
–, –, –, , –, Waltereit, Richard , , –
–, –, –, –, , , Warner, Anthony
, , , –, –, , , Weinreich, Uriel
, , –, –, , , Werner, Heinz ,
Trips, Carola Whaley, Lindsay J.
Trousdale Graeme , , , , Whitehead, A. N.
, , , , Wichmann, Anne
Wiemer, Björn , ,
Van Bogaert, Julie , , Wilkins, David
Van der Auwera, Johan , , Willis, David , –
Van de Velde, Freek , Wischer, Ilse , –, , , ,
Van der Wal, Jenneke ,
Van Gelderen, Elly , , –, , Wolff, Ekkehard
, –, Wright, Susan ,
Van Olmen, Daniel Wu, Xiu-Zhi Zoe –
Van Rompaey, Tinne
Velupillai, Viveka Xing, Janet Zhiqun ,
Vennemann, Theo
Verhagen, Arie , Yang, Lynne
Verstraete, Jean-Christophe Young, Robert
Vincent, Nigel –, , –, Yuzawa, Kōkichirō
–
Vindenes, Urd –, Zariquiey, Roberto
Visconti, Jacqueline , , Zavala, Roberto
Visser, Frederikus Ziv, Yael
Viti, Carlotta Zwicky, Arnold
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
Language index
Afrikaans – French , , , , , , , ,
Albanian , , , , , –, , –, ,
, –, , –, ,
Balkan languages –, French, Old , –, ,
Bantu languages –, , , ,
Bari German , , , , –, , ,
Basque –, – –, , , –, , , ,
Belarusian –, –, , , , –,
Bulgarian –, –, , , –, –, –
–, Germanic languages , , , ,
Greek , ,
Chamus –
Cherokee , , Hebrew ,
Chinese (Mandarin) , , , , Hungarian
–, –
Croatian Indo-European languages –, , ,
, , ,
Dutch , , , Iroquoian languages , ,
Italian –, , –,
English –, , , , , –, –,
–, –, , , –, , –, Japanese –, –, , , ,
–, , –, , , , , , , , –, –,
, , , , –, , Japanese, Old , –
–, , , –, –, ,
, –, , –, , –, Kabba –
, , –, , –, Kenya Pidgin Swahili , –
–, , , , , –, Korean , ,
, –, –, –, –,
–, , , , , –, , Ladin
, Lahu
English, Old –, , –, , , , Lamang
, , –, –, , , , Latin –, , , –, , ,
, – , , –, –, ,
Estonian , – ,
Ewe , – Latvian –
Finnic languages – Maa –
Finnish – Maasai, see Maa
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
LANGUAGE INDEX
Maltese Sanskrit –
Mandarin, see Chinese Serbian ,
Mayan Shoshone
Mexican Spanish, see Spanish Slavic languages –,
Mixe-Zoquean languages Spanish –, –, , , , –,
, –,
Old Church Slavonic Sranan , –
Swahili , –, –, , –, –,
Papago , , –,
Pennsylvania Dutch, see Pennsylvania
German Tamil , ,
Pennsylvania German – Tariana , –,
Pidgin Swahili, see Swahili Thai ,
Polish , Tigak –
Pomattertitsch – Tolowa
Portuguese –, , , , Tümpisa, see Shoshone
Proto-Indo-European , Turkic languages ,
Rapanui Ukrainian
Rhaeto-Romance –
Romance languages , –, , , Vedic Sanskrit, see Sanskrit
, , , , –, –,
, !Xun –
Romani –
Romanian , , Yiddish –, –
Russian , , , , , Yoruba ,
, Yupʼik ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
Subject index
ability, see PI-possibility cline (of grammaticalization) , ,
ablative, also: source –, , , , –, , , , ,
, , clitic , , –, , –, , , ,
agreement , , , , , , , , , , ,
analogical leveling , , ,
analogy , , , , –, , coalescence , , , ,
–, , , –, , comitative –, –, ,
andative –, , comparative (form, construction) ,
anterior, see also perfect , , , , , ,
, comparative method , –
antigrammaticalization , –, compositionality , –,
applicative , , , concessive , , , –, , ,
automation , , –, , ,
autonomy –, , , , , , concessive conditional –, ,
, –, , conditional , , , ,
avertive , conjunction , , , –, –,
, , , , –, , ,
benefactive –, , ,
bidirectionality , , constructional change , ,
bleaching, see also erosion , , , constructionalization , –,
–, , , , , , , ,
, contact-induced
bondedness , , –, , grammaticalization –,
borrowing –, , , –, , context extension , –, ––, –,
, , , , , , , –, , , –, ,
boundness –, , –, , , , ,
bridging context –, –, , conventionalization , –, –, ,
