English Linguistics

English Linguistics
A Coursebook for Students of English

by
Thomas Herbst

De Gruyter Mouton
ISBN 978-3-11-020367-7
e-ISBN 978-3-11-021548-9

Library of Congress Cataloging-in-Publication Data

Herbst, Thomas.
English linguistics : a coursebook for students of English / by
Thomas Herbst.
p. cm.
Includes bibliographical references and index.
ISBN 978-3-11-020367-7 (pbk. : alk. paper)
1. Linguistics. 2. Language and languages. 3. English lan-
guage ⫺ Textbooks for foreign speakers. I. Title.
P121.H56 2010
428.214⫺dc22
2010017332

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

쑔 2010 Walter de Gruyter GmbH & Co. KG, Berlin/New York

Cover image: Patrick Heron, detail from Window for Tate Gallery St Ives ⫺ ” Estate of
Patrick Heron. ” Tate, London 2008. All rights reserved. DACS, UK and VG-Bild-Kunst,
Bonn, Germany.
Printing: AZ Druck und Datentechnik GmbH, Kempten
⬁ Printed on acid-free paper
Printed in Germany
www.degruyter.com
Contents

Preface....................................................................................................... xiii

The English language and linguistics

1 Facts about English...................................................................1
1.1 English world-wide...................................................................1
1.2 Regional and social variation....................................................3
1.3 Historical variation ...................................................................4
1.4 The character of English...........................................................6
1.4.1 English as a Germanic language...............................................6
1.4.2 Language typology ...................................................................9
1.5 The linguistic analysis of English...........................................10
2 Principles of modern linguistics .............................................12
2.1 Basic concepts of linguistic structuralism ..............................12
2.1.1 Principles of linguistics since de Saussure .............................12
2.1.2 The character of the linguistic sign.........................................14
2.1.3 Synchronic and diachronic study of language ........................16
2.1.4 The importance of relations ....................................................17
2.1.4.1 The value of the linguistic sign...............................................17
2.1.4.2 Syntagmatic and paradigmatic relationships ..........................19
2.1.5 Schools of structuralism .........................................................20
2.2 Linguistics and descriptivity...................................................20
2.3 The principles of structuralism and foreign language
teaching...................................................................................22
2.4 Areas of investigation .............................................................25
3 Language, intuition and corpora .............................................27
3.1 Language ................................................................................27
3.1.1 Some basic distinctions...........................................................27
3.1.2 Competence and performance: the language of the indi-
vidual ......................................................................................28
3.1.3 Language as a social phenomenon .........................................29
3.1.4 System and Norm – language use............................................29
3.2 Finding data: traditional methods ...........................................31
3.2.1 Principal options .....................................................................31
vi Contents

3.2.2 Introspection and elicitation ...................................................32

3.2.3 Authentic language material: citations and corpora ...............33
3.3 Corpus linguistics ...................................................................33
3.3.1 Corpora of English..................................................................33
3.3.2 What we can do with corpora .................................................37
3.3.2.1 Corpus analysis.......................................................................37
3.3.2.2 Corpora and foreign language teaching ..................................40
3.3.3 Corpus design and corpus size................................................41
3.4 Introspection, corpus analysis and views of language............42

Sounds
4 The sounds of English: phonetics ...........................................43
4.1 Sounds as the starting point of linguistic analysis ..................43
4.2 Phones.....................................................................................43
4.3 Articulatory, auditive and acoustic phonetics.........................45
4.4 Description of sounds in articulatory terms............................48
4.5 Syllables..................................................................................53
4.6 Suprasegmental elements........................................................54
5 Phonology...............................................................................56
5.1 The function of speech sounds ...............................................56
5.1.1 Phonemes and allophones.......................................................56
5.1.2 Phonetics and phonology........................................................58
5.2 The description of phonemes..................................................58
5.2.1 Consonant phonemes ..............................................................58
5.2.2 Vowel phonemes ....................................................................60
5.2.3 Phonemic principle of pronunciation dictionaries ..................64
5.3 Phonotactics............................................................................65
6 Phonetic “reality” ...................................................................67
6.1 Problems of the phoneme concept ..........................................67
6.1.1 The problem............................................................................67
6.1.2 Phonetic value of phonological features.................................68
6.1.3 The bi-uniqueness requirement...............................................71
6.2 Pronunciation in connected speech.........................................73
6.2.1 Weakening of elements...........................................................73
6.2.2 Linking phenomena ................................................................74
6.2.3 Weak forms.............................................................................75
7 Contrastive aspects of phonetics and phonology....................76
7.1 Levels of contrast....................................................................76
7.2 Phoneme and phone inventories of English and German .......76
 Contents vii

7.3 Rule-governed differences......................................................79

7.4 Suprasegmental differences ....................................................80
7.5 Pedagogical implications ........................................................81

Meaning-carrying units
8 Morphology ............................................................................83
8.1 The concept of the morpheme ................................................83
8.2 Types of morpheme ................................................................85
8.3 Problems of a static morpheme concept .................................87
8.3.1 The problem............................................................................87
8.3.2 Portmanteau morphs ...............................................................87
8.3.3 Zero-morphs ...........................................................................88
8.3.4 Morphological and phonological conditioning.......................89
8.4 Inflectional morphology: historical background ....................91
8.5 Further problems of morphological analysis ..........................92
9 Word formation ......................................................................95
9.1 Words......................................................................................95
9.1.1 Words and lexemes.................................................................95
9.1.2 New words..............................................................................98
9.2 Word formation ....................................................................100
9.2.1 Introduction ..........................................................................100
9.2.2 Formal types of word formation: a survey ...........................102
9.2.3 Semantic description of word formations.............................105
9.3 Word formation and morphology .........................................108
9.3.1 The overlap between word formation and morphology........108
9.3.2 Explanatory value of the analysis .........................................111
9.4 Productivity and restrictions .................................................113
9.5 Possible words – nonce formations – institutionalized
words ....................................................................................115
9.6 Psychological aspects of morphology ..................................120
10 Phraseology ..........................................................................125
10.1 Prefabs ..................................................................................125
10.2 Statistically significant collocations .....................................128
10.3 Institutionalized collocations ................................................131
10.4 Idioms ...................................................................................134
10.5 The idiom principle and the mental lexicon .........................136
10.6 Phraseological units ..............................................................138
viii Contents

Sentences – models of grammar

11 Syntax: traditional grammar .................................................141
11.1 Syntax and grammar .............................................................141
11.1.1 Descriptive frameworks........................................................141
11.1.2 Sentence and clause ..............................................................142
11.1.3 Subject and predicate............................................................144
11.2 The elements of clause structure in CGEL ...........................147
11.2.1 Elements of clause structure as functional units ...................147
11.2.2 Criteria for the distinction between different elements
of clause structure .................................................................148
11.2.3 CGEL’s clause types ............................................................151
11.2.4 Problems of traditional terminology .....................................152
11.3 Phrases ..................................................................................153
11.3.1 Types of phrase.....................................................................153
11.3.2 The role of the phrase ...........................................................157
11.4 Word classes .........................................................................157
11.4.1 Criteria for the establishment of word classes ......................157
11.4.2 CGEL’s word classes............................................................160
11.4.3 Verbs.....................................................................................161
11.4.4 Central and peripheral members of word classes – word
classes as prototypes .............................................................162
11.4.5 Multiple-class membership...................................................164
11.4.6 The distinction between determiners and pronouns .............165
11.4.7 The distinction between prepositions and subordinating
conjunctions..........................................................................167
11.4.8 Word classes in English........................................................168
12 Valency theory and case grammar........................................171
12.1 Two types of hierarchy .........................................................171
12.1.1 Constituency .........................................................................171
12.1.2 Dependency ..........................................................................173
12.1.3 Case grammar and valency theory........................................176
12.2 Case grammar: semantic roles ..............................................176
12.2.1 Basic principles of case grammar .........................................176
12.2.2 Advantages and drawbacks of case grammar .......................178
12.2.3 Some useful participant roles................................................180
12.3 The basic principles of valency theory .................................183
12.3.1 Introduction ..........................................................................183
12.3.2 Complements and adjuncts ...................................................183
12.3.3 Qualitative and quantitative aspects of valency....................185
 Contents ix

12.3.4 Valency carriers ....................................................................187

12.3.5 Components of a valency description...................................188
12.3.6 Valency patterns ...................................................................191
12.4 A valency based approach to English syntax .......................192
12.4.1 Combining aspects of clause structure and valency .............192
12.4.2 A modified view of phrase structure.....................................194
12.4.2.1 Head complexes....................................................................194
12.4.2.2 Noun phrases, adjective phrases and adverb phrases ...........195
12.4.2.3 Particle phrases .....................................................................197
12.4.2.4 Clauses as verb phrases ........................................................197
12.4.3 Description of units ..............................................................198
12.4.4 Example ................................................................................198
13 Theories of grammar and language acquisition ....................200
13.1 Chomsky’s approach ............................................................200
13.1.1 Basic assumptions.................................................................200
13.1.2 Transformations – deep structures and surface structures ....202
13.1.3 Claims and evidence .............................................................205
13.1.4 Language acquisition ............................................................208
13.1.4.1 The language acquisition device...........................................208
13.1.4.2 Universal grammar ...............................................................209
13.2 Usage-based approaches.......................................................210
13.2.1 Construction grammar ..........................................................210
13.2.2 Argument structure constructions.........................................212
13.2.3 The usage-based view of language acquisition.....................215

Meaning
14 Semantics: meaning, reference and denotation.....................220
14.1 Meaning ................................................................................220
14.2 Meaning and reference .........................................................221
14.2.1 Bloomfield’s misconception of meaning..............................221
14.2.2 Denotation ............................................................................223
14.2.3 Reference ..............................................................................224
14.2.3.1 The general notion of reference............................................224
14.2.3.2 Definite and indefinite reference ..........................................226
14.3 The scope of meaning...........................................................229
15 Meaning relations .................................................................233
15.1 Polysemy and homonymy ....................................................233
15.1.1 Polysemy and homonymy in linguistic analysis...................233
15.1.2 Psycholinguistic and lexicographical implications...............237
x Contents

15.2 Ambiguity.............................................................................237
15.3 Problems of identification of meanings and lexical units.....238
15.4 Structural semantics..............................................................239
15.4.1 The idea of contrast ..............................................................239
15.4.2 Semantic relations.................................................................240
15.4.2.1 Hyponymy: unilateral entailment .........................................240
15.4.2.2 Synonymy: bilateral entailment............................................241
15.4.2.3 Semantic oppositions ............................................................243
16 Ways of describing meaning ................................................247
16.1 Componential analysis..........................................................247
16.2 The structure of vocabulary ..................................................252
16.3 Vocabulary and conceptualization........................................253
16.4 Prototype theory ...................................................................256
16.4.1 Colour terms .........................................................................256
16.4.2 Prototypes .............................................................................258
16.4.3 Basic level categories ...........................................................261
16.4.4 Problems of prototype theory ...............................................263

Utterances
17 Pragmatics ............................................................................265
17.1 Word, sentence and utterance meaning ................................265
17.1.1 Sentence meaning .................................................................265
17.1.2 The meaning of utterances....................................................266
17.2 Principles ..............................................................................268
17.2.1 The co-operative principle and conversational implica-
ture ........................................................................................268
17.2.2 Further principles..................................................................270
17.3 Speech acts ...........................................................................271
17.3.1 Performatives and constatives ..............................................271
17.3.2 Locutionary, illocutionary and perlocutionary acts ..............275
17.3.3 Felicity conditions ................................................................277
17.3.4 Types of speech act...............................................................278
17.3.4.1 Searle’s taxonomy ................................................................278
17.3.4.2 Direct and indirect speech acts .............................................280
17.3.4.3 Problems of classification.....................................................281
18 Texts .....................................................................................283
18.1 The notion of text .................................................................283
18.1.1 Cohesion and coherence .......................................................283
18.1.2 Texts as utterances................................................................287
 Contents xi

18.2 Cohesive relations.................................................................287

18.2.1 Explicit linking expressions..................................................287
18.2.2 Grammatical aspects of relating referents and meanings......288
18.2.3 Lexical aspects of cohesion and coherence ..........................291
18.3 Thematic structure and information structure.......................295
18.3.1 Theme and rheme – given and new information ..................295
18.3.2 End-focus and marked focus ................................................297
18.4 Spoken and written texts.......................................................298

Variation
19 Variation in language............................................................302
19.1 Registers and dialects ...........................................................302
19.2 Accent, dialect, standard and prestige ..................................305
19.2.1 Standard English and its pronunciations...............................305
19.2.2 Quality judgements...............................................................307
19.3 Levels of differences between regional and social varie-
ties.........................................................................................309
20 Linguistic change..................................................................315
20.1 Types of linguistic change ....................................................315
20.2 Sound change........................................................................316
20.2.1 The phoneme systems of Old English and RP......................316
20.2.2 Types of sound change .........................................................318
20.2.3 Important sound changes in the history of English ..............319
20.2.3.1 I-mutation .............................................................................319
20.2.3.2 The Great Vowel Shift..........................................................319
20.2.3.3 Quantitative changes.............................................................321
20.2.3.4 Present-day reflections..........................................................321
20.3 Lexis .....................................................................................322
20.3.1 New words............................................................................322
20.3.2 Changes of meaning .............................................................324
20.3.3 Homonymy ...........................................................................325
20.4 Grammar ...............................................................................325
20.4.1 Differences between Old English and Modern English........325
20.4.2 Analogy ................................................................................327
20.4.3 Grammaticalization...............................................................328

Postscript ...................................................................................................330
Bibliography..............................................................................................332
Index..........................................................................................................365
Preface

English Linguistics is intended as an introduction to a field that, as such,

perhaps does not even exist. The idea of this book is to introduce students
of English to basic concepts of linguistics that are relevant to the descrip-
tion and analysis of the English language and to ideas and approaches that
are relevant in this context. These, of course, apply not only to English but
also to other languages.
In view of the wide range of different subjects and the great number of
different theoretical and methodological approaches comprised by linguis-
tics, it is perfectly obvious that a selection must be made with respect to the
topics and approaches that can be discussed within the scope of an intro-
ductory book. Any selection of this kind will inevitably involve a personal
element in the choice and treatment of particular topics. The present book
has emerged from a manuscript that has been used for courses during the
first year in English linguistics at the Friedrich-Alexander-Universität Er-
langen-Nürnberg, which means that it also includes aspects of the analysis
of English that appear particularly relevant in a foreign language context,
which is why reference is occasionally made to foreign language teaching
and to other languages, in particular German.
This book attempts to be a general introduction to central aspects of the
academic study of the English language. In my view, this entails two equal-
ly important components:
– On the one hand, this book provides the reader with an introduction to
important areas of investigation and the basic concepts as well as the
terminology used for the analysis of the respective fields. Thus, the
book deals with the description of sounds, words, sentences, texts etc.
introducing terms such as phoneme, compound, subject or coherence
that are central to their analysis.
– On the other hand, however, it is one of the main concerns of this intro-
duction that the reader should become aware of the fact that there are no
(or not many) easy answers or straightforward solutions in linguistics. It
is important to realize, for instance, that although many linguists use the
term object in the analysis of sentences, they do not necessarily agree on
what they mean by it, or – more radically, that some linguists (or
xiv Preface

schools of linguistic thought) are firmly convinced of certain assump-

tions that should be made about language and principles that should be
followed in the analysis of language that are strongly rejected by others.
In other words, one must recognize that in linguistics there is very little in
the way of “truth”, but that different linguists have developed different
ideas about how we can account for language or for certain linguistic phe-
nomena. In some cases, the best or most appropriate way of describing a
particular phenomenon may depend on the purpose of the description; in
other cases, it may be a matter of personal conviction. Nevertheless, it is
important to understand that it is precisely this struggle between different
approaches that determines the nature of this subject (as is true of any other
academic subject). It is for this reason that this introduction not only tries to
describe and explain central concepts of linguistics, but also to indicate that
the concepts outlined represent one (possibly the most established and
commonly used) way of seeing things. Of course, it would be beyond the
scope of an introduction of this kind – and terribly confusing for the reader
– to mention all approaches to a particular phenomenon that might deserve
mentioning. In some cases, however, reference is made to alternative ac-
counts – often in footnotes or in special sections which are then marked by
grey backgrounds in the chapter headings as being intended for more ad-
vanced readers.
This means that this book does not necessarily have to be read from
cover to cover in one go. It is perfectly possible for beginners to work
through the chapters dealing with more introductory subjects first and then
to come back to the more advanced chapters of the same sections at a later
stage.
This book owes a lot to help and support from a number of colleagues
and friends. I would like to thank in particular Dr. Susen Faulhaber, Dr.
Katrin Götz-Votteler, David Heath, Dr. Michael Klotz, Kevin Pike, Dr.
Stefan Thim and Peter Uhrig for reading the entire manuscript and for their
valuable comments and suggestions, their patience and the many discus-
sions we had on different aspects of the text. Furthermore, I owe thanks to
Eva Klein, Dominic Losse, Thomas Maisel, Dr. Brigitta Mittmann, Elisa-
beth Reber, Dr. Christina Sanchez and many of my students for comment-
ing on earlier versions or parts of the manuscript. Barbara Gabel-
Cunningham deserves special thanks for her extremely competent work in
getting the manuscript into a publishable form, and equally I would like to
thank Christian Hauf and Sabine Menz for their assistance in preparing it.
 Preface xv

I very much hope that the present introduction succeeds in demonstrat-

ing the fascination of many ideas discussed in linguistics today and the
importance of knowing about linguistics for everybody who teaches lan-
guage.
 The English language and linguistics

1 Facts about English

1.1 English world-wide

English is spoken – in different forms or varieties – by people all over the

world. At the beginning of the twenty-first century, it is the most important
language used in international communication. Estimates suggest that about
1.5 billion people have some command of English today. This does not
only include people who have English as their mother tongue (or first lan-
guage L1), but it is more than likely that a Norwegian and an Italian or
someone from Japan and someone from Belgium will make use of English
as a lingua franca when communicating with one another. In fact, there are
considerable differences between the functions that English has for differ-
ent people and the status the language has in different countries of the
world.1
− English as a native language (ENL): English is spoken as a native lan-
guage by the majority of people living in North America, the British
Isles, Australia or New Zealand, for example.
− English as a second language (ESL): in some countries people whose
native language is not English use English (which they have acquired
after or at the same time as another language) as a second language in
some situations such as certain educational, commercial or governmen-
tal contexts. This may be the case in countries such as Canada, where
many French-speaking Canadians also speak English. It is also the case

1
This account is based on the approach taken in CGEL (I.4-5). For an account
of the distinction between ENL, ESL and EFL in terms of “social-linguistic
constellations in which English is used” see Mair (2008: 151–156). Cf. also
Trudgill and Hannah (52008: 4–8).
2 The English language and linguistics

in countries such as India or Nigeria, in which several native languages

co-exist.2
− English as a foreign language (EFL): the group of speakers of English
as a foreign language (EFL) is very heterogeneous as far as the actual
command of the language is concerned.
Estimates of the number of speakers of different languages at the beginning
of the twenty-first century vary: Crystal (2006: 424) gives the following
figures:3
L1 speakers 400 million
L2 speakers 400 million
foreign language speakers 600 – 700 million
Although English neither has the largest number of L1 speakers in the
world (Chinese) nor in the European Union (German) the remarkably high
figures for foreign speakers of English are an indication of its status as a
world language. Another important aspect of the role of English in the
world is that countries in which English is at least the main language are
spread over four continents: Europe (Britain and Ireland), Australia, Amer-
ica (USA and parts of Canada), Africa (South Africa).4 House (2002: 246)
mentions four factors for English becoming a “truly global language”: “the

2
According to CGEL (1985: 1.4) English is the only official language in Nige-
ria, whereas in India several languages have that status. For the situation in
Nigeria compare, however, Gut (2008: 35), who points out that “currently
about 20% of the population have some command of English”. For the status
of English in East Africa or Ghana see Schmied (2008: esp. 154–155) and
Huber (2008: esp. 73) respectively, for India see Schneider (2007: 161). For
Africa see Schmied (1991: esp. 23-45).
3
Obviously, such figures have to be treated with great caution, see e.g. Crystal
(2003: 108–109), who gives an estimate of 329m L1 and 433m L2 speakers.
Compare the figures given by Viereck, Viereck, and Ramisch (2002: 242):
355m L1 speakers, 100m L2 speakers and 150m speakers of English as a for-
eign language. For other languages see Viereck, Viereck, and Ramisch (2002:
236): Chinese 1,110m, Spanish 305m, Arabic 220m, Portuguese over 160m,
Russian 160m, French 124m, German 121m, Italian 66m.
4
For details see Crystal (1988: 8–9). For factors contributing to the status of
English as international language – such as political and economic factors, its
use in advertising, as a means of communication in aviation etc. or academic
publications – see Crystal (2006: 426–431) and Viereck, Viereck and
Ramisch (2002: 238–245).
 1 Facts about English 3

worldwide extension of the British Empire; the political and economic rise
of the United States to world power status after the Second World War; the
unprecedented developments in information and communication technolo-
gies; and the recent economic developments towards globalisation and in-
ternationalisation”. It is important to realize that this kind of status a lan-
guage or a variety has is entirely derived from external factors and has
nothing to do with particular qualities or characteristics of the language in
question. The present status of English is a reflection of historic and econ-
mic developments but one must doubt whether it has very much to do with
the way plurals or the present perfect are formed in English.