– , –, , , ,
cooptation , , –,
case –, , , , , –, , cycle , –, , , , –,
–, , , , –, , ,
causative –, , , –,
, dative , , –,
chain (of grammaticalization) , , , de-volitive (future) , ,
, , –, , , , , debonding , ,
, decategorialization , , –, –,
change-of-state , , , –, , –, , , , ,
chunking , , , , , , , , ,
classifier , –, –, , ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUBJECT INDEX
degrammaticalization –, , , grammaticalization, primary , , , ,
–, , , –, –, , ,
degrammation , , grammaticalization, secondary , , ,
deinflectionalization –, , , , , , , , ,
delexicalization –, ,
demonstrative , –, , , , , –,
, –, , , , –, , habitual (aspect) ,
demorphologization –, hearer orientation , –, ,
deobligatorification –, , –, –
deontic (modality) –, , , heterosemy –, ,
–, honorific , , ,
desemanticization , , –, , ,
–, , , –, , , , , imperative , , , –, –, ,
, , , , , –,
discourse marker , , –, , imperfective , , –, , –,
, –, –, –, , , –,
discourse orientation , –, , impersonal , , ,
–, implicature, conversational , ,
discursively primary , , , , ,
discursively secondary –, , , , inclusive ,
distal (demonstrative) , , indefinite article , –, ,
divergence , –, –, , , –, ,
dual , –,
durative , , indefinite pronoun ,
inference , , , , , –,
elative , –, –, ,
emphatic , , ingressive ,
endocentric (construction) , , innovation , , –, , , ,
epistemic (modality) , , , , , –, –, ,
–, , , , –, , instrument(al) –, –, , –,
ergative –, , , ,
erosion , –, , –, , –, , integrity , , , , ,
, , –, , , , , intensifier , ,
evidential , –, intention, see B–necessity
evidential, inferred , interjection , ,
evidential, reported , intersubjectification , , , ,
exaptation –, –, –, –, –, ,
exclusive , intersubjectivity –, , ,
,
focus , , iterative , ,
functional renewal ,
future (tense) –, , –, –, , layering , , ,
–, –, –, , , , , lexicalization –, –, , ,
–, , , –, –, –, , –,
–, loss-and-gain model ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUBJECT INDEX
mechanism , , , –, –, proclitic , , ,
, , , –, –, progressive –, , –, ,
metatextual , , –, , , –, –, , , , ,
, –, , , ,
model language , , –, proximative –, ,
,
morphological cycle, see cycle reanalysis , , , , –, –,
morphologization , –, , , , , ,
recipient , –,
B(oulomaic)-necessity (intention) –, reflexive , , –, ,
, , , regrammation
D(eontic)-necessity (obligation) , , reinforcement –, , ,
–, , renewal –, , , ,
E(pistemic)-necessity (probability) –, repetitive, see iterative
, replica language , , , –,
nominalization , –, ,
nominalizer , , , replication –, , , –, ,
non-specific marker , , , –, ,
resultative , , ,
obligation, see D-necessity retraction , ,
obligatorification , , , ritualization , ,
obligatoriness , –, –, , , root possibility, see C-possibility
, routinization , –, ,
paradigmaticity , –, , scope , –, , , –, , ,
passive , , –, –, , , , , , , ,
–, scope, structural –, –, ,
perfect , , , , , –, , ,
, serial verb , , , ,
perfective , , –, –, source, see ablative
, source, lexical (of grammaticalization) –,
persistence , , , , , , , , , –,
phonogenesis –, , source construction , ,
phonologization –, source meaning –, –, , ,
polysemy –, , –, –, ,
–, , –, source schema –, , –,
polysemy copying –, speaker orientation –, –,
C(ircumstantial)-possibility (root –, , ,
possibility) –, specialization , , ,
PI(participant-internal)-possibility specific marker ,
(ability) , , speech act orientation , ,
presentative marker , , , speech community , , –,
procedural (meaning) , , subjectification , –, , ,
–, –, –, , ,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
SUBJECT INDEX
subjectivity –, , , , , transcategoriality –,
, , transcategorization , ,
subordinator , –, , ,
substitution –, uniformitarian principle, or
switch context , –, , uniformitarianism ,
univerbation , , –,
target (of grammaticalization) –, ,
–, –, , , , , , variability, paradigmatic –, –,
, , , , ,
thetical , –, variability, syntagmatic , , , –,
topic , , , , –,
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
OUP CORRECTED PROOF – FINAL, 23/1/2021, SPi
You might also like
- Theoretical and Empirical Issues in Grammaticalization100% (1)Theoretical and Empirical Issues in Grammaticalization382 pages
- Beaugrande R., Discourse Analysis and Literary Theory 1993No ratings yetBeaugrande R., Discourse Analysis and Literary Theory 199324 pages
- A Computational Phonology of Russian - Peter A. Chew PDF100% (1)A Computational Phonology of Russian - Peter A. Chew PDF426 pages
- What Makes Learning Second Language Grammar Difficult A Review of Issues DeKeyser 2005No ratings yetWhat Makes Learning Second Language Grammar Difficult A Review of Issues DeKeyser 200525 pages
- Continuing Discourse On Language - A Functional Perspective Volume 2100% (1)Continuing Discourse On Language - A Functional Perspective Volume 250 pages
- (Linguistic Inquiry Monographs) Alec Marantz-On The Nature of Grammatical Relations (Linguistic Inquiry Monographs, 10) - The MIT Press (1984)No ratings yet(Linguistic Inquiry Monographs) Alec Marantz-On The Nature of Grammatical Relations (Linguistic Inquiry Monographs, 10) - The MIT Press (1984)355 pages
- Everything That Linguists Have Always Wanted To Know About - McCawley, James D - 2019 - University of Chicago Press - 0226556182 - Anna's Archive100% (1)Everything That Linguists Have Always Wanted To Know About - McCawley, James D - 2019 - University of Chicago Press - 0226556182 - Anna's Archive532 pages
- Systemic Grammar (Eng115) Powerpoint PresentationNo ratings yetSystemic Grammar (Eng115) Powerpoint Presentation10 pages
- Syntactic Theory and The Evolution of Syntax: Draft ManuscriptNo ratings yetSyntactic Theory and The Evolution of Syntax: Draft Manuscript36 pages
- The Limits of Structuralism - Forgotten Sources in The History of Modern Linguistics100% (1)The Limits of Structuralism - Forgotten Sources in The History of Modern Linguistics336 pages
- (HANDBOOKS in LANGUAGE and LINGUISTICS) Dikken M.D. (Ed.) - The Cambridge Handbook of Generative Syntax-Cambridge University Press (2013)100% (4)(HANDBOOKS in LANGUAGE and LINGUISTICS) Dikken M.D. (Ed.) - The Cambridge Handbook of Generative Syntax-Cambridge University Press (2013)1,166 pages
- Tense and Aspect in Han Period Chinese PDF50% (2)Tense and Aspect in Han Period Chinese PDF538 pages
- Introduction To English Linguistics 2nd Edition Ingo Plag Download100% (1)Introduction To English Linguistics 2nd Edition Ingo Plag Download54 pages
- Studies On Semantics in Generative Grammar Noam Chomsky100% (2)Studies On Semantics in Generative Grammar Noam Chomsky103 pages
- Hopper 6 Traugott - 2003 Grammaticalization Cap.7No ratings yetHopper 6 Traugott - 2003 Grammaticalization Cap.712 pages
- The History of Linguistics in The Classical Period (Daniel J. Taylor) (Z-Library)100% (1)The History of Linguistics in The Classical Period (Daniel J. Taylor) (Z-Library)310 pages
- Epistemological and Theoretical Foundations in Language Policy and PlanningNo ratings yetEpistemological and Theoretical Foundations in Language Policy and Planning145 pages
- Text Linguistics The Subject and Evolution of Text LinguisticsNo ratings yetText Linguistics The Subject and Evolution of Text Linguistics5 pages