1.2 Regional and social variation

When we say that the English language is spoken by 300 or 400 million
people as a mother tongue today, then this is not to say that the actual lan-
guage they use is absolutely the same. Quite the opposite is true, of course.
Not only does language differ from one individual to another – in linguis-
tics the language of an individual is termed their idiolect –, there are also
remarkable differences between the language of different groups of speak-
ers, for which the term dialect – or the more neutral term variety (> 19.1) –
is used.
One can make a distinction between

regional dialects, which are determined by the geographical distribu-
tion of certain linguistic forms,

social dialects (or sociolects), which are determined by the social group
to which their speakers typically belong.
It has to be borne in mind, however, that such varieties hardly ever exist in
a discrete and clearly distinguishable form.5 Furthermore, regional variation
can be described in different degrees of specificity: within the British Isles,
for example, we can describe dialects such as Scottish English, Irish Eng-
lish, Northern English English or South-Western English or Belfast Eng-

5
Furthermore, speakers are not necessarily very consistent in their speech, in
that people tend to modify the kind of language they are using depending on
who they are talking to, what they are talking about, whether they are spea-
king or writing etc. See Chapter 19.1.
4 The English language and linguistics

lish, Birmingham English or Cockney (a dialect spoken in London) etc.6

Taking a world-wide perspective, one can identify varieties of English such
as Canadian English, Australian English, South African English, which are
sometimes called national varieties. Furthermore, the world-wide spread of
English has lead to language contact with other languages, which has re-
sulted in the emergence of new varieties of English, sometimes called New
Englishes, which have become a very important area of study.7
It is one of the interesting facts about the English language that there is
no official institution – such as the Paris Académie Française for French –
that would take any decisions about how English should be used. Neverthe-
less there exists a type of Standard English, which differs from other dia-
lects or varieties in terms of the functions it has because it is the variety that
is commonly used in printing and serves as a model of foreign language
teaching (> 19.1). It is possible to identify different national standards, of
which British English and American English are the most important in-
ternationally.
Standard English can be pronounced with a range of different accents
(the term accent referring to variation with respect to pronunciation). The
most neutral (in the sense of least regional) pronunciations of Standard
English are generally referred to as General American (or Network Eng-
lish) for American English and BBC English or Received Pronunciation
(RP) for British English. RP has a special status because it shows practi-
cally no regional variation within England and Wales, but is to a very high
degree associated with a particular social class (> 19.2).

1.3 Historical variation

Apart from variation according to region and social group, there is also
variation in time. The extent to which English has changed over the centu-
ries is illustrated by the following examples:8

6
For the dialect areas and the Survey of English Dialects see Barnickel (1980:
145–151).
7
For different phases in the evolution of postcolonial Englishes see Schneider
(2007: 29–55).
8
Sources: P.G. Wodehouse: Blandings Castle, Harmondsworth: Penguin
(1935/1954: 1). – William Shakespeare: The Complete Works. Original-
Spelling Edition, edited by Stanley Wells and Gary Taylor, Oxford: Oxford
University Press (1986: 1108). – The Riverside Chaucer, 3rd edition (edited
 1 Facts about English 5

(1) ‘Beach,’ said Lord Emsworth.

‘M’lord?’
‘I’ve been swindled. This dashed thing doesn’t work.’
‘Your lordship cannot see clearly?’
… ‘Perhaps if I were to remove the cap at the extremity of the instru-
ment, m’lord, more satisfactory results might be obtained.’
… ‘Ah!’ There was satisfaction in Lord Emsworth’s voice. He twiddled
and adjusted, and the satisfaction deepened. ‘Yes, that’s better. That’s
capital.’...
P.G. Wodehouse: Blandings Castle
(2) LADY MACBETH He has almost supt: why haue you left the cham-
ber?
MACBETH Hath he ask’d for me?
LADY MACBETH Know you not, he ha’s?
William Shakespeare: Macbeth
(3) Whan that Aprill with his shoures soote
The droghte of March hath perced to the roote,
And bathed every veyne in swich licour
Of which vertu engendred is the flour ...
Geoffrey Chaucer: Canterbury Tales
(4) þa gesette se munuc ealla þa gerecednesse an anre bec, and eft ða þa
seo boc com to us binnan feawum gearum, þa awende we hit on Englisc,
swa swa hit heræfter stent.
Ælfric: King Edmund
As one can see from these examples, like varieties of present-day English,
historical varieties differ in pronunciation, spelling, vocabulary, grammar
etc. Even the Wodehouse text, which was first published in 1935, contains
a number of expressions that one would not expect to be used today or that
one would consider old-fashioned, and the changes between the earlier
periods and today are rather dramatic. The sample from Ælfric’s account of
the life of King Edmund, written more than 1000 years ago, cannot be un-
derstood by native speakers of English today. It reflects a stage of the lan-
guage at which English was much closer in character to German in that, for
____________________________
by Larry D. Benson, Oxford: Oxford University Press, 1987: 23). – Ælfric:
St. Edmund (taken from: Rolf Kaiser: Medieval English, Berlin, 5th impres-
sion 1961: 152).
6 The English language and linguistics

instance, nouns, adjectives, pronouns inflected for case and number in a

way that they no longer do in modern English.
The texts given here represent the four phases into which the develop-
ment of the English language is generally divided:

Old English, ca. 600 – ca. 1100, characterized by a complex system of
inflexions, by a relatively homogeneous vocabulary of predominantly
Germanic origin

Middle English, ca. 1100 – ca. 1500, characterized by an increasing
loss of inflexions and a tremendous change in vocabulary, with a large
number of words of French origin coming into English

Early Modern English, ca. 1500 – ca. 1750, a period in which English
was subject to the Great Vowel Shift and a standard language gradually
emerged

Modern English, ca. 1750 – today. Sometimes, the term Late Modern
English is used for the period between 1750 and 1900 and the English
after 1900 is referred to as Present-Day English.9

1.4 The character of English

1.4.1 English as a Germanic language

As far as its place within the world’s languages is concerned, English is a

Germanic language, i.e. a member of the Indo-European family of lan-
guages.10

9
For slightly different periodisations of English compare Brinton and Arno-
vick (2006: 10–11), Denison and Hogg (2006: 2–3) or Mair (2008: 189–195).
For a discussion of periodisation see Lutz (2002b).
10
Based on Denison and Hogg (2006: 5). Compare Brinton and Arnovick
(2006: 94–103) and Crystal (1997: 300), also with respect to Yiddish and
Flemish.
 1 Facts about English 7

Indo-European languages

Albanian
Germanic

Tocharian
Armenian
Anatolian
Celtic

Hellenic

Iranian
Slavic
Balto-

Indo-
Italic

West North East

English Icelandic Gothic

Frisian Faeroese
Low German Norwegian
Dutch Swedish
Afrikaans Danish
High German

Although as far as its historical origins are concerned English is quite

clearly to be regarded as a Germanic language, it has to be noted that the
vocabulary of present-day English can be seen as a mixture of words of
Germanic and Romance origin. The reasons for the development from a
language with an almost entirely Germanic vocabulary to the present situa-
tion lie, at least partly, in general historical developments in England and
the kind of language contact in which they resulted.
The first contact that must have taken place after the arrival of the An-
glo-Saxons in England in the middle of the fifth century is that between
English and the language spoken by the Celts inhabiting the island at the
time. However, the Celts were driven by the Anglo-Saxons into the fringe
regions of Scotland, Wales, Ireland, Cornwall and Brittany, where in some
parts Celtic languages are still spoken today. Until recently, it was believed
that there had been very little contact between the English and the Celts,
partly because very few Celtic words can be found in English. More recent
research suggests, however, that the contacts may have been more intensive
and are reflected linguistically in certain peculiarities of the Old English
grammatical system compared with that of other Germanic languages (>
20.1).
8 The English language and linguistics

As far as vocabulary is concerned, the languages to which English owes

most are Latin, French, Scandinavian languages and Greek.11
1. Latin: The Romans, who had occupied parts of Britain between 43 and
407, had already withdrawn from the island when the Anglo-Saxons ar-
rived. There had been contact between the Anglo-Saxons and the Ro-
mans on the Continent, however, which had indeed resulted in borrow-
ings such as port or mile. Furthermore, there is the possibility that the
Celts living in Britain during the Roman occupation took over Latin
words, which were later taken over by the Anglo-Saxons. During the
Old English period, towards the end of the sixth century, the Christiani-
zation of the Anglo-Saxons set in, starting in Canterbury. Christian
monasteries were established and became centres of learning. It is
through translations of religious and scientific texts from Latin into Eng-
lish that were carried out at the monasteries that Latin words were intro-
duced into English on a large scale (disciple, decline).
Later, during the Renaissance, a great number of Latin and Greek
words were taken over on a large scale during the period of Neoclassi-
cism.
2. Scandinavian languages: During the ninth and tenth centuries, the
North-Eastern part of England, the so-called Danelaw, was settled by
Scandinavians and was politically a part of Denmark. This, in the long
term, resulted in Scandinavian vocabulary being taken over by English.
Although the number of Scandinavian loan words is relatively small, the
Scandinavian influence is important because many highly frequent
words such as take (which replaced Old English niman) and the pro-
nouns they, their, them were taken over from the Scandinavian lan-
guages.
3. French: By far the most dramatic change in the history of England and
the history of English is represented by the Norman Conquest in 1066,
which resulted in a situation in which practically the whole ruling class
of the country was replaced by French-speaking nobility. For three hun-
dred years, French was an official language in England, it was the lan-
guage of the courts and of government; it was as late as 1362 that the
London Parliament was opened in English for the first time. Thus the

11
For a short outline of British history in this context see e.g. Denison and
Hogg (2006: 8–29). For developments in vocabulary see also Chapter 20.3
and the sources given there, esp. Kastovsky (2006).
 1 Facts about English 9

vast intake of words from French (such as crown, government, accuse,

appetite, to give just a few examples) during that period is hardly sur-
prising.
The result of these processes is the so-called mixed vocabulary of present-
day English. There are word counts that suggest that more than 50%12 of
the vocabulary of English is Romance in character (taken over from
French, Latin or Italian), although one must not overlook the fact that many
(although by no means all) of these words have a considerably lower fre-
quency than many Germanic words. More detailed research on this is cer-
tainly needed to obtain reliable data. This is one of the areas where modern
corpus linguistics provides a new and much more reliable empirical basis.
A first analysis of the 3600 words that according to the Longman Diction-
ary of Contemporary English (LDOCE4, 2003) make up the 3000 most
frequent words of the spoken language and the 3000 most frequent words
of the written language in the British National Corpus has revealed that of
these words about 34% are of Germanic origin and 54% of Romance ori-
gin.13
It is sometimes argued that the mixed character of its vocabulary makes
English particularly suitable for the status of a world language and lingua
franca which it occupies today. It must be said, however, that there is no
empirical evidence to support such a view and that any explanation for the
emergence of English as an international language in terms of cultural and
political dominance is probably more powerful.

1.4.2 Language typology

Languages cannot only be classified according to their historical origins,

but, in a typological way, according to certain features they show. The fol-
lowing types of languages are commonly distinguished:14

12
Cf. Scheler (1977: 72), where a much more differentiated account is given.
See also Lutz (2002a: 410).
13
The basis of these figures is provided by LDOCE4 (for frequency) and the
Oxford English Dictionary (OED2) (for the etymological source). Thanks to
Peter Uhrig for his help and particularly for designing a computer program to
work out these figures.
14
This kind of classification goes back to Wilhelm von Humboldt and August
Wilhelm Schlegel; cf. Crystal (1980: 367) and Bußmann (42008: 664–666).
10 The English language and linguistics

agglutinating languages (such as Turkish, Finnish, Hungarian or

Basque), in which grammatical relations are expressed by unambiguous
endings, which are added to words.

inflecting languages (such as Latin, Arabic, French or German), in
which grammatical relations can be expressed by endings, which can,
however, have different meanings (Fahrzeug-e: nominative plural and
accusative plural), and by changes of the stem (singen – sang).

isolating languages (such as Chinese or Vietnamese), in which words
are invariable and grammatical relations are expressed by word order or
separate words, but not by grammatical endings.
Although these types present prototypical descriptions and further types
can be established, one can say that English has developed from a language
of predominantly inflecting character to a language which shows strong
characteristics of an isolating language.

1.5 The linguistic analysis of English

This short characterization of the English language opens up a large spec-

trum of topics which can be investigated in linguistics: questions of lan-
guage typology are related to questions of language change, which in turn
is related to linguistic variation. Research on how a language developed can
go hand in hand with research on how speakers of different varieties or of
different languages behave when they communicate with one another, etc.
In principle, such questions arise in the study of any language. Never-
theless, it has to be said that the international status of English has affected
its linguistic analysis. Firstly, because the kind of language contact phe-
nomena one encounters in the study of English are manifold in that they
include its status with respect to minority languages as in Wales, for exam-
ple, the status of English in countries such as India or in worldwide com-
munication and the influence that English has on other languages through
its present position. The mere power of this influence and the reactions
against, say, the use of anglicisms in Germany, or, perhaps in an even more
pronounced way in France, gives English a special status in this respect.
This also applies to other areas of linguistic description, of course. The
mere fact that English is a or the world language at present has resulted in
____________________________
See Crystal (1980) for short definitions of these and further distinctions made
in language typology. For analytic and synthetic see Bußmann (42008: 664).
 1 Facts about English 11

enormous research being carried out on it, especially where commercial

interests are involved. The amount of teaching materials, dictionaries or
grammar books on English available in Germany, for example, by far su-
percedes that for any other language and, as a result, fostered by large po-
tential sales figures and by commercial competition, a very high standard
has been reached in many such publications.
English is probably the best described language in the world, partly be-
cause of the commercial aspects just mentioned, but also because a great
deal of linguistic research is carried out in the United States and Britain.
This has resulted in a situation in which many innovative ideas and re-
search tools ranging from computational corpus linguistics to generative
language theory, which are relevant for the study of all languages, have
been developed in the context of English. In fact, many claims made about
universal characteristics of language as such are made on the basis of an
analysis of English so that one might argue that English has replaced Latin
not only as lingua franca but also in this respect. In any case, it is probably
fair to say that for these reasons English today is a particularly important
language to study. At the same time it is probably equally fair to say that
students of English with a different mother tongue have an important con-
tribution to make in this enterprise because a different background of think-
ing and the experience of another language may make them see certain
aspects of the English language in a different way from native speakers and
provide insights which are equally valuable to work in applied and theo-
retical linguistics.
2 Principles of modern linguistics

2.1 Basic concepts of linguistic structuralism

2.1.1 Principles of linguistics since de Saussure

... I am more and more aware of the immense amount of work that would be
required to show the linguist what he is doing ... The utter inadequacy of
current terminology, the need to reform it and, in order to do that, to dem-
onstrate what sort of object language is, continually spoils my pleasure in
philology, though I have no dearer wish than not to be made to think about
the nature of language in general. ...15 (Ferdinand de Saussure)
What we today call modern linguistics emerged at the beginning of the
twentieth century and is generally associated with the name of the Swiss
linguist Ferdinand de Saussure, who, in a series of lectures given in Geneva
between 1907 and 1911, established important principles of language de-
scription, which were published posthumously by his pupils in the famous
Cours de linguistique générale (1916).
Of course, the study of language has a much longer history. Its origins
lie not only in Europe, but also in other parts of the world such as Arabia16
or India.17 In ancient Greece philosophers such as Aristotle or the Stoics
were concerned with the description of the nature of language. The ideas
they developed can be seen as establishing a tradition of language descrip-
tion which was continued by Latin grammarians such as Donatus (4th cen-
tury AD) and Priscian (ca. 500 AD) and has remained influential through
the Middle Ages and the Renaissance up to this day. In fact, our present
system of word classes owes a lot to the discussions of Ancient Greek phi-
losophers, particularly the ideas of Dionysius Thrax (ca. 100 BC), and ideas
similar to established concepts of modern sentence analysis can be found in

15
F. de Saussure: ‘Letter of 4 January 1894’, in ‘Lettres de F. de Saussure à
Antoine Meillet’, Cahier Ferdinand de Saussure 21 (1964: 95); quoted in
Culler (1976: 15).
16
Cf. Robins (1967: 97–99) and Brekle (1985: 68–87).
17
Cf. Robins (1967: 136–148).
 2 Principles of modern linguistics 13

the work of Apollonius Dyscolus (2nd century AD).18 A very important

source of influence, in this context, have been grammars developed for the
teaching of Latin and other languages, which were based on the Latin tradi-
tion.
During the eighteenth and especially the nineteenth century a focus of
interest in the history and comparison of languages developed – instigated
by the study of Sanskrit. The nineteenth century is the period in which his-
torical linguistics emerged as a scientific discipline, a development which is
linked with such names as that of Jacob Grimm (1785–1863).19 It is impor-
tant to realize that de Saussure’s work must be seen against this background
of an established discipline of comparative philology.20 In fact, de Saussure
himself, who had studied at Leipzig, which was then a centre of the histori-
cally oriented school of the Junggrammatiker, had published on Indo-
European and Sanskrit and taught Sanskrit and historical linguistics in Paris
and Geneva before he turned to the teaching of general linguistics.21
What is so important about de Saussure’s Cours de linguistique gé-
nérale is that it raised a few matters of linguistic principle in a very system-
atic manner and addressed the general question of the task of the linguistic
science as such. Considering it a milestone in the history of linguistics does
not mean that all work preceding de Saussure should or could be disre-
garded. In fact, many ideas that are being fruitfully exploited today have
their roots in pre-de Saussurian linguistics: the leading American linguist of
the second half of the twentieth century, Noam Chomsky, for instance, has
repeatedly related his ideas to those of the German philosopher Wilhelm
von Humboldt (1767–1835).22

18
Cf. Robins (1967: 36–39 and 56–58).
19
Cf. Robins (1967: 171). For a more detailed account of the history of linguis-
tics see Robins (1967), Helbig (1970) or Sampson (1980).
20
Cf. Culler (1976: 9; 13–15). For a discussion of de Saussure’s view of lin-
guistics after 1800 see Culler (1976: 53–79).
21
Cf. Culler (1976: 15).
22
For such a view see Robins (1967: 174): “Wilhelm von Humboldt was one of
the most profound thinkers on general linguistic questions in the nineteenth
century, and one wonders whether, if his style had been less diffuse and his
ideas more worked out and exemplified than they were, and his voluminous
works were better known and more widely read, he would not be accorded a
position comparable to that given to de Saussure as one of the founders of
modern linguistic thought.”
14 The English language and linguistics

Nevertheless, it was de Saussure who with hitherto unknown sys-

tematicity outlined principles of modern linguistics, most of which are still
widely accepted today. They concern the following aspects:

the character of the linguistic sign and the dichotomy between signifié
and signifiant (> 2.1.2)

principles of the description of language such as the distinction between
the synchronic and the diachronic study of language (> 2.1.3)

the distinction between langue and parole (> 3.1.3)

the view of language as a system which has structure and which can be
analysed in terms of the relations of the elements establishing structure
(> 2.1.4)

2.1.2 The character of the linguistic sign

One of the basic units of linguistic description identified by de Saussure is

the linguistic sign. The linguistic sign consists of two components, which
de Saussure labelled signifiant and signifié, and for which different terms
have been used: Ausdrucks- und Inhaltsseite des sprachlichen Zeichens in
German, signifier and signified, or more commonly, form and meaning in
English.23

signifiant / form /"aIl@nd/

signifié / meaning ‘island’

Lipka (32002: 55) characterizes de Saussure’s concept of the linguistic sign

as follows:
Saussure stresses repeatedly that the linguistic sign is a mental unit (“une
entité psychique à deux faces”), and does not link a thing and a name, but a
concept and a phonic image. This image is for him nothing material, physi-
cal, but a mental impression of sound. The connection of concept and image

23
Palmer (21981: 5–6) points out that the term sign is often used to refer to the
signifier in de Saussure’s sense. For an outline of de Saussure’s concept of
the linguistic sign see Lipka (32002: 55–56).
 2 Principles of modern linguistics 15

acoustique, of concept and sound picture, for Saussure constitutes the signe
linguistique, the linguistic sign. The notions “concept” and “image
acoustique” are later replaced by him by the terms signifié and signifiant,
which have since become internationally accepted technical terms ...
The three characteristics of the linguistic sign that de Saussure emphasizes
are
− the linearity of the signifiant, which means that the form of the linguis-
tic sign consists of a chain of sounds, which is uttered and perceived in
linear order,
− the conventionality of the linguistic sign, which means that signifiant
and signifié are linked by convention within a speech community and
that as a result the linguistic sign cannot be altered by an individual,
− the arbitrariness of the linguistic sign, which means that there is no
causal relationship between form and meaning in the sense that there is
a reason why a particular meaning should be expressed by a particular
form in a language, i.e. why, for instance, the concept of ‘island’ should
be expressed as island in English, eiland in Dutch, Insel in German,
Ailön in Frisian, ø in Danish, saari in Finnish, île in French, or isola in
Italian.24
The notion of arbitrariness can be extended, as Culler (1976: 21) points out,
to the notion of signifiant as such:
It is, Saussure says, all too easy to think of language as a set of names and
to make the biblical story of Adam naming the beasts an account of the very
nature of language. If one says that the concept ‘dog’ is rendered or ex-
pressed by dog in English, chien in French and Hund in German, one im-
plies that each language has an arbitrary name for a concept which exists
prior to and independently of any language.
If language were simply a nomenclature for a set of universal concepts,
it would be easy to translate from one language to another. One would sim-
ply replace the French name for a concept with the English name. If lan-
guage were like this the task of learning a new language would also be
much easier than it is. But anyone who has attempted either of these tasks
has acquired, alas, a vast amount of direct proof that languages are not no-

24
Onomatopoetic words like cuckoo, which in a way are motivated by the thing
they refer to, are in de Saussure’s view also arbitrary in that they differ to a
certain extent in different languages and are subject to linguistic change. For
a conflicting view see Lipka (32002: 56). For a discussion of iconicity see
Lyons (1977: 99–109) and Ungerer and Schmid (22006: 300–312).
16 The English language and linguistics

menclatures, that the concepts or signifieds of one language may radically

differ from those of another. The French ‘aimer’ does not go directly into
English; one must choose between ‘to like’ and ‘to love’. ... Each language
articulates or organizes the world differently. Languages do not simply
name existing categories, they articulate their own.
Arbitrariness thus can be interpreted to mean not only that the relationship
between form and meaning is not motivated, especially not motivated by
any features of the objects of the real world which are covered by the
meaning of a linguistic sign, but also that the way in which different lan-
guages divide up the spectrum of reality in terms of meanings differs from
one language to another.

2.1.3 Synchronic and diachronic study of language

Another important distinction that was established by de Saussure which is

still generally accepted in linguistics is that between the study of the system
of a language at a given point in time, which de Saussure termed syn-
chronic, and the study of the changes that can be observed in the course of
the history of a language, which he called diachronic.
The importance attributed to the distinction must be seen against the
background of a kind of historical linguistics prevailing at the time, which
did not investigate language as a system at a particular point in time but
concentrated on describing developments. It is important to realize that
synchronic linguistics must not be equalled with the analysis of present-day
language since a synchronic description of language can be one of a his-
torical state of the language. The distinction between synchronic and dia-
chronic is primarily a methodological distinction in that any diachronic
study presupposes a synchronic study.25

25
A synchronic study of Old English is provided, for instance, by Quirk and
Wrenn (21957); a history of the English language that is very systematically
built upon the distinction between synchronic and diachronic is that by Strang
(1970).
 2 Principles of modern linguistics 17

Of course, one must realize that the idea of a synchronic study of a lan-
guage at a given point in time is to some extent a methodological abstrac-
tion since language is always in transition and never presents a stable sys-
tem as such. Nevertheless the distinction between synchronic and dia-
chronic must not be understood as denying the essentially historical nature
of language.26

2.1.4 The importance of relations

2.1.4.1 The value of the linguistic sign

One reason why de Saussure attributed so much importance to the distinc-

tion between synchronic and diachronic is that he wanted to emphasize that
at any point in time the actual value of a linguistic sign is independent from
its history and solely determined by its relationship to the other linguistic
signs of the system at that point in time. This is no contradiction to saying
26
Cf. Culler (1976: 36).
18 The English language and linguistics

that a particular sign and its place in the system of language as such are the
result of a historical process.
Thus, for the purposes of describing the synchronic state of a language,
“diachronic information is irrelevant” (Culler 1976: 36). Culler illustrates
this by the example of the English second person pronoun you, which trans-
lates as three different forms into German, for example: du, ihr and Sie.
Present-day English you is historically plural, then came to be also used as
a respectful way of addressing one person until the singular form thou died
out completely. While this is interesting in itself, it does not help you un-
derstand how present-day English you is used.
This kind of argument is crucial to a structuralist approach to the
description of language, in which language is seen as a system of linguistic
signs, which gain their identity or value not by some intrinsic property they
possess but through their mutual relationships with each other. De Saussure
(1916, English translation by Roy Harris 1983: 115) describes this as fol-
lows:
… The Slavic languages regularly distinguish two verbal aspects: the per-
fective aspect represents an action as a whole, as a single point, taking no
development into account, whereas the imperfect aspect represents the same
action in the process of development, taking place in time. These categories
are difficult for a Frenchman, because his language does not recognise
them. If they were predetermined categories, there would be no such diffi-
culty. In all these cases what we find, instead of ideas given in advance, are
values emanating from a linguistic system.
Thus it is differences and relations that matter, not substance. By the way,
this irrelevance of substance also reflects the view that the linguistic sign is
arbitrary (because if it were not, it would have substance) and, similarly,
why the linguistic sign can be subject to diachronic change.
De Saussure’s most famous analogy to illustrate this view is that with a
game of chess, where the precise shape of the pieces in a particular game,
their material, their size etc., i.e. all the factors contributing to substance, do
not matter as long as the functional value of the different pieces – which
can be recognized by relating them to each other – are clear. In the same
way, to use another analogy,27 when one identifies a train as, say, the 16.50
from Paddington, one does not mean that the same train in terms of sub-
stance (in the sense of consisting of the same engine, carriages etc.) leaves
Paddington every day at that time, but one refers to a unit that has a place

27
Cf. Culler (1976: 27).
 2 Principles of modern linguistics 19

or value in the system of all trains (so that it is still the 16.50 from Padding-
ton even if it has left Paddington late).
In a slightly different way, the example of colour terms is often men-
tioned to illustrate that linguistic signs do not have meaning as such but that
their meaning only exists in terms of their relationship to other linguistic
signs. The argument runs that a term such as blue does not mean or signify
anything as such, but that its meaning or value only becomes clear if the
other colour terms in the system are also considered and a term such as blue
is seen in relation to the other colour terms of the language.28

2.1.4.2 Syntagmatic and paradigmatic relationships

The crucial idea of a structuralist approach to language is to see linguistic

signs as determined by the relations they enter with other linguistic signs.
De Saussure identifies two types of such relationships: a sign enters
− a syntagmatic relationship with signs with which it can occur in a
sentence (indicated by the horizontal arrow below), and
− a paradigmatic relationship with signs which could occur in the
same place in the sentence (symbolized by the vertical arrow).

This boat left at 5 p.m.

A ship sank yesterday
The ferry arrived last year

Such relations can be established not only between linguistic signs such as
words or grammatical elements, but also at the level of sound:

b i; t
p & d

28
Compare the approach of prototype semantics described in Chapter 15. This
has been questioned by psychologists in recent works.
20 The English language and linguistics

Linguistic signs can thus be described in terms of the syntagmatic and

paradigmatic relations they enter.

2.1.5 Schools of structuralism

The principles of structuralist linguistics were shared by various schools of

linguistics, which emerged in different places of the world during the first
half of the twentieth century, which can all be subsumed under the heading
of structuralism:
− American structuralism, also called distributionalism, which
focussed on formal properties of linguistic elements, especially their dis-
tribution (associated with linguists such as Leonard Bloomfield, Charles
Hockett, Zellig Harris), which provided an important basis for genera-
tive grammar, the kind of linguistics which was instigated by Noam
Chomsky in the second half of the century and which has dominated
linguistic discussion ever since, especially in the English-speaking
world
− the Prague School, also known as functionalism, which concen-
trated on the function of linguistic elements, particularly in the sound
system of a language and in the way that sentences are organized (de-
veloping a theory called functional sentence perspective) (associ-
ated with linguists like Trubetzkoy, Jakobson, Mathesius, Daneš)
− the Copenhagen School, also known as glossematics, centering on
the character of the linguistic sign (founded by the Danish linguist
Hjelmslev)
− dependency grammar (developed by the French linguist Tesnière),
which concentrates on the relationship of elements within a sentence,
taking the verb as its centre, which found its most important reflection
in the development of valency theory, a model of syntax widely used
in Germany (> Chapter 17).

2.2 Linguistics and descriptivity

Although there are many differences between various modern schools of

linguistics in how they approach the subject of language and which aspects
of language they focus on, there is unanimous agreement with respect to
 2 Principles of modern linguistics 21

one principle, namely the descriptive nature of linguistic research. It is

generally understood that the task of linguists is
− to describe how people speak and
− not to tell them how to speak.
This has not always been so. In fact, in the past, it was quite common for
grammarians to take a normative or prescriptive approach by formulating
rules which did not reflect the way that the language was actually spoken
but stated how, for one reason or another, it ought to be spoken.
This means that modern linguists would reject any kinds of statement
that would imply value judgments about the use of particular forms. Such
discussions could arise with respect to utterances like the following:29
(1a)QE Whom did you see?
(1b)QE Who did you see?
(2a)QE It is I.
(2b)QE It is me.
(3a)QE I didn’t see anybody.
(3b) I didn’t see nobody.
(4a)BNC Students could be given the opportunity of actively searching for infor-
mation about some topic in which they are interested.
(4b)BNC That’s the only thing anybody seems to be interested in.
Normative grammarians and usage guides have used a number of different
arguments to justify judgements saying that one of the forms given is to be
preferred over the other and why this second form could or should be re-
garded as incorrect. Thus in the case of (1) whom is to be preferred because
it is the historically older form (or because it is the objective case), (2a) is
seen as preferable because it corresponds to the Latin rule of case assign-
ment, (3b) is rejected because it is “illogical” since two negatives can be
seen as cancelling each other out. However, from the point of view of de-
scriptive linguistics the only criterion to go by is whether a particular form
is used by the speakers of the variety we are describing. There is no reason

29
For an intensive discssion of these problems see Palmer (1971: 13–26), where
examples (1a) – (3a) are taken from.
22 The English language and linguistics

whatsoever why English should follow the rules of Latin grammar.30 Fur-
thermore, language use certainly is not logical – otherwise the response to a
question such as
(5a)BNC Can you tell me what time it is?
should be yes or no, and not
(5b)BNC It’s nine-thirty.
which is perfectly natural, however. Nevertheless, it has got to be said that
multiple negation is not used by speakers of Standard English today, al-
though it was a feature of earlier stages of the language. For this reason,
one would not consider (3b) to be a grammatical sentence of Standard Eng-
lish, although, of course, it is acceptable in many dialects of English.
Example (4b) could be objected to on the basis of a normative rule that
prepositions should never occur at the end of a sentence. Here, Palmer
(1971: 18) quotes the famous anecdote about Winston Churchill, who, after
one of his secretaries had altered a sentence to avoid it ending with a prepo-
sition, commented: “‘This is the kind of pedantry up with which I will not
put’”.
Within a descriptive approach to the analysis of language, the only rele-
vant criterion for calling a form acceptable is whether a form is established,
i.e. used by a certain number of speakers in the variety of the language one
is describing.

2.3 The principles of structuralism and foreign language

teaching

While it is not necessarily to be expected that insights gained in the scien-

tific or academic exploration of a subject are directly applicable for practi-
cal purposes, linguistics and foreign language teaching have always shown
great affinity. On the one hand, theoretical and descriptive linguistics can
gain important insights from investigating aspects of language (or of a spe-

30
See Palmer (1971: 20–21), who points out that double negatives are common
in other languages such as Russian or Spanish and continues: “It was the
same in classical Greek but not in Latin. This should be hardly surprising; if
Latin had had double negatives they would have found favour, not disfavour,
with English grammarians! There can, then, be no logical reason for exclu-
ding double negatives. No rules are broken by I didn’t see nobody.”
 2 Principles of modern linguistics 23

cific language) that become particularly obvious when one deals with a
foreign language or problems of teaching a foreign language. In particular,
many important grammars and dictionaries of English, which were written
with the foreign language perspective in mind, have turned out to be valu-
able contributions to the description of the English language as such. On
the other hand, developments in the scientific investigation of language can
indeed be applied to improving teaching materials and sometimes even
teaching methodology. As far as structural linguistics is concerned, two
important consequences can be drawn with respect to analyses which are
relevant to foreign language teaching:
− Firstly, the dichotomy of form and meaning makes it essential to keep
these two levels apart in any analysis of language.
− Secondly, the recognition of great differences between languages – both
with respect to the levels of form and meaning – shows the necessity
that a language can only be described appropriately in terms of catego-
ries that have evolved from the analysis of that language and not by us-
ing concepts or terminology that has arisen from a different language.
It is important to note that both criteria have often been violated in the
teaching of foreign languages. For a very long time foreign language teach-
ing was carried out in the framework of so-called traditional grammar,
which often used semantic criteria for the definition of formal categories
such as word classes (> 11.4), for example.31
In the description of English one such problem arises in the identifica-
tion of tenses. First of all, it is important to make a distinction between time
and tense. Time is a semantic category because it relates to meaning – time
being either ‘past time’, ‘present time’ or ‘future time’. Tense, however, is
a formal category relating to particular verb forms.32
As far as the tense system of English is concerned, the question is
whether any kind of verbal construction that relates to time should be con-
sidered a tense or whether the term tense should be restricted to forms
which can be distinguished in terms of endings (or suffixes, > 8.2) and not

31
Cf. Burgschmidt and Götz (1974: 27–30). For the weaknesses of traditional
classification of word classes see 11.4 and Herbst and Schüller (2008: 12–14
and 31–35).
32
Note that even terms such as past tense violate the principle of keeping the
levels of form and meaning strictly apart because the label ‘past’ is a seman-
tic label indicating a very frequent but by no means the only meaning that can
be expressed by this form.
24 The English language and linguistics

applied to constructions involving auxiliary verbs, for instance. In this latter

view, which is taken in the Comprehensive Grammar of the English Lan-
guage by Quirk, Greenbaum, Leech, and Svartvik (CGEL 1985: 4.3), for
example, English has only two tenses – a present tense and a past tense
(walks – walked).33
The most important point in this respect, however, is the fact that it
would be very difficult to make out a case for a category future tense in
English. Both great reference grammars of English – CGEL (1985: 4.3) and
the Cambridge Grammar of the English Language (CamG 2002: 209–210)
– reject the use of this term for the description of English. First of all, there
is no inflected future tense in English as in Latin (amabo). Secondly, there
are a number of forms that can be used to refer to future time – will + in-
finitive, be going to, be about to, present progressive, present simple, etc.
Thirdly, it is not really possible to draw a dividing line between temporal
uses of the will-construction and modal uses of the will-construction.34
A similar case is presented by the use of the term gerund with reference
to present-day English, for which there is no justification at all. While in
Latin gerunds (laudandi) are distinguished from present participles (lau-
dans) by different endings, there is no such formal distinction in English
(praising – praising). For this reason, the standard reference grammars do
not make a terminological distinction either: CGEL (1985: 17.54) merely
uses the term participle and CamG (2002: 80–83) introduces the term ger-
und-participle. In fact, the gerund is a good example of how the teaching of
English may be made unnecessarily complicated by the use of inappropri-
ate terminology. If students are told in a pedagogic grammar that “gerund-
constructions can only function as adverbials when combined with a prepo-
sition”35, they may well be led to assume that sentences of the type
(6)BNC Knowing her children, the teacher had decided that this activity was
suitable for them.