- Lexical and Phonological - Two Levels of Stylistics: An Analytical Study of Ted Hughes' PoemsNo ratings yetLexical and Phonological - Two Levels of Stylistics: An Analytical Study of Ted Hughes' Poems6 pages
- A Kurdish Grammar - Descriptive Analysis of The Kurdish of Sulaimaniya, IRAQ (Ernest N McCarus) - ED089545No ratings yetA Kurdish Grammar - Descriptive Analysis of The Kurdish of Sulaimaniya, IRAQ (Ernest N McCarus) - ED089545146 pages
- What, If Anything, Is Typology?: Johanna NicholsNo ratings yetWhat, If Anything, Is Typology?: Johanna Nichols8 pages
- Minimalist Essays On BP Syntax - Jairo NunesNo ratings yetMinimalist Essays On BP Syntax - Jairo Nunes126 pages
- Ross 1967 Constraints On Variables in SyntaxNo ratings yetRoss 1967 Constraints On Variables in Syntax529 pages
- (Bernd - Heine, - Derek - Nurse) A Linguistic Geography of Africa PDF67% (3)(Bernd - Heine, - Derek - Nurse) A Linguistic Geography of Africa PDF390 pages
- The Limits of Grammaticalizatio - Giacalone-Ramat, Anna (Editor)No ratings yetThe Limits of Grammaticalizatio - Giacalone-Ramat, Anna (Editor)309 pages
- Lars Johanson Code Copying The Strength of LanguagesNo ratings yetLars Johanson Code Copying The Strength of Languages165 pages
- Lecture 1. Stylistics As A Science: 1. The Concept of Style in LinguisticsNo ratings yetLecture 1. Stylistics As A Science: 1. The Concept of Style in Linguistics6 pages
- Prescriptive vs. Descriptive LinguisticsNo ratings yetPrescriptive vs. Descriptive Linguistics12 pages
- Lingg 206: Grammatical Analysis I Angelina A. Aquino100% (1)Lingg 206: Grammatical Analysis I Angelina A. Aquino25 pages
- (Leech, G. (1985) The Study of Meaning (Semantics)No ratings yet(Leech, G. (1985) The Study of Meaning (Semantics)31 pages
- Grammaticalization Heiko Narrog Full Digital ChaptersNo ratings yetGrammaticalization Heiko Narrog Full Digital Chapters139 pages
- Ateneo Law Journal Legal Citation Primer PDF75% (4)Ateneo Law Journal Legal Citation Primer PDF21 pages
- Landmark Judgment On Interpretation of Commercial Contract: Lawweb - inNo ratings yetLandmark Judgment On Interpretation of Commercial Contract: Lawweb - in35 pages
- Awwa Standard For Pilot Operated Control ValvesNo ratings yetAwwa Standard For Pilot Operated Control Valves32 pages
- M2 - Five Facets of Music Education - SingingNo ratings yetM2 - Five Facets of Music Education - Singing5 pages
- Family Violence in Cross-Cultural Perspective: David LevinsonNo ratings yetFamily Violence in Cross-Cultural Perspective: David Levinson21 pages
- Wrangled by The Alien Rancher 40 - Ursa DaxNo ratings yetWrangled by The Alien Rancher 40 - Ursa Dax177 pages
- YOkeno 4 S 6 Oi COXCGCTABUQBWk BB 1 F FNo ratings yetYOkeno 4 S 6 Oi COXCGCTABUQBWk BB 1 F F13 pages
- Alvin C. Burns Louisiana State University Ronald F. Bush University of West Florida Prentice Hall PublishersNo ratings yetAlvin C. Burns Louisiana State University Ronald F. Bush University of West Florida Prentice Hall Publishers31 pages
- English Faculty (2013), Advanced English Grammar - Eng204 - TextNo ratings yetEnglish Faculty (2013), Advanced English Grammar - Eng204 - Text82 pages
- Atunus Furniture Catalog - NPC (2024-09-29 02 - 53 - 55)No ratings yetAtunus Furniture Catalog - NPC (2024-09-29 02 - 53 - 55)81 pages
- Nusrat Jahan Anika Internship Report 2 2No ratings yetNusrat Jahan Anika Internship Report 2 256 pages
- Lesson 3 (Prelim) - Language of MathematicsNo ratings yetLesson 3 (Prelim) - Language of Mathematics50 pages