33
CGEL (4.17) classifies forms such as have painted or had painted as realiza-
tions of a perfective aspect, whereas CamG (2002: 15) speaks of secondary
tenses.
34
For instance, it is difficult to see why I’ll drop by sometime and You will
come back, won’t you? should be given as examples of future tense and We
will stay here and I will call you when I’m ready as examples of a modal verb
will, as in the Cobuild English Grammar (1990: 4.195 and 5.53/60).
35
“Gerund-Konstruktionen können nur in Verbindung mit Präpositionen ad-
verbiale Funktion haben” (Englische Grammatik heute 1999: 199).
 2 Principles of modern linguistics 25

are ungrammatical in English. Quite obviously they are not – only that this
particular grammar would classify knowing in (6) as a participle and not as
a gerund.
This example shows that traditional terminology taken over from the
teaching of Latin can be quite detrimental when applied to English (and
may put many students off grammar).36 The fact that many linguists use
the term gerund without finding it necessary to justify this in the light of
the arguments that such a category is not appropriate to the description of
present-day English is at least worrying and shows that de Saussure’s
qualms about “the inadequacies of current terminology” have by no means
been solved a hundred years later.

2.4 Areas of investigation

The structuralist way of seeing language and of approaching its analysis

has been so influential in establishing many of the categories serving for
the analysis of the sound systems of languages (phonetics and phonol-
ogy), the analysis of the structure of words (morphology and word for-
mation) and sentences (syntax), and also the analysis of word and sen-
tence meanings (semantics) that some of the chapters that follow will to a
large part be based on the concepts and ideas developed by linguists work-
ing within a structuralist framework.
Apart from analysing these structural aspects of language, several sub-
disciplines of linguistics concentrate on different aspects of how language
is used by speakers. For example, they are concerned with
− units larger than the sentence (text linguistics, discourse analysis)
− the social importance of language use (sociolinguistics)
− the analysis of language as human behaviour (pragmatics, speech act
philosophy)

36
For questions of terminology and foreign language teaching see Schröder
(2005) and Herbst (2005). Compare Palmer’s (1971: 15) comments on
mother tongue teaching: “Since most English grammar teaching was based
upon Latin the students were often at a loss. They could not see why English
had a subjunctive or a dative case, but when they learnt Latin it all became
clear. Latin helped them with their English grammar, but only because Eng-
lish grammar was Latin grammar all the time.”
26 The English language and linguistics

− psychological aspects of language use, language acquisition, language

production and perception (psycholinguistics and cognitive linguis-
tics).
The later chapters of this book will introduce some aspects of these topics.
3 Language, intuition and corpora

3.1 Language

3.1.1 Some basic distinctions

Since modern linguistics is a descriptive discipline, one of its main tasks is

to describe language (or languages) as used by native speakers. This leads
to two questions:
− How do we know or find out how a language is used?
− What exactly is it that we are describing or could be aiming at describ-
ing?
The question of data will be addressed in sections 3.2 and 3.3; with respect
to the second question one important aspect of the problem is whether the
research carried out is primarily aiming at
− finding out about human language as such, addressing such questions as
to whether there is an inborn language faculty in the mind, whether
there are general principles operating in all languages (so-called univer-
sals) etc. or
− describing a particular language to show its properties, state its rules of
grammar, describe its vocabulary etc., as one would, for instance, when
one is writing a grammar or a dictionary of English.
This is linked, up to a point at least, with the question of whether one is
aiming at a description of
− the language of an individual, which seems appropriate when one inves-
tigates questions of language and the mind,
− a language (or a variety of a language) as such, as used by a speech
community, i.e. language as a social phenomenon.
That these are not identical is fairly obvious: no one would expect any in-
dividual speaker to know all the words of a language. Thus one can safely
assume that most speakers of German will not be familiar with all of the
following words: Duckdalbe, Menhir, Menkenke, Hyperurbanismus, Pfahl-
bürger, Pfahlmuschel – all to be found in the Duden Universalwörterbuch
4
2001). While it would hardly make sense to base the selection of items to
28 The English language and linguistics

be included in a general dictionary on the words known to one person, find-

ing out which words a person knows and uses at a particular age can be
highly relevant when exploring language acquisition, for example.
Thus the scope of what a descriptive linguist might aim to describe is
relatively wide. A number of important distinctions have been made in
linguistics which may help to clarify (or at least illustrate) the problem,
some of which will be mentioned in the following sections.

3.1.2 Competence and performance: the language of the individual

One such distinction is that between competence and performance, which

was introduced by the American linguist Noam Chomsky, who, in the late
1950s, instigated a theoretical framework known as generative grammar,
which was to dominate linguistic theory for the second half of the twentieth
century (> 13.1). In his book Aspects of the Theory of Syntax Chomsky
(1965: 4) writes:
We ... make a fundamental distinction between competence (the speaker-
hearer’s knowledge of his language) and performance (the actual use of
language in concrete situations). ... A record of natural speech will show
numerous false starts, deviations from rules, changes of plan in mid-course,
and so on.
Elsewhere, Chomsky (1966: 9–10) describes the difference by saying that
competence is “what a speaker of a language knows implicitly” and per-
formance “what he does”. It is important to emphasize the word implicitly,
because native speakers very rarely are able to state rules of the grammar of
their language (unless they have received some special training as teachers,
for example), but they know the rules of the language in the sense that they
apply them correctly.
The difference between competence and performance can be illustrated
with reference to errors that people make when learning a foreign language.
Performance errors are the kinds of errors that learners make but which
they either put right immediately or which they would correct when look-
ing at the text again. Competence errors cannot be corrected by the learners
themselves, of course.
 3 Language, intuition and corpora 29

3.1.3 Language as a social phenomenon

One problem for the description of language is that what can be observed
directly is the performance of a speaker or indeed the performance of many
speakers/writers of a speech community but not their competence. How-
ever, it would be inappropriate to describe a language simply in terms of
performance, i.e. in terms of the utterances its speakers produce. Rather,
many linguists believe that it is possible to construct an underlying “system
of rules and relations” (Lyons 1968: 52) on the basis of these utterances. De
Saussure (1916) introduced the notions of langue and parole to capture this
difference, which Lyons (1968: 51) describes by saying “that all those who
‘speak English’ (or are ‘speakers of English’) share a particular langue and
that the set of utterances which they produce when they are ‘speaking Eng-
lish’ constitute instances of parole”.37
De Saussure (1916, English translation by Roy Harris 1983: 13) charac-
terizes langue as follows:38
The individual’s receptive and co-ordinating faculties build up a stock of
imprints which turn out to be for all practical purposes the same as the next
person’s. How must we envisage this social product, so that the language it-
self can be seen to be clearly distinct from the rest? If we should collect the
totality of word patterns stored in all those individuals, we should have the
social bond which constitutes their language. It is a fund accumulated by
the members of the community through the practice of speech, a grammati-
cal system existing potentially in every brain, or more exactly in the brains
of a group of individuals; for the language is never complete in any single
individual, but exists perfectly only in the collectivity.

3.1.4 System and Norm – language use

Chomsky (1965: 4) draws a parallel between his own distinction of compe-

tence and performance and langue and parole, but criticizes de Saussure’s

37
Compare the formulation of Chomsky (1995: 14), where an individual’s
competence is described as “knowledge and understanding” and performance
as “what he does with that knowledge and understanding”.
38
Cf. Culler (1976: 29): “La langue is the system of a language, the language as
a system of forms, whereas parole is the actual speech, the speech acts which
are made possible by the language”.
30 The English language and linguistics

concept of langue as being merely a systematic inventory of items.39 This

reflects a rather interesting difference of views about the nature of language
as such. Chomsky (1965: 4) argues that the “problem for the linguist” or a
“child learning the language, is to determine from the data of performance
the underlying system of rules that has been mastered by the speaker-hearer
and that he puts to use in actual performance”.
This raises the question – which is one core problem of modern linguis-
tics and will be taken up in a number of chapters of this book – to what
extent language can be seen as a system of rules or not. In this respect, a
distinction introduced by the Romance scholar Coseriu (1973: 44) is of
interest – namely that between System and Norm. The Norm in Coseriu’s
sense40 is the totality of traditional realizations, which Coseriu distinguishes
from the System, which is the totality of possible realizations. The concept
of Norm is quite useful to explain why certain forms or combinations of
forms which should be possible according to the System nevertheless do not
seem to occur: thus the fact that in present-day English there is sensitivity
next to sensitiveness, while there is only kindness (and not *kindity) and
idiomaticity (and no *idiomaticness) can be seen as a typical Norm phe-
nomenon.
A similar kind of emphasis on the role of item-specific or idiomatic
elements in language can be found in so-called usage-based approaches
within the framework of cognitive grammar.41 With respect to language
acquisition, the basic assumption of such theories is, as Tomasello (2003:
5) puts it, “that language structure emerges from language use”. The idea is
that on the basis of usage events, i.e. in de Saussure’s terminology in-
stances of parole, children build up inventories of constructions and make
certain abstractions or generalizations and that “difference between young
children’s inventories and those of adults” can be seen as “one of degree”
(Lieven 2008: 64) (> 12.3.3). This means that grammar is no longer envis-

39
Cf. Chomsky (1965: 4): “A grammar of a language purports to be a descrip-
tion of the ideal speaker-hearer’s intrinsic competence”. See Sampson (1980:
49–50) for a discussion of competence vs. performance and langue vs. parole.
40
The term Norm must not be interpreted in the sense of prescriptive or norma-
tive grammar here.
41
For convention and usage see Langacker (1987: 65–66). For the importance
of idioms in construction grammar cf. Croft and Cruse (2004). For item-
specific knowledge and generalizations in usage-based models see Goldberg
(2006: 63–65).
 3 Language, intuition and corpora 31

aged as “a fixed synchronic system”, as Bybee and Hopper (2001: 3) point

out:
The notion of language as a monolithic system has had to give way to that
of a language as a massive collection of heterogeneous constructions, each
with affinities to different contexts and in constant structural adaptation to
usage.
If we look upon language acquisition and the resulting linguistic compe-
tence as being usage-based,42 then this must allow for differences in the
emerging linguistic “systems” – which, up to a point, seems compatible
with de Saussure’s characterization of language in 3.1.3.43

3.2 Finding data: traditional methods

3.2.1 Principal options

One of the principal problems of descriptive linguistics is the question of

how one can find out which forms occur or can occur within a language.
Basically, there are three methods that can be used:
− relying on one’s own intuition
− carrying out tests with a sufficiently large number of native speakers
− analysing authentic language material
or combining various methods of that kind.44

42
See, for instance, Goldberg (2006: 227): “Speakers’ knowledge of a language
consists of systematic collections of form-function pairings that are learned
on the basis of the language they hear around them”.
43
It is less compatible with Chomsky’s concept of competence, which is devel-
oped with reference to the construct of the ideal speaker-hearer “in a com-
pletely homogeneous speech community” (Chomsky 1965: 3). This is, of
course, a considerable abstraction, which has been criticised from a number
of different perspectives. For a discussion of the distinction between compe-
tence and performance with respect to language variation see e.g. Dufter,
Fleischer and Seiler (2009: 6–9). Compare also Chomsky’s (1986: 20–22)
discussion of I-language and E-language in this context.
44
For a discussion of the use of attested data in different approaches see Stubbs
(1996: 28–32). Compare also Mair (1997: 9–13). For the advantages and dis-
advantages of these methods in lexicography see Herbst and Klotz (2003:
267–280).
32 The English language and linguistics

3.2.2 Introspection and elicitation

Many grammars and dictionaries are based on the linguistic intuitions of

their authors, i.e. their own competence in that language. This involves
three main dangers:
− first of all, the competence of the individual in question may not be rep-
resentative of the speech community as a whole,
− secondly, linguists may be guided in their acceptability judgements by
the kind of analysis they are carrying out and the kinds of claims they
wish to make,45
− thirdly, it has been shown that acceptability judgements vary to a con-
siderable extent.
These problems can be reduced by basing one’s findings not on the intui-
tion of a single person. This has been done, for example, in the case of the
Grammar of Contemporary English, a large grammar published by
Randolph Quirk, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik in
1972. For the purposes of writing that grammar, elicitation experiments
were carried out in order to clarify certain points of English grammar. The
tests used involved so-called judgement tests, in which subjects were asked
to rate given forms as acceptable or not etc., and so-called performance
tests, in which subjects were asked to change sentences in a particular way.
In order to find out whether a phrase such as none of the children is used
with a plural or a singular verb, for instance, subjects were given a past
tense sentence such as None of the children answered the question and
required to put it into the present tense (which forces a decision as to plural
or singular verb).46 Performance tests have two advantages: they test actual
behaviour (sometimes, as in this example, without subjects being aware of
what is being tested) and – unlike judgement tests – they do not depend on
conscious reflection about language, which is often unreliable. Still, even
tests of this kind do not always lead to clear results: experience shows that
it is not uncommon for the same subjects to rate the same sentences differ-
ently when the test is repeated. Furthermore, results can depend on the de-
sign of the tests: Greenbaum and Quirk (1970: 6) report that in the case of
None of the children “a singular verb (prescribed by schoolroom precept)

45
For the problems of intuition-based linguistics see Heringer, Strecker, and
Wimmer (1980: 63).
46
Cf. Greenbaum and Quirk (1970: 23–24).
 3 Language, intuition and corpora 33

was preferred in a judgment test more frequently than it was preferred in a

selection test”, which shows that people’s attitude towards the acceptability
of a form does not always coincide with their actual behaviour.

3.2.3 Authentic language material: citations and corpora

The alternative to relying entirely on introspection and native speaker in-

formants is to draw upon authentic language material. In fact, there is a
long tradition for dictionaries and grammars to make use of authentic ex-
amples, often drawn from literary sources.
A remarkable attempt to base linguistic description on authentic cita-
tions was the Appeal to the English-Speaking and English-Reading Public
launched in 1879 by Murray in the process of compilation of the famous
Oxford English Dictionary, which up to this day is the most important his-
torical dictionary of English. By the time the first edition appeared, 5 mil-
lion citations had been collected by over 1300 voluntary readers, 1.8 mil-
lion of these citations were actually used in the dictionary.47
While the method of employing citations is based on language use in
that it provides evidence for particular instances of language in use, it does
not allow any kind of statistical analyses, for example. A much more com-
prehensive way of investigating language in use consists in the analysis of
large collections of texts, so-called corpora, which will be addressed in the
next section.48

3.3 Corpus linguistics

3.3.1 Corpora of English

Not many people outside linguistics are aware of the fact that the basis for
the empirical study of English – and increasingly so, also of other lan-
guages – has changed dramatically over the last few decades. This is appar-

47
Cf. Mugglestone (2009: 243–244) and Herbst and Klotz (2003: 271–273).
48
For the difference between “citation of instances” and “concordancing of
texts” see Sinclair (1991: 39–41). For the use of the OED as a corpus-like re-
source see Hoffmann (2004). Compare also Mukherjee (2009: 13 and 125–
128).
34 The English language and linguistics

ent from the introduction of a book by John Sinclair (1991: 1), where he
talks about “the emergence of a new view of language” and says:
Over the last ten years, computers have been through several generations,
and the analysis of language has developed out of all recognition.
The big difference has been the availability of data. The tradition of lin-
guistics has been limited to what a single individual could experience and
remember. Instrumentation was confined to the boffin end of phonetics re-
search, and there was virtually no indirect observation or measurement. The
situation was similar to that of the physical sciences some 250 years ago. ...
Thirty years ago when this research started it was considered impossible
to process texts of several million words in length. Twenty years ago it was
considered marginally possible but lunatic. Ten years ago it was considered
quite possible but still lunatic. Today it is very popular.
The main reason for this is the enormous developments in computer tech-
nology, which have resulted in possibilities of accessing, within seconds,
data from linguistic databases of unprecedented size. These technological
innovations have given huge impetus to the development of a corpus lin-
guistics – a methodology49 (or a subdiscipline of linguistics) that systemati-
cally analyses large text corpora and derives its results from the analysis of
such a corpus.
Although the method of basing linguistic analysis on corpora is also
characteristic of the approach taken by American structuralists in the first
half of the previous century,50 the beginnings of what we now refer to as
corpus linguistics can probably be more appropriately seen in the late
1950s or early 1960s. This is the time when Randolph Quirk founded the
Survey of English Usage, a corpus for grammatical analysis, which served
as an important source of information for the 1972 Grammar of Contempo-
rary English. At about the same time, the first computer corpus of Ameri-
can English was set up by Francis and Kucera at Brown University, and a
parallel corpus for British English – the Lancaster-Oslo/Bergen (LOB)

49
See Leech (1992: 105). Cf. also Aarts (2000: 7) for the relation between cor-
pus linguistics and theoretical linguistics. See also Leech, Hundt, Mair, and
Smith. (2009: 24).
50
Cf. Matthews (1993: 131–133). Compare also Leech (1991: 8): “For such
linguists the corpus – a sufficiently large body of naturally occurring data in
the language to be investigated – was both necessary and sufficient for the
task in hand, and intuitive evidence was a poor second, sometimes rejected
altogether. But there is virtually a discontinuity between the corpus linguists
of that era and the later variety of corpus linguists ...”.
 3 Language, intuition and corpora 35

Corpus – was developed in the seventies.51 That such important steps in

corpus linguistics should have happened at this time is remarkable insofar
as linguistic research, especially in the area of syntax, was dominated by
the development of generative grammar with its emphasis on intuition as it
had been formulated by Noam Chomsky.
In the early days of corpus linguistics corpus size was a major problem
– and, in fact, it remains difficult to determine how large a corpus must be
in order to be appropriate for a particular research task.52 However, the
extent of the problem has been reduced considerably by developments in
computer technology. While the Brown and the LOB corpora contained 1
million words, the Cobuild Corpus established by John Sinclair at Bir-
mingham, which can claim to be the first corpus to have been systemati-
cally exploited for the purposes of writing a general English dictionary, in
1987, when the Cobuild Dictionary of the English Language first appeared,
already contained 7 million words. The size of corpora dramatically in-
creased during the 1990s; Cobuild, which was then renamed the Bank of
English, at one point reached up to 500 million words running text. An-
other equally important corpus available to us today is the British National
Corpus, containing 100 million words, with a 10 per cent component of
spoken English. Although the possibility that a particular form or construc-
tion does not occur in the corpus cannot be totally ruled out today, the
problem quite obviously presents itself in a completely different perspec-
tive than 40 years ago.
Increasingly, corpora are being designed with the purpose of comparing
different historical stages of English (such as the Helsinki Corpus of Eng-
lish Texts: Diachronic Part) or different varieties of English such as the In-

51
For a short survey of English corpus linguistics see Garside, Leech, and
Sampson (1987), Leech (1991) and Aijmer and Altenberg (1991: appendix)
and Mukherjee (2009).
52
Referring to a study of the negation of the verbs need and dare using the
Brown Corpus, which found only 32 instances of need negated by not and
only 11 of dare, Greenbaum (1988: 84) points at a general problem of corpus
analysis: “If we are looking at syntactic data, it may be a matter of chance
that a particular syntactic feature is absent or rare in our corpus. Only for very
common constructions can we be certain of finding adequate evidence. We
cannot know that our sampling is sufficiently large or sufficiently representa-
tive to be confident that the absence or rarity of a feature is significant.”
36 The English language and linguistics

Name Editors Content

Survey of English Us- R. Quirk and S. 1 m British English (50%
age Greenbaum written – 50% spoken)
(University College, from 1953–1987
London)
Brown Corpus W. N. Francis and H. 1 m American English
Kucera from 1961
(Brown University,
Providence)
Lancaster-Oslo/Bergen G. Leech, S. Johansson 1 m British English (to
Corpus (LOB) and K. Hofland (Lancas- match Brown) from 1961
ter, Oslo, Bergen)
Bank of English/ J. Sinclair in 1987: 7 million; later
Cobuild Corpus (University of Birming- over 500 m British and
ham) American English, spoken
and written
British National Cor- BNC consortium53 100 m words of British
pus English (90% written –
10% spoken)
Helsinki Corpus of M. Rissanen, O. Ihalainen diachronic corpus: 1.6 m
English Texts: Dia- and M. Kytö (University British English from 850–
chronic and Dialectal of Helsinki) 1720
International Corpus project initiated by S. ICE-GB, ICE-India etc.
of English Greenbaum 1 million (60% spoken)
(University College Lon- texts from 1990–1993
don) each

Some important corpora of English

ternational Corpus of English, which is to comprise 25 subcorpora of L1-

varieties such as British English, American English, New Zealand English
and second language varieties such as East African, Indian or Singapore
English.54 Furthermore, FROWN and F-LOB – the Freiburg Brown and the

53
The BNC consortium includes: Oxford University Press, Addison-Wesley
Longman, Larousse/Kingfisher Chambers, Oxford University Computing
Services, University Centre for Computer Corpus Research on Language at
Lancaster University, and the British Library’s Research and Innovation Cen-
tre.
54
For details of ICE corpus design see Mukherjee (2009: 48–50). For Brown,
LOB, Frown and F-LOB see Leech et al. (2009: 24–31).
 3 Language, intuition and corpora 37

Freiburg LOB Corpora – were designed as parallel corpora to Brown and

LOB, containing texts from 1991–1992, to be paralleled by corresponding
corpora with texts from 1901 and 1931 which provide a basis for research
on change in English in the twentieth century.

3.3.2 What we can do with corpora

3.3.2.1 Corpus analysis

Corpora have become an indispensable tool for a considerable amount of

research in the fields of syntax, phraseology and lexis. When Sinclair
(1991: 1) speaks of a new view of language, then this is because corpus
research has indeed shown the importance of phraseological units and pre-
fabricated items in language use (see Chapter 12). It is not by chance that
the development of corpora such as COBUILD or the BNC should have
been supported by the major EFL dictionary publishers. All recent editions
of English learners’ dictionaries are based on extensive corpus research and
can thus provide much better descriptions of the language than earlier edi-
tions. Likewise, modern grammars are based on the insights of corpus re-
search – notably the Longman Grammar of Spoken and Written English by
Biber, Johansson, Leech, Conrad and Finegan (1999).
Corpora provide interesting insights into the use of words or construc-
tions. They can be used to gain information on:

the frequency of words or particular constructions (in the spoken or
written language or in particular types of text),

whether a particular form typically occurs in particular types of texts
(such as spoken texts, written texts, literary language, newspaper lan-
guage),

whether a particular form is typically used by particular groups of
speakers (depending on factors such as age, region, sex),

which other words typically occur together with a particular word (by
providing collocational profiles showing the statistical significance of
particular combinations),

the syntactic constructions and valency patterns in which a word occurs.
For more detailed analyses of the uses of a word, the most convenient start-
ing point is a so-called KWIC-concordance (keyword in context):
38 The English language and linguistics

Concordance from the BNCweb interface (British National Corpus)55

55
Cf. Hoffmann et al. (2008).
 3 Language, intuition and corpora 39

Thus, with the software currently available, it is relatively easy to find out
that in the BNC (and thus, one would hope, in British English) I suppose is
much more frequent than I assume (69.49 pmw versus 3.17 pmw (per mil-
lion words)) or that guilty conscience (0.4 pmw) is more frequent than bad
conscience (0.06 pmw).
Similarly, checking a problem such as whether none is used with a plu-
ral or a singular verb in a corpus such as the BNC is quicker and probably
more reliable than carrying out judgment or performance tests of the type
described above. Interestingly, the results show that when none is immedi-
ately followed by a verb, singular forms tend to occur almost twice as often
as plural forms, but after phrases of the type none of the + plural noun plu-
ral verb forms are to be found more often (suggesting that the occurrence of
56
a plural form in the noun may prime the use of a plural form in the verb).
It is no exaggeration to say that many rules of grammar found in older
grammar books or dictionaries had to be thrown overboard or modified in
the light of the new evidence. For instance, it was often claimed that the
difference between try to open the window and try opening the window is
that the former would be used in a situation where for some reason it may
be difficult to open a particular window (because it got stuck etc.) and the
latter in a situation in which a window should be opened to achieve some-
thing else (such as getting fresh air into a seminar room). Corpus research
shows beyond doubt, however, that this kind of rule is inadequate in that
try doing can be used equally for both types of situation, as is shown by the
concordances above.
These examples show to what extent corpus research can contribute to
an accurate description of the way a language is used.57

56
Such figures have to be treated with a certain amount of caution, however.
What was calculated here was none and none of the _NN2 [collocational pro-
file + verb span R1R1 frqn (node, collocate at least 1) frqn (collocate at least
1)], but obviously more complex structures need also be taken into account.
57
Corpora play an important role in many other areas of research as well; for
first language acquisition see e.g. Behrens (2008); for a corpus-based study of
gender-specific elements of language use see Grimm (2008).
40 The English language and linguistics

3.3.2.2 Corpora and foreign language teaching

Corpus linguistics has of course had an immensely important impact on the

quality of teaching materials such as textbooks, grammars and dictionaries.
In advanced language classes, having direct access to a large corpus of the
foreign language is a good way for the teacher to provide authentic infor-
mation on language use, especially in areas such as collocation and phrase-
ology (> 10.5).58
Furthermore, corpora can be used in research on foreign language learn-
ing. In this context the International Corpus of Learner English (ICLE)
designed by Sylviane Granger, which contains subcorpora of the English
used by university students with different L1-backgrounds and provides an
ideal basis for finding out differences between the use of English by learn-
ers and native speakers.59 This does not only concern the identification of
errors and investigation of L1 influence, but the comparison of learner cor-
pora with native speaker corpora has also shown overuse and underuse of
particular forms to be an important characteristic of learner language: Lo-
renz (1999: 168–171), in a study on the intensification of adjectives, has
shown that German learners of English tend to overuse high-frequency all-
purpose intensifiers such as really, so and quite and to underuse particu-
larly, extremely or crucially and vitally, Gilquin (2007: 288) comes to the
conclusion that French learners underuse collocations with the verb make,
and Paquot (2009: 109) provides evidence for the overuse of for example
and for instance by learners with different mother tongue backgrounds.60 It
is obvious that this kind of research can be applied fruitfully to foreign
language teaching: thus Granger (forthcoming) argues that “the overuse of
the adjective important is a prompt for lexical expansion exercises aimed to
raise learners’ awareness of other adjectives (major, crucial, significant,
etc.) that can be used instead”.

58
For the use of corpora in the classroom see Mukherjee (2002). For the use of
corpora with respect to the design of teaching materials see Mindt (1988).
59
See, for instance, Granger (1998 and 2009) and Gilquin (2007). For a survey
of learner corpora see Nesselhauf (2004).
60
For the different types of chunks used by native speakers and advanced learn-
ers see de Cock (2000).
 3 Language, intuition and corpora 41

3.3.3 Corpus design and corpus size

Despite the tremendous possibilities that have been opened for linguistic
research by the availability of large machine-readable corpora, one has to
be aware of the problems associated with corpus linguistics, particularly
those concerning corpus size and corpus design.61
Thus the question of how large a corpus must be for which research
purpose is by no means easy to answer. One has to be aware of the fact that
there is an enormous discrepancy between the 100 million words of the
British National Corpus (and even larger databases used by dictionary pub-
lishers) and the size of 1 million words of the subcorpora of the ICE-project
and the corpora belonging to the Brown family (like Frown and Lancaster
1931).
Equally important is the question of how representative a corpus is of
the overall use of the language (Mukherjee 2009: 21), as is also pointed out
by Hausser (1999: 291–292):
The value of a corpus does not reside in the content of its texts, but in its
quality as a realistic sample of a natural language. The more representative
and balanced the set of samples, the higher the value of the corpus for, e.g.,
computing the frequency distribution of words and word forms in the lan-
guage.
A very important demand to be made on a corpus today is that it should be
balanced. This means that it should contain texts from
− different genres
− spoken and written language
− speakers of different regions, age groups, social backgrounds and gen-
ders
in an appropriate mixture.
Apart from the obvious difficulties involved in designing a balanced
corpus, there are further reasons for not taking corpus evidence as unques-
tionable truth in any way. For instance, there is the problem that corpora,
especially corpora of the spoken language, may contain errors made by
native speakers or may also contain utterances by non-native speakers.
Thus one would certainly not take the fact that one instance of succeed +
to-infinitive can be found in the BNC – as opposed to 1215 examples of

61
For principles of corpus design see also Mukherjee (2009: 21–23) and also
Hoffmann et al. (2008).
42 The English language and linguistics

succeed + in + V-ing – as evidence for the existence of a to-infinitive pat-

tern for the verb succeed.
It should just be mentioned in passing that the question of not being able
to verify whether an utterance was produced by a native speaker (and, if so,
of which variety of English) is particularly problematic when it comes to
using examples from the internet on the basis of a Google search etc. This
is a particularly serious problem in the case of English because the internet
contains many texts that were translated into English or written by non-
native speakers, which is an important argument in favour of using care-
fully designed corpora.

3.4 Introspection, corpus analysis and views of language

Although there is no immediate connection between the method used to

discover what the facts of language are and the particular view of language
held by a researcher, it is obvious that the different tools outlined above are
appropriate in different degrees for different purposes. In particular, one
must be aware of the fact that corpora only contain utterances that have
already been made, i.e. they are an account of usage. While this means that,
up to a point, they can also be used to account for the creative potential of
language, the fact that the BNC displays no instances of the verb sneeze of
the kind discussed by Goldberg (2006: 42) with respect to a certain argu-
ment within construction grammar, namely, She sneezed the foam off the
cappuccino cannot be taken to mean that such uses do not exist. As a con-
sequence, it is important to bear in mind that although corpora offer fasci-
nating possibilities they can never be the only source of linguist research,
as is underlined by Johansson (1991: 313):
In spite of the great changes in the less than three decades since the first
computer corpus, there is one way in which the role of the corpus in lin-
guistic research has not changed. The corpus remains one of the linguist’s
tools, to be used together with introspection and elicitation techniques.
Wise linguists, like experienced craftsmen, sharpen their tools and recog-
nize their appropriate uses.
Recommended further reading:

− Hoffmann (22000a) and Robins (1967)

− Palmer (1971)
− Mukherjee (2009)
 Sounds

4 The sounds of English: phonetics 62

4.1 Sounds as the starting point of linguistic analysis

One can approach the systematic description of a language either from top
to bottom or from bottom to top, i.e. to begin with either the largest or the
smallest units that can be identified in the analysis. Since it is not immedi-
ately obvious what the largest unit is that can sensibly be identified for such
a purpose – the text? the sentence? – and since smaller units are perhaps of
greater relevance to the description of larger units than vice versa, the
sounds of speech will be taken as a starting point in this chapter.
The fact that it is speech sounds, and not, as one might think, letters, that
are attributed such importance in the description of language has to do with
the fact that speech precedes the written language both ontogenetically and
phylogenetically. Just as children learn to speak before they learn to write,
writing systems can be considered a kind of secondary phenomenon in the
history of the development of languages as such. For these reasons, struc-
turalists such as de Saussure made it an important principle to base the
study of language primarily on the spoken and not on the written form.

4.2 Phones

Describing the sounds of a language presupposes knowledge of what the

sounds of a language actually are. This is by no means as obvious as it may
seem. As a rule, speakers do not produce speech sounds in isolation, but
sequences of sounds with no breaks or pauses between individual sounds,

62
The main purpose of Chapters 4 and 5 is to provide a basis for the discussion
of the phonology of English. For a more detailed account of English phonet-
ics and phonology see, for instance, Gimson (1962/41989), Arnold and Han-
sen (1975/81992), Scherer and Wollmann (1972), Dretzke (1998), Roach
(42009) or Gut (2009).
44 Sounds

or indeed words. Thus the first task of a phonetician is to identify the sound
inventory of a language. This can be done by analysing stretches of sound
chains and trying to find recurrent elements, i.e. units that appear repeat-
edly. The result of this process of segmentation, of dividing the contin-
uum of a chain of sounds into small, recurrent elements, can then be repre-
sented in the form of a phonetic transcription of the following kind:
[ðɪˈɪŋɡlɪʃˈhævnəʊrɪˈspektfəðeəˈlæŋɡwɪdʒ|əndwɪɫnɒtˈtʰiːtʃðeəˈtʃɪɫdrəntəˈspiːkɪt]

The English have no respect for their language, and will not teach their
children to speak it (G.B. Shaw: Pygmalion).
The phonetic transcription indicates that it is possible to identify segments
in a stretch of speech which are identical or at least sufficiently similar to
be regarded as instances of the same speech sound or phone. 63 Further-
more, the transcription shows that some syllables are more prominent than
others (indicated by the symbol ˈ preceding such a syllable). What cannot
be shown in this way are features of speech that extend to larger units such
as intonation. Nevertheless, the kind of segmentation procedure reflected in
the transcription can be used to establish a phone inventory of a language.
Thus, on the basis of the passage quoted above, it is possible to identify the
following sounds as belonging to the sound inventory of English:
[ð] the, their
[ɪ] the, English, English, will, it
[ŋ] English, language
[g] English, language
[l] English, language
[ɫ] will, children
[t] respect, it

63
This is a slightly idealized generalization. Cf. Garman (1990: 190): “Different
utterances of the same word or phrase by the same speaker under the same
conditions give rise to variable forms. These variables are, individually, un-
predictable; but, over repeated utterances they will tend to cluster around
common articulatory-acoustic centres. They can therefore be termed probabil-
istic variants.” Garman also points at differences due to distinct vocal tracts,
age and sex.
 4 The sounds of English: phonetics 45

[th] teach64
etc.
After segmentation, the next step is the description and classification of
the phones identified for a language.

4.3 Articulatory, auditive and acoustic phonetics

Speech sounds can be described in various ways, depending on the kind of

property that is taken as the focus of investigation:65

Articulatory phonetics describes sounds with respect to the way in
which they are produced, i.e. with respect to the speech organs involved.

Auditory phonetics investigates the way in which speech sounds are
perceived in the ear.

Acoustic phonetics studies the physical properties of sounds such as
their frequency.
Acoustic measurements are often represented in the form of so-called spec-
trograms, which show wave lengths and the type of vibration produced.66

64
The symbol h indicates aspiration.
65
For an introduction to the principles of acoustic and auditory aspects of pho-
netics see Gut (2009: 137–195).
66
Frequencies are shown on the vertical axis in this spectrogram. The different
shades of colour refer to intensity, black areas indicating frequencies of high
intensity. See Gut (2009: 147).
46 Sounds

Broadband spectrogram of the utterance “Please help me”.

(taken from Gut 2009: 148)

Acoustic measurements of this kind are probably the most precise method
of analysing speech sounds. However, it is still customary to make use of
the terminology of articulatory phonetics for most purposes of linguistic
description.
One fundamental distinction that is made in the classification of speech
sounds is that between vowels and consonants. This distinction can be
made on the basis of the phonetic properties of the sounds or on the basis of
their function in the syllable.
Phonetically, the distinction refers to the different ways in which vowels
and consonants are produced: 67

67
Cf. Gimson and Cruttenden (62001: 27): “This type of definition might define
vowels as median (air must escape over the middle of the tongue, thus ex-
cluding the lateral [l]), oral (... thus excluding nasals like [n]), frictionless
(thus excluding fricatives like [s]), and continuant (thus excluding plosives
like [p]); all sounds excluded from this definition would be consonants.”
Compare also the definitions given by Jones (1918/1960: 23), who draws the
 4 The sounds of English: phonetics 47

in the case of vowels, the air stream can pass from the lungs through the
mouth across the central part of the tongue without any obstacle that
would stop it or cause audible friction,

whereas this is not the case with consonants.
With respect to the syllable (> 4.5), one can say that

vowels have a central function in the syllable,

whereas consonants have a marginal function (Gimson 41989: 30).
These two definitions coincide in the case of the three phones of the word
teach in that [i;] has a central function in the syllable and shows no closure
of the speech tract during its articulation and thus is clearly a vowel and [tS]
and [th] are not central in the syllable and do show such closure. However,
this is less clear in the case of the [w] in will, which, like [j] in yes, is con-
sonantal as far as its position in the syllable is concerned but phonetically is
best described as a vowel glide.68 Gimson (41989: 33–34) thus describes
[w] and [j] as semi-vowels, although more recently it has become common
to include them under the class of approximants (Roach 42009: 50–52,
Gimson and Cruttenden 62001: 211–216).69

____________________________
line between vowels and consonants in such a way that “consonants ... in-
clude (i) all sounds which are not voiced (e.g. p, s, É), (ii) all sounds in the
production of which the air has an impeded passage through the mouth (e.g.
b, l, rolled r), (iii) all sounds in the production of which the air does not pass
through the mouth (e.g. m), (iv) all sounds in which there is audible friction
(e.g. f, v, s, z, h)”. Compare also Pike (1943: esp. 145), who distinguishes be-
tween vocoids and contoids at the phonetic level and between syllabic and
non-syllabic with respect to the function of sounds within the syllable. For an
elaborated discussion of different terminologies referring to phonetic quality
and function in the syllable see Abercrombie (1967: 79–80).
68
Cf. Gimson (41989: 33–44).
69
See also Gut (2009: 55). For [R] see Gimson (41989: 33) and Gimson and
Cruttenden (62001: 27). That the distinction between vowels and consonants
is not perfectly straightforward also becomes apparent when one considers
acoustic criteria, according to which consonants contain a noise component
which vowels lack. This distinction does not fully coincide with that made on
articulatory grounds as is pointed out by Arnold and Hansen (1975/81992:
26). Arnold and Hansen (41992: 32) use the term Sonoranten for [l] and the
nasals. Compare also the classification of consonants provided by Dretzke
(1998: 47). For the distinction between sonorants and obstruents see also
Gimson (41989: 34).
48 Sounds

4.4 Description of sounds in articulatory terms

Speech sounds are produced by modifying air streams. This can be done by
using organs such as the tongue, the lips and the velum (or soft palate).70

The vocal tract. The three cavities and the articulators of the articulatory
system. (1= pharyngeal cavity; 2 = oral cavity; 3 = nasal cavity)
(taken from Gut 2009: 24)

Articulatory phonetics describes phones according to the way these modifi-

cations of air streams are achieved:
− Depending upon whether the lungs are involved in producing an air
stream or not, phones are classified as pulmonic or as non-pulmonic.
− If the air stream is pushed out of the lungs, the sound is classified as
egressive, if it is produced while breathing in, the phone is classified as

70
For a detailed description of the respiratory system, the phonatory system and
the articulatory system see Gut (2009: 13–27).
 4 The sounds of English: phonetics 49

ingressive. All phones of English are normally egressive, but an ingres-

sive speech mode can be used too in some languages.71
− If the vocal cords are so far apart that they allow the air stream to pass
without modification, the resulting sound is called voiceless, if they are
so close together that they cause the air to vibrate the speech sound is
called voiced.
− Depending on whether the air stream is directed through the mouth, the
nose or both, sounds are classified as oral, nasal or nasalized.
These criteria are usually applied in the description of any speech sound.
As far as further criteria are concerned it is usual to use separate sets of
criteria for vowels and for consonants.
Consonants can be described in terms of criteria such as the following:

the place of articulation, which indicates where the air stream is
modified and which articulatory organs are involved in this modifica-
tion: Since [th] in English teach is produced by pressing the tongue
against the alveolar ridge (the part of the palate behind the teeth), this
sound is classified as alveolar, for example. The following places of ar-
ticulation are relevant in the description of English:72

bilabial lips [ph] [p] [b] [m]

labio-dental lower lip and upper teeth [f] [v]
dental tongue and upper teeth [θ] [ð]
alveolar tip and rims of the tongue and [th] [t] [d] [s] [z] [n]
upper alveolar ridge and side [l]
teeth

71
Cf. Abercrombie (1967: 24–25): “Linguistic use of an ingressive pulmonic
air-stream is certainly not common, but it can nevertheless be found in many
parts of the world in a variety of circumstances. Thus in English the first part
of the word yes, spoken in a somewhat off-hand manner, is often pronounced
by many people with an ingressive pulmonic air-stream ... The quality of the
voice is considerably changed when this air-stream is used ... and in a number
of communities an ingressive air-stream is used as a disguise when the
speaker cannot be seen and does not wish to be recognized.”
72
For a much more detailed description of manner and place of articulation of
English consonants see Gimson (41989: Chapter 8).
50 Sounds

post-alveolar tip, blade, and rims of the tongue [r]

and rear part of upper alveolar
ridge
palato- tip, blade, and rims of the tongue [ʃ] [ʒ]
alveolar and upper alveolar ridge and side [tʃ] [dʒ]
teeth
velar tongue and soft palate [kh] [k] [g] [ŋ]
glottal vocal cords [ʔ] [h]

the manner of articulation, which describes how the articulatory

organs operate to produce this sound. In the case of [th], for in-
stance, tongue and alveolar ridge form an obstruction which is then
opened so that the air stream is suddenly released. Accordingly, [th] is
termed plosive.

plosive / stop articulatory organs form obstruc- [ph] [p] [th] [t] [kh] [k]
tion; air stream is held up; sud- [ʔ]
den release of air [b] [d] [g]
fricative articulatory organs brought so [f] [v] [θ] [ð]
close together that friction of air [s] [z] [ʃ] [ʒ] [h]
stream occurs
affricate 73 plosive with friction during re- [tʃ] [dʒ]
lease stage
lateral partial closure so that air stream [l]
can escape on one or both sides
of obstruction
approximant contraction of tongue; air stream [r] [w] [j]
can escape without friction
nasal air stream released through nose; [n] [m] [ŋ]
articulatory organs form obstruc-
tion

73
Gimson (41989: 177–179) also identifies two post-alveolar affricates [tr] and
[dr].
 4 The sounds of English: phonetics 51

the amount of energy involved, resulting in a distinction between for-

tis (high energy) and lenis (little energy).

the fact whether there is aspiration, a short air stream following the
articulation, which is the case with the [ph] in English pit but not with
[p] in English top, for instance.
Vowels are generally described in terms of

the place of the highest point of the tongue in the mouth during
the articulation of the vowel. It can be located according to the horizon-
tal and the vertical dimension: depending on whether the front, the cen-
tral part or the back of the tongue takes the highest position, a distinc-
tion is made between front, central and back vowels whereas the
degree of closeness of the highest point of the tongue towards the
palate results in a distinction between close, close-mid (half-
close), open-mid (half-open) and open vowels.

the degree of lip rounding, where a distinction between rounded, neu-
tral and spread is made.

length (in terms of long and short vowels).
Furthermore, a distinction can be made between monophthongs, (i.e. rela-
tively pure vowels such as [&]), diphthongs and triphthongs (which involve
a glide from one position to another such as [I@] and [aI@]).74
The place of the highest point of the tongue is often indicated in the so-
called vowel diagram in terms of a front-back dimension and a close-open
dimension, which can be seen as an abstraction from the shape of the
mouth.75 In order to arrive at a precise description of vowel quality, refer-
ence can be made to the so-called cardinal vowels, which was outlined in
detail by the famous British phonetician Daniel Jones.76 Jones (1918/91960)
postulated eight primary cardinal vowels, which represent the extreme
points of the possible position of the tongue and which can be located in the
vowel diagram in the following way:

74
Cf. Gimson and Cruttenden (62001: 129) and Gut (2009: 60).
75
Compare also the English Pronouncing Dictionary (EPD17 2006: viii) and
Herbst, Stoll, and Westermayr (1991: 47).
76
For the history of the idea of a cardinal vowel system and references to A. M.
Bell, A. J. Ellis and H. Sweet see Kohler (1977). For primary and secondary
cardinal vowels see Jones (1918/91960: 35–36).
52 Sounds

It is important to realize that the eight cardinal vowels have to be thought of

as idealized realizations of particular vowel qualities and not as sounds
actually occurring in any particular language. The English vowels /I/ (as in
bid) and /&/ (as in bad) can be represented in this diagram in the following
way:

The purpose of the system of cardinal vowels is to serve as a reference grid

for the exact description of vowels in real languages and to make it easier
to compare vowels of different languages. So the difference between the
English short [ɪ] in a word such as bitter and the German short [ɪ] in a word
such as bitte can be expressed by saying that the German [ɪ] comes much
closer to the position of cardinal 1 than the English one.
A phonetic description of phones of English could then take the follow-
ing form:
 4 The sounds of English: phonetics 53

ð voiced lenis dental fricative oral egressive pulmonic

æ voiced front open/open-mid oral egressive pulmonic
n voiced lenis alveolar nasal nasal egressive pulmonic
g voiced lenis velar plosive oral egressive pulmonic
h
t voiceless fortis alveolar plosive aspirated oral egressive pulmonic
t voiceless fortis alveolar plosive oral egressive pulmonic

4.5 Syllables

Syllables are units above the individual sound segments since they can
consist of one or more phones. The structure of the syllable can be de-
scribed by distinguishing between onset, peak, and coda (Gimson and Crut-
tenden 62001: 51, Crystal 22003: 246), where the peak is the central and
usually most sonorous element of the syllable, whereas onset and coda are
marginal:

t i; tS
str æ ndz
pl i; z
b OI
aI z
aI

In terms of the distinction between vowels and consonants as made in 4.3,

the peak of a syllable consists of a vowel, whereas the marginal elements
consist of one or more consonants. The approximants [j, w, r], which pho-
netically are like vowels, behave like consonants in this respect. In words
54 Sounds

such as little or button, so-called syllabic [l̩ ] and syllabic [n!] can function as
the peak of the syllable.77
Syllables ending in a vowel (buoy, sea), are called open syllables, syl-
lables containing a coda, i.e. ending in a consonant (boys, girl) are called
closed syllables.

4.6 Suprasegmental elements

Apart from the sound segments identified in 4.4 there are a number of pho-
netic properties that extend to more than one segment, so-called su-
prasegmental elements.
One such element that is associated with the unit of the syllable and not
with an individual sound segment is stress. Thus the syllable house is
stressed in a word such as housekeeper but unstressed in lighthouse. Stress
is usually associated with more energy in the pronunciation of a syllable.
From the point of view of perception stressed syllables are generally de-
scribed as having more prominence, which arises from such factors as
loudness, length, pitch and vowel quality.78
In phonetic transcription, accented (or stressed) syllables are indicated
by the symbol " for primary stress and % for secondary stress within a word:
lighthouse ["laIthaUs]
lighthousekeeper ["laIthaUs%ki;pə]
Another such element is pitch change, where Gimson (41989: 272–274)
distinguishes between a falling tone, a rising tone, a falling-rising tone and
a rising reinforcement of a fall.79
Variations of pitch and the direction of pitch within an utterance result
in its intonation. A sentence such as
(1)SP The English have no respect for their language

77
See Gimson and Cruttenden (62001: 52) and Arnold and Hansen (1992: 165–
166) for a more detailed discussion.
78
See Roach (42009: 73–74). Compare also Gimson (41989: 224) and Gimson
and Cruttenden (62001: 272–274), where the term stress is avoided altogether.
79
Other linguists make different kinds of distinctions, cf. Esser (1979), Halliday
(1970a) or Cruttenden (1986). Gut (2009: 118) distinguishes the following
tones: fall, rise, fall-rise, rise-fall, rise-fall-rise and level.
 4 The sounds of English: phonetics 55

can be pronounced with a falling or a rising intonation, for example,

the former is likely to be interpreted as a statement, the latter as a
question or an exclamation. Stress and pitch change can signal the
prominence of a word in a sentence as in
(1a) The English have no respect for their language
or
(1b) The English have no respect for their language
or mark a primary word accent. 80
As far as rhythm is concerned, English, like German and unlike
French or Italian, is often said to belong to the so-called stress-timed lan-
guages, which means that in connected speech intervals between stressed
syllables tend to be of the same length, i.e. the stressed syllables are
isochronous. Thus the interval of time between the stressed syllables in
(2)QE "Which is the "train for "Crewe "please
is said to be roughly the same, even if the accented syllable is followed by
two unaccented syllables in the case of which, by one in the case of train
and by none in the case of Crewe. This results in a different kind of rhyth-
mic patterning from that of syllable-timed languages, which require
roughly the same amount of time for each syllable.81

80
Cf. Gimson (41989: 225 and 269).
81
See Abercrombie (1967: 97–98) for a detailed discussion of this example. It
has often been stated, though, as Cruttenden (1986: 25) points out, on a rela-
tively poor empirical basis, that English shows isochrony, which is de-
scribed by Gimson (41989: 263–264) as follows: “It is noticeable that the
rhythmic beats of an utterance occur at fairly equal intervals of time. As a re-
sult of this, the speed at which the unstressed syllables are uttered – and the
length of each – will depend upon the number occurring between the strong
beats.” Compare also the account given in Gimson and Cruttenden (62001:
251), where the lack of instrumental verification is stressed: “the occurrence
of full vowels generally predicts the rhythm of English rather more usefully
than any notion of stress ...”.
5 Phonology

5.1 The function of speech sounds

5.1.1 Phonemes and allophones

Phonetics deals with a description of phones, i.e. with formal properties of

speech sounds. A next step in the analysis of a language can be to further
analyse and classify the phones identified, particularly with respect to their
function within the language. Here, it must be emphasized that speech
sounds as such do not have meaning. If one accepts the de Saussurean view
of the linguistic sign as a combination of form and meaning, then the func-
tion of speech sounds can be described as combining with other speech
sounds to make up such forms, which can be distinguished from other
forms with different meanings. In those terms, the function of speech
sounds is to distinguish between meanings. For instance, the phones [th],
[i;] and [tS] enter into the combination of [thi;tS], which contrasts with other
linguistic forms such as the following:

[thi;tS] (teach) – [ri;tS] (reach) – [li;tS] (leach, leech) – [phi;tS] (peach)

[thi;tS] (teach) – [thVtS] (touch) – [thO;tS] (torch)
[thi;tS] (teach) – [thi;k] (teak) – [thi;l] (teal) – [thi;m] (team) – [thi;n] (teen) –
[thi;T] (teeth)

Minimal pairs such as teach – reach can be taken to show that each of the
three sounds in [thi;tS] is relevant to the identity of the form [thi;tS]. Thus it
can be argued that one function of sounds is to contrast linguistic forms
with different meanings. However, not all speech sounds in a language
contrast in this way. Thus, replacing [th] by [t] does not result in a form
with a new meaning, not only in the case of [thi;tS] but nowhere in English.
[ti;tS] could be considered an unusual or impeded pronunciation of teach,
but not as a form representing a different meaning from [thi;tS].
Observations of this kind have led to the distinction between phonemes
and allophones:
 5 Phonology 57

A phoneme is an abstract linguistic unit at the level of sound that

serves to distinguish between linguistic forms with different meanings.

An allophone is one of several phonetic realizations of a phoneme in
the form of a speech sound.
h
[t ] and [t] are thus two allophones of the phoneme /t/. These two allo-
phones are in complementary distribution in that each of them can only
occur in a position in which the other cannot occur. Thus, whenever the
allophones of a phoneme are in complementary distribution,82 it is predict-
able which allophone will realize that phoneme. For the English plosives
/p/, /t/, /k/ the following distributional rule can be stated:83

Aspirated and non-aspirated allophones of the English phonemes /p/ /t/ /k/:
[ph] [th] [kh] initial position in accented syllables [phi;tS] [thi;tS] [khi;]
(strong aspiration)
[p] [t] [k] in all other positions [spi;k] [rI"spekt]
(no or weak aspiration)

According to these principles, the two speech sounds [l] (clear l) and [ł]
(dark l), to be found in English pronunciations of language and will, for
instance, can be subsumed under a single phoneme /l/. This is because [l]
and [ł] can never serve to contrast forms with different meanings in English
(although this may well be possible in other languages). Again, [l] and [ł]
are in complementary distribution:

Distribution of allophones clear l and dark l in English:

[l] before vowels and /j/ ["l&NgwIdZ] ["INglIS]
[ł] before consonants [bVłk]
at the ends of words [wIł]

82
Cf. Gimson (41989: 49): “No two realizations of a phoneme (its allophones)
are the same. This is true even when the same word is repeated ... When the
same speaker produces slightly different pronunciations of the word cat, the
different realizations of the phonemes are said to be ‘free variants’.”
83
For a much more detailed account of the allophones of the English plosives
see Gimson (41989: 152–155).
58 Sounds

A common principle to establish the phoneme inventory of a language is

the one used above, namely to find so-called minimal pairs, i.e. linguistic
forms with different meanings which are phonetically identical with the
exception of one sound such as [thi;tS] (teach) – [ri;tS] (reach).

5.1.2 Phonetics and phonology

The difference between phonetics and phonology is basically that phonetics

describes the properties of speech sounds as they occur in language,
whereas phonology is concerned with the kinds of contrasts at the level of
sound that can be seen as creating differences of meaning in a particular
language. Phonetics thus deals with the formal or physical analysis of con-
crete entities, the phones, whereas phonology deals with phonemes, which
have to be regarded as abstract units of linguistic analysis.84

5.2 The description of phonemes

5.2.1 Consonant phonemes

Phonemes are generally described with reference to the articulatory features

of their phonetic realization. Since the phoneme is an abstract unit and
since the idea behind the phoneme concept is one of functional contrast, it
is unnecessary and inappropriate to draw upon the whole inventory of fea-
tures used for the description of the sounds of a language. In phonology,
only those features are used that are necessary to distinguish all the pho-
nemes of a language from one another; features that all sounds of a lan-
guage share (such as ‘pulmonic’ or ‘egressive’ in the case of English) can
be ignored as can features that are not shared by all allophones of a pho-
neme (like aspiration in the case of the English plosives).
It is possible to establish the contrast between the consonant phonemes
of English (BBC English/Received Pronunciation and General American)85
in terms of three sets of distinctive features:
84
There are many different definitions of the term phoneme. Note that the fa-
mous British phonetician Daniel Jones (1944/1973: 169, 178) believed “pho-
nemes to be undefinable” and described the phoneme as “a family of sounds
in a given language ...”.
85
Cf. EPD17 (2006: x), Gimson and Cruttenden (62001: 148) and Giegerich
(1992: 41). Giegerich and Gimson and Cruttenden also identify a bilabial
 5 Phonology 59

Manner of articulation Place of articulation voiceless – voiced

/p/ /b/ plosive bilabial voiceless – voiced

/t/ /d/ plosive alveolar voiceless – voiced
/k/ /g/ plosive velar voiceless – voiced

/tS/ /dZ/ affricate palato-alveolar voiceless – voiced

/f/ /v/ fricative labio-dental voiceless – voiced

/T/ /D/ fricative dental voiceless – voiced
/s/ /z/ fricative alveolar voiceless – voiced
/S/ /Z/ fricative palato-alveolar voiceless – voiced
/h/ fricative glottal

/m/ nasal bilabial

/n/ nasal alveolar
/N/ nasal velar

/l/ lateral (approximant) alveolar

/r/ approximant post-alveolar
/j/ approximant palatal
/w/ approximant velar

It is important to realize that, since phonemes are abstractions made in the

process of linguistic analysis, the establishment of a phoneme system is
dependent on a number of decisions of a relatively arbitrary nature.

____________________________
fricative /„/, which, however, does not occur in all varieties; similarly a velar
fricative /x/ is listed by Giegerich and EPD17. In Gimson and Cruttenden
(62001) /l/ is listed as an approximant.
60 Sounds

One such decision concerns the phoneme status of /T/ and /D/, which
might also be treated as allophones of one phoneme since in present-day
English minimal pairs are notoriously difficult to find: thigh – thy is unsat-
isfactory since thy is restricted in its use to archaic church language, which
leaves a few minimal pairs such as loath – loathe, mouth (noun) – mouth
(verb) and wreath – wreathe. One argument in favour of a phoneme treat-
ment may be seen in the fact that it would not be possible to establish rules
of complementary distribution for [T] and [D] as allophones, another is that
the phonemic contrast between voiced and voiceless members can be found
for all other English fricatives, which in a way makes it aesthetically tempt-
ing to postulate such a contrast in the case of [T] and [D], too.86 A similar
problem concerning minimal pairs is presented by /h/ and /N/, which are in
complementary distribution, /h/ only occurring in syllable-initial position,
/N/ never occurring in syllable-initial position. Here, however, lack of pho-
netic similarity is regarded as an argument against treating the two as allo-
phones of the same phoneme.
A third case of possible controversy is presented by the affricates, which
could also be analysed as a sequence of a plosive and a fricative. One rea-
son mentioned by Gimson (41989: 174) for treating /tS/ and /dZ/ as one
phoneme is that “the native speaker does not regard /tS, dZ/ as composite
sounds”.
A problem of a slightly different kind concerns the choice of the distinc-
tive features that characterize particular phonemes. So it is possible to use
the ‘voiceless’ – ‘voiced’ contrast to establish the distinction between pho-
nemes such as /t/ and /d/, but it is equally possible to express the opposition
in terms of ‘fortis’ and ‘lenis’ since in Received Pronunciation voiceless
consonants tend to be fortis and voiced ones to be lenis.87 Which of the two
types of opposition is chosen to be a distinctive feature in the phonological
description is up to a point a matter of personal preference; to use both
would make the description redundant, however.

5.2.2 Vowel phonemes

As with consonants, not all phonetic features of English vowel sounds have
the status of distinctive features of the corresponding vowel phonemes.

86
There is a tendency for /T/ to occur in nouns or adjectives (thing, thick) and
for /D/ to occur in function words such as the or they.
87
This is a generalization of some kind, as will be pointed out later.
 5 Phonology 61

Since there is no phonemic contrast between egressive and ingressive

vowel articulation in English, egressiveness is not relevant to a phonologi-
cal description of English. The same is true of lip rounding.
As far as vowel quality and vowel quantity are concerned, the decision
as to which of these ought to be treated as a distinctive feature in phono-
logical terms is not quite as straightforward. In some accents of English
such as Received Pronunciation differences in vowel length and differences
in vowel quality coincide. Thus (in identical contexts) the i- and u-vowels
whose quality comes relatively close to that of the cardinal vowels 1 and 8,
for instance, are longer than those which are more centered. Phonologi-
cally, this means that the difference between the vowels of bead and bid or
of pool and pull can be expressed by either treating quantity or quality as a
distinctive feature, which is similar to the fortis/voiceless and lenis/voiced
problem outlined above. In the first case one can express the difference by
a length mark, in the second by different symbols representing the quality
of the sounds:

beat bit pool pull

phonetic realization [bi;t] [bIt] [pu;ł] [pUł]
phonological representation based on
/bi;t/ /bit/ /pu;l/ /pul/
quantity (used in EPD13)
phonological representation based on
quality (used e.g. by Giegerich 1992: /bit/ /bIt/ /pul/ /pUl/
45, Brown 1977: 33 or Gut 2009: 63)

While both systems are used in discussions of English phonology, many

standard reference books and pronunciation dictionaries represent RP vow-
els in a redundant notation indicating both differences in quality and quan-
tity.

phonological representation based on

quality and quantity (used e.g. by Gim- /bi;t/ /bIt/ /pu;l/ /pUl/
son/Cruttenden 62001: 91 or EPD17)
62 Sounds

Thus the vowel phonemes of RP can be represented as follows (Roach,

Hartman, and Setter EPD17, 2006: viii–ix):

Monophthongs:

Diphthongs:

It has to be said that the number of vowel phonemes identified and the
symbols used for different varieties of English vary to a certain extent.88
The list below shows the twenty vowel phonemes identified by Roach,
Hartman, and Setter (EPD17) and Gimson and Cruttenden (62001: 91–93)
for British English (BBC English or Received Pronunciation). For compari-

88
See, for example, Brown (1977: 33) or Giegerich (1992: 48–75). Some lin-
guists include triphthongs such as /aI@/ (fire) or /aU@/ (power) for RP, see
Trudgill and Hannah (52008: 16) and Gut (2009: 63).
 5 Phonology 63

son, the symbols used for American English in EPD17 are given on the
right.89

RP/BBC English US American

/I/ bit, bid /I/ bit, bid
/e/ bet, bed /e/ bet, bed
/æ/ bat, bad /æ/ bat, bad
/ʌ/ hut, putt /ʌ/ hut, putt
/ɒ/ hot, golf
/ʊ/ foot, put /ʊ/ foot, put
/ə/ tomato, about /ə/ tomato, about
/iː/ sea, see /iː/ sea, see
/ɑː/ park, father /ɑː/ father, golf, hot
/ɔː/ caught, north /ɔː/ north, port
/uː/ cool, queue /uː/ cool, queue
/ɜː/ bird, first
/eI/ bay, may /eI/ bay, may
/aI/ buy, light /aI/ buy, light
/ɔI/ buoy, employ /ɔI/ buoy, employ
/aʊ/ now, house /aʊ/ now, house

89
For similar overviews see Trudgill and Hannah (52008: 16 and 42) or Gut
(2009: 63). Compare also the Longman Pronunciation Dictionary (LPD
3
2008: xvi) and the descriptions provided by Wells (1982: 118–127) or
Giegerich (1992: 43–61). For the use of length marks for American English
see Roach, Hartman and Setter (EPD17: ix). Compare also Wells (1982:
120), who uses length marks for RP but not for General American. See also
Giegerich (1992: 72–73), who does not make use of length marks in phoneme
symbols but whose phonetic description of a phonemic vowel symbol such as
/I/ includes the statement: “In GA and RP it is also of shorter duration than /i/
in comparable contexts.”
64 Sounds

/əʊ/ blow, boat

/oʊ/ boat
/ɪə/ near
/eə/ where
/ʊə/ moor
ɚ “r-coloured schwa” as in mother
(EPD17)
ɝ “r-coloured bird vowel” as in bird
(EPD17)

5.2.3 Phonemic principle of pronunciation dictionaries

Pronunciation dictionaries such as the Cambridge English Pronouncing

Dictionary (EPD 172006) or the Longman Pronunciation Dictionary (LPD
3
2009) generally follow the principles of a phonemic description. EPD17
(viii, xiv) mentions two exceptions to this principle – namely the use of [t}]
to show the voicing and flapping of /t/ in words such as getting in Ameri-
can English. The second exception concerns the use of the symbol [i] e.g.
for the final vowel in busy on the grounds that in certain positions the pho-
nological distinction between /i;/ and /I/ is neutralized.90
city "sIti seedy "si;di holidaymaker "hQlədi%meIkə
One consequence of the generally phonemic character of the transcriptions
provided by pronunciation dictionaries is that users must always be aware
that the symbols used represent abstractions whose phonetic realization
depends on the variety referred to. Thus both LPD and EPD provide tran-
scriptions for the place name Newcastle (upon Tyne) for three different
varieties of English:

90
Note, however, that Wells (LPD 32008: xxv) includes /i/ and /u/ under the
weak vowel phonemes, whereas Roach, Hartman and Setter (EPD17: xiv)
point out that /i/ “is not, strictly speaking, a phoneme symbol”. Wells (LPD
3
2008: 539) points out that there are two environments in which /I/ and /i:/
cannot be distinguished so that the opposition between these phonemes is
“neutralized”: with vowels in weak syllables at the end of a word (happy) or
with vowels in weak syllables before another vowel (radiation).
 5 Phonology 65

British English: "nju;%kA;s@l

American English: "nu;%kæs@l
Northern English English: nju;"kæs@l
This must not be taken to mean that the American and the local pronuncia-
tion of the vowel transcribed as /æ/ were very similar. In fact, in Northern
English, it tends to be pronounced as [a], whereas the American vowel is
noticeably closer – the symbol /æ/ indicates that both in American English
and Northern English English castle has the same vowel phoneme as bat.

5.3 Phonotactics

There are certain restrictions on the types of combinations of phonemes that

are possible in a particular language. The branch of phonology that studies
the distributional and combinatorial properties of phonemes is known as
phonotactics.
It is interesting to see, for instance, that the short vowels /e/, /&/, /V/ and
/Q/ do not occur in word final position or that only a limited number of
combinations of two consonants in word initial position are possible in
English (Gimson, 41989: 243):91

/p/ + /l/ /r/ /j/

/t/ + /r/ /j/ /w/
/k/ + /l/ /r/ /j/ /w/
/b/ + /l/ /r/ /j/
/d/ + /r/ /j/ /w/
/g/ + /l/ /r/ /j/ /w/
/m/ + /j/
/n/ + /j/
/l/ + /j/
/f/ + /l/ /r/ /j/
/v/ + /j/
/T/ + /r/ /j/ /w/

91
Note that such lists of phonotactic possibilities of a language clearly depend
on what one considers to be a word of that language. Cruttenden, in his sixth
edition of Gimson’s book (62001: 240-241), adds combinations such as /vl/ or
/Sw/ on the basis of such words as Vladivostok and Schweppes, which quite
clearly present borderline cases.
66 Sounds

/s/ + /l/ /j/ /w/ /p/ /t/ /k/ /m/ /n/ /f/
/S/ + /r/
/h/ + /j/

The options for word initial clusters consisting of three consonants are even
more restricted (Gimson 41989: 245). Interestingly enough, the first conso-
nant in such a combination must always be an /s/ and it can only be fol-
lowed by a fortis plosive /p/, /t/ or /k/:

/s/ + /p/ + /l/ /r/ /j/

/s/ + /t/ + /r/ /j/
/s/ + /k/ + /l/ /r/ /j/ /w/

What makes phonotactic rules of this kind interesting beyond their purely
descriptional value is that one may safely assume that knowledge of such
regularities may well help speakers – and especially listeners – recognize
word boundaries and thus be instrumental in word recognition processes.
6 Phonetic “reality”

6.1 Problems of the phoneme concept

6.1.1 The problem

Phonetics (> Chapter 4) is concerned with the description of speech sounds

(phones), whereas phonology (> Chapter 5) investigates the function of so-
called phonemes in terms of establishing meaning distinctions in a lan-
guage. The concept of the phoneme is a useful one for certain purposes of
linguistic analysis. For instance, it helps to describe certain regularities in
the area of morphology (such as phonological conditioning of allomorphs;
see 8.3.4).
It would be tempting to imagine that word recognition takes the follow-
ing form:
− hearers perceive a series of sounds [a], [b], [c] etc.
− relate these to a series of corresponding phonemes /A/, /B/, /C/ etc.
− and then arrive at a meaningful unit (a morpheme or word) consisting of
a representation of this string of phonemes and a representation of some
meaning.
However, it seems things are more complicated than that because
− the way a phoneme is realized phonetically may depend on its position
in the word (> 5.1.1),
− phonemes are influenced in their phonetic realization by sounds follow-
ing or preceding them (> 6.2),
− one and the same phone may represent several phonemes (> 6.1),
− not all varieties of a language would necessarily be analysed as having
identical phoneme inventories (> 19.3) even if speakers of these varie-
ties can understand each other without any problems,
− speakers’ anatomical properties influence the quality of sounds they
produce.
68 Sounds

It is thus an open question to what extent the notion of the phoneme is rele-
vant to speech perception.92

6.1.2 Phonetic value of phonological features

The concept of the phoneme involves a good deal of abstraction from pho-
netic reality. This can be illustrated by looking at pairs such as /p – b/ or /t
– d/ in English. At the phonemic level, the contrast between these pho-
nemes is often described in terms of the distinctive features ‘voiceless’ and
‘voiced’. This is misleading in so far as not every phone that realizes the
‘voiced’ phoneme /b/, for example, needs to be phonetically ‘voiced’; in
fact, RP /b d g/ seem to have full voicing only when they occur between
voiced sounds (Gimson and Cruttenden 62001: 152).
In fact, in words such as peat or bead (i.e. when occurring in word-
initial position followed by a stressed vowel) the difference between plo-
sives such as /p/ and /b/ can best be described in terms of voice onset time,
i.e. the time that passes between the release of the closure of the plosive
and the voicing of the following sound. Word-initial /p/ and /b/ need not
involve any voicing at all. Rather, /p/ tends to be aspirated, which means
that there is a longer voice onset time than with /b/.93 Brown (1977: 28)
describes this as follows:94
peat _________ ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,ʔ ________________

p i∑ i ʔ t
bead ______,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, ___________
b i d
_______ closure
,,,,,,,,,,,,,, voicing

92
For a more detailed account of speech perception (and factors such as voice
onset time) and word recognition see Gut (2009: 195–200) and Wimmer and
Perner (1979: 106–119).
93
See Catford (1988: 190–192) for details. Compare also Gimson and Crutten-
den (62001: 152–153).
94
Cf. Brown (1977: 28): “Initially, the ‘voiced’ stops are realized by a period of
voiceless closure (i.e. no vibration of the vocal cords) with, as the closure is
released, immediate onset of voicing in the following segment. The differ-
ence between initial ‘voiceless’ and ‘voiced’ stops lies, then, in the timing of
onset of voicing immediately following the release of the closure. The behav-
iour of the vocal cords during the period of closure itself is no different.”
 6 Phonetic “reality” 69

Thus the difference in the realization of the consonant phonemes /p/ (with
the phonological feature ‘voiceless’) and /b/ (with the feature ‘voiced’) is
created by aspiration and voice onset time, but not by any voicing of [b].
In word final position, ‘voiced’ phonemes such as /b d g/ can actually be∑
realized by a partially voiced or completely voiceless lenis phone [b̥ d̥ g ]
(Gimson and Cruttenden 62001: 152). The problem caused by this for pho-
nological description could perhaps be avoided by referring to the ‘for-
tis’/‘lenis’-distinction (> 4.4) instead of to the ‘voiced’/‘voiceless’-
distinction in the description of these phonemes. What is much more im-
portant, however, is the fact that the contrast between final /p t k/ and final
/b d g/ manifests itself in the length of the preceding vowel. Thus the main
phonetic difference between bat /bæt/ and bad /bæd/ is not to be found in
the articulation of the final consonant but in the length of the preceding
vowel; the vowel in bad being considerably longer than that in bat.95

bat bad
phonological contrast /t/ /d/
(main) phonetic contrast [æ] [æ;]

This is not only interesting with respect to linguistic theory but also with
respect to foreign language teaching. If the length contrast in the vowel is
more important for perception than the voicing of the consonant, then it is
important to focus on the vowel rather than the consonant in teaching. In
any case, it is essential to realize that a phonological feature of one pho-
neme can be realized phonetically in the articulation of another phoneme.
How complicated the interaction of phonological and phonetic features
can be is shown by the fact that so-called long-vowel phonemes can be
shorter in articulation before ‘voiceless’ plosive phonemes than so-called
‘short’-vowel phonemes before ‘voiced’ plosive phonemes.96

95
Cf. also Gimson and Cruttenden (62001: 158): “The voiceless series /p,t,k/
will, of course, be distinguished in final positions from the voiced series
/b,d,g/ either by the reduction of length of the sounds preceding /p,t,k/ or by
the presence of some voicing in /b,d,g/, or by a combination of both factors.
The non-release of final plosives is a feature of colloquial RP.” For glottal re-
inforcement of final /p,t,k/ see Gimson and Cruttenden (62001: 159).
96
Compare Brown (1977: 29): “A much more important point to dwell upon,
since this is not usually stressed in manuals of English pronunciation and it is
70 Sounds

/bi;d/ [bi;…d] (bead) > /bId/ [bI;d] (bid) > /bi;t/ [bi…t] (beat) > /bIt/ [bIt] (bit)
This can be confirmed by measurements of the duration of vowels (Wiik
1965: 114):97
/i;/ + voiced fricative 36 csec.
/I/ + voiced fricative 18.6 csec.
/i;/ + voiceless fricative 13 csec.
/I/ + voiceless fricative 8.3 csec.
This means that the opposition between long and short vowel phonemes
can only be interpreted phonetically as holding within the same context, but
not in absolute terms. In fact, “/i;/ is typically shorter in a word such as seat
(12.3 csec.) than /I/ in a word such as hid (14.7 csec.)” (Gimson and Crut-
tenden 62001: 96).
What is important about these observations is the insight that the con-
trast between two phonemes need not necessarily be expressed in the seg-
ment that is generally regarded as the phone realizing that particular pho-
neme. A particularly interesting case in point is that of the pronunciation of
words such as write – ride and writer – rider in U.S. English (discussed at
length by Lass 1984: 30–31):98
− In write and ride the contrast between /d/ and /t/ is also reflected in a
difference in the length of the diphthong:
phonemic /raIt/ /raId/
phonetic [raIt] [rA>Id]
− In intervocalic position, both phonemes /t/ and /d/ are realized by the
alveolar tap [R] and not by [t] or [d].

____________________________
consequently often unknown to foreign teachers of English, is the way the
word final distinction between ‘voiceless’ and ‘voiced’ stops is made. The
main distinction is between a relatively short vowel, with ‘tight’ voicing and
glottal stop preceding a ‘voiceless’ stop, and a relatively long vowel with
‘full’ voicing preceding a ‘voiced’ stop. ... The main distinction between
‘voiceless’ and ‘voiced’ stops in word final position lies in the realization of
the preceding vowel – not in the articulation of the stop itself or of its re-
lease.”
97
See also Gimson and Cruttenden (62001: 95–96 and 152–153).
98
Compare Lass’s (1984: 30) discussion of violations of the linearity condition,
namely that “phonemic and phonetic differences must match up one-to-one in
linear strings”.
 6 Phonetic “reality” 71

− The contrast between writer and rider is then expressed not by the con-
sonant segments /t/ or /d/ but by the preceding vowel:
phonemic /raIt@r/ /raId@r/
phonetic [raIR´r] [rA>IR´r]
Lass (1984: 31) comes to the following conclusion regarding the role of
[aI] and [A>I] on the one hand and [t] and [d] on the other:
... two normally non-contrastive phones carry meaning difference, while
two (supposedly) contrastive phones fail to contrast. There are, as it hap-
pens, perfectly respectable ways of dealing with this – but not in the
framework of classical phonemics. They necessitate a fairly extensive revi-
sion of the conceptual framework, and lead to a more complex picture of
what natural-language phonologies are probably like.

6.1.3 The bi-uniqueness requirement

A further element of complexity arises through the fact that there is not
always one-to-one correspondence between phones and phonemes. The fact
that a phoneme can be realized by different allophones is central to the
theory and does not present any problems for decoding as long as each
phone can be attributed to one phoneme only (which can be referred to as
bi-uniqueness).99 There are cases, however, where one phone is used to
realize different phonemes: thus in many varieties of English the fortis plo-
sives /p t k/ can be realized by a glottal stop [ʔ] in word-final position, as in
the following examples:
/sæp/ [sæp] or [sæʔ]
/sæt/ [sæt] or [sæʔ]
/sæk/ [sæk] or [sæʔ]
It is clear that [ʔ] cannot be regarded as an allophone of just one of these
phonemes. Lass (1984: 28) arrives at the following conclusion:

99
Compare the description of the bi-uniqueness requirement given by Lass
(1984: 27): “The idea is that the listener’s ‘decoding’ of the speech signal is a
segmentation-and-classification routine: and this is not feasible unless a given
phone in a given (phonological) environment is always an allophone of one
particular phoneme.”
72 Sounds

The biuniqueness solution is thus counter-intuitive and messy. The obvious

solution is to permit extensive overlapping, giving a realization pattern
like: ...

[p] [t] [k]

  

/p/ /t/ /k/

  

[ʔ]
Such considerations make it clear that the concept of the phoneme is by no
means uncontroversial. One can only agree with the conclusions drawn by
Lass (1984: 29):
Such situations display theoretical ‘weak points’: here speaker intuition vs.
theoretical constraints lead to conflict. Theories have hidden consequences,
which emerge only through confrontation with data; and these conse-
quences may force us into revision or rethinking.
While this is not the place to go into alternative approaches to phonology
such as the feature theory developed by Roman Jakobson or generative
phonology,100 the purpose of these few examples was to illustrate that the
traditional concept of the phoneme is hardly able to cope with all problems
of analysis and that perhaps much more sophisticated concepts are needed.
Lass (1984: 164) argues that languages “may have different systems for
different word-positions, different accentual conditions, even different
morpho-syntactic or lexical categories”, which is a theoretical view of lan-
guage which he refers to as “at least potentially polysystemic, not mono-
systemic as in most classical phonological analysis.”101 In any case, the

100
Compare Jakobson, Fant, and Halle (1951) and Chomsky and Halle (1968).
For features see also Lass (1984).
101
For psycholinguistic aspects of this problem see Garman (1990: 186–188):
“Speech processing cannot be carried out initially on a phoneme-by-phoneme
basis, since the rate of their arrival at the ear ... would exceed the resolving
capacity of the auditory system. So there can be no question of simple ‘pho-
neme-detectors’ in speech perception. The implication is that what are de-
tected must be much smaller than phonemes: acoustic cues, which can be
processed within the time constraints, and which provide sufficient informa-
 6 Phonetic “reality” 73

limited explanatory power of the phoneme concept should not only be

borne in mind in the context of further theoretical research but especially
also when it comes to questions of contrasting different languages and of
teaching pronunciation.

6.2 Pronunciation in connected speech

6.2.1 Weakening of elements

The descriptive apparatus provided by phonetics and phonology can be

used to describe the phonological (or phonetic) form particular lexical units
have (which is what pronunciation dictionaries do, for example). It is clear,
however, that using words in connected speech, especially in fast informal
spoken language, means that these phonological forms can be subject to
considerable modifications.
A first type of modification that can be observed is that sounds can be
influenced in their quality by neighbouring sounds. This process is called
assimilation.102 Different degrees and directions of assimilation can be
distinguished: according to whether a sound takes over some or all of the
features of a neighbouring sound, there can be partial or total assimila-
tion. Depending on whether the sound that gets modified precedes or fol-
lows the sound it assimilates to, there is regressive or progressive as-
similation, and if two sounds influence each other that is a case of recip-
rocal assimilation.
dɪ"saɪsɪf"fæktə v
f decisive factor regressive – total
"kɒməmwelθ n
m Commonwealth regressive – partial
tʃɪkŋ ɪn
ŋ chicken progressive – partial
"dəʊnt"mɪʃʃə"treɪn sj
ʃʃ don’t miss your train reciprocal – total

A second type of modification occurring in rapid speech is elision, which

means that sounds get omitted altogether.

____________________________
tion for subsequent identification of phonemes.” Cf. Wimmer and Perner
(1979: 111) who point out that a /d/ in syllable initial position cannot be iden-
tified if it is not followed by a vowel.
102
For a detailed description of the processes of assimilation and elision see
Brown (1977).
74 Sounds

Furthermore, sounds can be weakened in their articulation so that /g/ is

realized by a weak [ƒ] in [s√m"taIm´ƒ´U] (some time ago), /k/ by a weak
[x] in ["lUxDe˘] (look there) or /b/ by a weak [B] in [wIvBIn] (we’ve
been).103 Brown (1977: 74–75) points out that this type of weakening is
rather common, especially with /k/ and /t/ as long as they are not initial in
an accented syllable. It is important to note that this kind of weakening may
not only be reflected in the actual acoustic quality of a phone, but also in a
weakening of the visual clues that accompany its articulation104 – some-
thing which, incidentally, considerably facilitates the dubbing of foreign
films.
All this shows that the actual form of a spoken utterance may differ
from its phonological representation in a quite substantial way.105 What
kind of implications this has for the plausibility of the phoneme concept
and for the process of understanding speech remains to be discussed.

6.2.2 Linking phenomena

The reduction of articulatory features is not the only way which character-
izes the pronunciation of words in connected speech. Rather, a kind of op-
posite phenomenon can be observed too, namely the insertion of sounds. In
English, there is a strong tendency to introduce a linking element between a
word final and a word initial vowel. This can be a vowel glide in the form
of a semi-vowel or the articulation of an [r]-sound:
to Exeter [tʊʷeksɪtə]
As far as the latter type of linking phenomenon is concerned, a termino-
logical distinction is made between a linking [r], i.e. the articulation of an
[r] which is represented in the spelling of the word but not realized in other
positions in those accents of English which are non-rhotic like RP, and an

103
Most of the examples in this chapter are taken from Brown (1977: 58 and
74), some from Gimson and Cruttenden (62001). Note also the remark on the
difficulties of transcribing weakened consonants appropriately.
104
See Brown (1977: 79–80) for detailed descriptions.
105
For the psycholinguistic implications of these facts see Garman (1990: 189):
“there is an important source of potential processing difficulty, if we believe
that speech processing proceeds on a segment-by-segment, syllable-by-
syllable basis.”
 6 Phonetic “reality” 75

intrusive [r], which refers to the insertion of an [r] in cases where no let-
ter <r> is present in the spelling.
more and more ["mɔːrənd"mɔː](linking [r])
idea of it [aɪ"dɪərəvɪt](intrusive [r])

6.2.3 Weak forms

Finally, one phenomenon ought to be mentioned in the context of con-

nected speech that is very typical of the English language, namely the fact
that with a number of highly frequent function words (articles, auxiliary
verbs, conjunctions, prepositions, pronouns) a distinction can be made be-
tween so-called strong forms and weak forms. The weak forms are usu-
ally shorter than the strong forms and show reduction of vowel length, of-
ten reduction of the vowel to /ə/ or elision of vowels or consonants.106
(1)ZD ‘And [@nd] when I’m [m] famous you can [k@n] sell it and [@n] make
your fortune.’ (weak forms)
‘So I can ["k&n].’ (strong form)
(2)ZD ‘We’re [@] living up at [@t] Higher Tregerthen, near Eagle’s Nest. Do
[dU] you know it?’ (weak forms)
‘She [SI] (weak form) does ["dVz].’ (strong form)
The difference between these weak forms and the weakening phenomena
described in 6.2.1 is that the use of weak forms is not optional in the way
that assimilation is. Assimilation and elision are processes that may or may
not happen in the articulation of a particular sound sequence by a particular
speaker. Weak forms, however, are always used in unstressed syllables; in
fact, one might describe the occurrence of strong and weak forms by saying
that it is the use of the strong form that is marked and signals particular
prominence as in situations of contrast, for example.
It has often been argued that the use of weak forms is linked with the
stress patterns of English and that it contributes to the tendency towards
isochrony that has sometimes been claimed for English.

106
For an extensive list and a description of weak forms in English cf. Gimson
(41989: 265–269).
7 Contrastive aspects of phonetics and
phonology

7.1 Levels of contrast

A contrastive analysis of two languages at the levels of phonetics and pho-

nology will be concerned with similarities and differences with respect to
the following aspects:107
− segmental phoneme and phone inventories of the two languages
− distributional properties of segmental phonemes and phones in the two
languages
− suprasegmental properties of both languages.

7.2 Phoneme and phone inventories of English and German

Comparing the phoneme system(s) of English as described in 5.2 and the

phoneme system of German as outlined in Duden Aussprachewörterbuch
(42000: 35 and 43) one will find that while English and German share a
considerable number of, in fact, most phonemes, there are a few that occur
in only one of the two languages:

consonants only English only German English and German

plosives /p//b/ /t//d/ /k//g/
108
fricatives /θ/ /ð/ /ç/ /x/ ich, doch /f//v/ /s//z/ /ʃ//ʒ//h/
109
affricates /pf/ /ts/ Pferd, Zaun /tʃ/ /dʒ/
nasals /m/ /n/ /ŋ/

107
For more detailed accounts of the sound systems of English and German see
Burgschmidt and Götz (1974: 196–213) and König and Gast (2007: 8–55)
108
For /θ/ and /ð/ in German see Duden Aussprachewörterbuch (42000: 43).
109
The Duden Aussprachewörterbuch (42000: 43) points out that the affricates
are treated as two separate phonemes in other accounts of German phonol-
ogy.
 7 Contrastive aspects of phonetics and phonology 77

lateral /l/
approxi- /r/
mant110
semi-vowels /w/ /j/111

While in the case of consonants and semi-vowels correspondences between

English and German can be established relatively easily, this is not so
straightforward in the case of vowels.112 Thus it is more appropriate to sim-
ply contrast the complete vowel systems of RP (> 5.2.2) and German
Duden Aussprachewörterbuch (42000: 35):

Monophthongs English (RP) German

oral /iː/ bee /iː/ Biene
/ɪ/ middle /ɪ/ Mitte
/eː/ Beeren
/εː/ Bären
/e/ bed /ε/ Bett
/æ/ bank /a/ Bank
/ɑː/ park /aː/ Tag
/ʌ/ hut
/ɒ/ lot /ɔ/ Tochter

110
For /r/ in German see Duden Aussprachewörterbuch (42000: 52–55).
111
/j/ is not treated as a semi-vowel in the Duden Aussprachewörterbuch (42000:
43) but listed as a fricative; Kohler (1977: 116) calls it a semi-vowel, how-
ever.
112
On the one hand, one could contrast the number of phonological oppositions
in the different vowel systems: Trudgill and Hannah (52008: 23 and 42) iden-
tify 23 vowel phonemes for RP (including the triphthongs /aI@/ and /aU@/ and
15 for US English whereas the Duden Aussprachewörterbuch (42000: 35)
lists 23 vowel phonemes for German. This includes nasalized vowels such as
/ɑ̃ː/ or /ɛ̃ː/ occurring in words such as Gourmand or Pointe, which play a mar-
ginal role in German. Thus one might argue that a purely numerical compari-
son is not particularly revealing. On the other hand, the quality of the vowel
phonemes in English and German in some cases differs considerably so that it
is difficult to decide whether English /з;/ in heard should be treated as in any
way corresponding to German /ø;/ in hören.
78 Sounds

/ɔː/ Cornwall /oː/ Sohn

/ʊ/ foot /ʊ/ kurz
/uː/ food /uː/ pusten
/ə/ about /ə/ alle
/зː/ heard /øː/ hören
/œ/ Hörnum
/yː/ Juist
/y/ Sylt
nasalized /ɛ̃ː/ Pointe
/ɑ̃ː/ Gourmand
/ɶ̃ː/ Parfum
/õː/ Garçon
Diphthongs /aɪ/ ride /ai/ reiten
/aʊ/ house /au/ Haus
/ɔɪ/ employ /ɔy/ täuschen
/eə/ there
/eɪ/ train
/ʊə/ poor
/əʊ/ low
/ɪə/ here

Apart from the fact that one should bear in mind that different analyses of
the phoneme systems of the same language may well vary as to the number
of phonemes they identify – depending, for instance, on whether particular
cases are analysed as affricates, diphthongs or triphthongs or as sequences
of phonemes – it must be clear that the phonological transcriptions chosen
by different analysts may suggest a higher degree of similarity or difference
than is actually justified. Thus the use of the symbol /æ/ for the vowel in
bank suggests a greater difference in vowel quality to German /a/ in Bank
than is perhaps appropriate with respect to recent developments in British
 7 Contrastive aspects of phonetics and phonology 79

English.113 On the other hand, the use of identical symbols in the case of
such vowels as /i;/ or /ɪ/ should not be taken to signal that the actual sounds
are identical; in fact, the English vowels tend to have a slightly more cen-
tralized and thus less extreme articulation. The same applies to the conso-
nants transcribed such as /ʃ/, whose phonetic realization in English involves
less lip rounding and less energy than in German.114 In other cases, pho-
netic differences are already made clear in the kinds of symbols chosen for
the respective phonemes, as in the case of /3;/ in heard (with no lip round-
ing) and /ø;/ in hören (with strong lip rounding).
Thus inventory differences do not only concern number and character of
phonological oppositions, but also the properties of the phones that realize
the respective phonemes. There are, of course, significant differences in the
phone inventories of two languages, which can also affect the allophones of
a phoneme to be found in a particular language. Thus, for example, while
both Standard German and RP have an /l/-phoneme, only RP contains the
dark l allophone [ł].

7.3 Rule-governed differences

A further type of difference does not concern the phoneme or phone inven-
tories of two languages, but rules which operate on the distribution of par-
ticular phonemes or phones. Thus, there are phonotactic differences of the
kind that particular clusters do not occur: for example word-initial /kn/
(knacken) is possible in German, but not in English, whereas word-initial
/sp/ (speak) or /st/ (stand) occur in English and not in German (with the
exception of some Northern German dialects). Such differences can also be
found with the distributional rules of the allophones of particular pho-
nemes. Thus, whereas the English fortis plosives are realized by a non-
aspirated (or weakly aspirated) allophone in word final position [hæt],
German fortis plosives can have aspiration in that position [hu;th].115

113
Cf. EPD15 (1997: ix): “The quality of this vowel is now more open than it
used to be, and the symbol /a/ might one day be considered preferable. We
have retained the /æ/ symbol partly because it is phonetically appropriate for
the corresponding American vowel.”
114
Cf. Arnold and Hansen (81992: 145).
115
Cf. Gimson (41989: 154–155) and Kohler (1977: 160).
80 Sounds

A very important distributional difference between English and Ger-

man116 can be found in the case of the phonemes /s/ and /z/:

word initial intervocalic word final

English /s/ seal /s/ hissing /s/ hiss
/z/ zeal /z/ rising /z/ rise
German /s/ reißen /s/ Kuss
/z/ sehen /z/ reisen

Another type of a rule-governed difference between German and English is

the phenomenon known as Auslautverhärtung in German, which results in
the neutralisation of the phonological contrast between ‘voiced’ and ‘voice-
less’ plosives at the ends of words in German. The phonotactic conse-
quence of this rule is that /b/, /d/ and /g/ do not occur in word final position:
/ta;k/ Tag /ta;ɡə/ Tage
/bo;t/ Boot /bo;tə/ Boote
In contrast to that, RP shows a tendency towards “non-release of final plo-
sives” (Gimson 41989: 157), which also means that the plosives cannot be
distinguished by phonetic voicing. Even when this happens, there is still no
neutralisation of the phonological opposition between /p t k/ and /b d g/ in
English, since the opposition is then expressed by the length of the preced-
ing vowel (> 6.1.2): /bæt/ [bæt] – /bæd/ [bæ;d]. This is not the case in Ger-
man, where Rat und Rad are homophones /ra;t/ [ra;th].

7.4 Suprasegmental differences

At the suprasegmental level, one could investigate whether particular into-

nation contours are used to express the same meanings in both languages or
not.117 Since there is no generally accepted framework for the description of

116
This applies only to those accents of German that show an /s/-/z/ contrast.
The phoneme inventories of many Southern accents of German do not pos-
sess /z/ at all.
117
Cf. Kohler (1977) and Esser (1979). For instance, in the case of wh-
questions, Esser (1979: 91) distinguishes between objective wh-questions
(“sachliche wh-Fragen”) with a falling intonation and polite wh-questions
 7 Contrastive aspects of phonetics and phonology 81

intonation (and the types of meaning expressed by different intonation pat-

terns) this is a relatively complicated task.
In more general terms, an interesting difference between German and
English at the suprasegmental level concerns the realization of pitch. In
English, changes of pitch tend to be performed during the articulation of a
stressed syllable, whereas in German they tend to happen between vowels,
so that English can be characterized as making use of pitch glides
whereas German typically shows pitch jumps (Cruttenden 1986: 54).

7.5 Pedagogical implications

A contrastive analysis may serve as a useful basis for pedagogical consid-

erations in the teaching of pronunciation. Although it cannot necessarily
serve to predict errors, it can be instrumental in explaining common errors.
Thus, the common phenomenon that German native speakers substitute
such phonemes as /θ/ or /w/ by /s/ or /v/ can quite easily be explained by
the absence of these phonemes from the German phoneme system. Simi-
larly, a mispronunciation of /kæt/ [khæt] as [khæth] can be traced back to the
different distributional rules of the allophones of plosives in English and
German, and the pronunciation of /bæd/ as [bæt] to the application of the
German rule of Auslautverhärtung to English.
A question that is particularly important in this context is to what extent
the distinction between the phonological and the phonetic levels of analysis
is relevant to the teaching or learning of pronunciation. Initially, there is
some logic behind the argument that phonological errors may result in mis-
understanding (such as when [bet] could represent /bet/, /bed/ or even /bæt/
or /bæd/) and that preventing such errors should be a prime aim in the
teaching of pronunciation118 and that allophonic errors (such as [beth] for
/bet/ or [bɔ;l] for [bɔ;ł]) are relatively unimportant in comparison. At sec-
ond sight, this argument must be treated with some caution. Firstly, the
number of contexts in which phonological errors would actually result in
misunderstandings is probably relatively limited. Secondly, it is to be
doubted that listeners perceive speech so systematically that they find a
clear l instead of a dark l less disturbing than a voiceless consonant instead
____________________________
(“höfliche wh-Fragen”) with a rising intonation. Echo questions are character-
ized by Esser as having rising intonation in English, too. Compare Kohler
(1977: 127).
118
Cf. Arnold and Hansen (1992: 52).
82 Sounds

of a voiced one. Thirdly, and most importantly, in the light of the reserva-
tions made in 6.1 it must be doubted whether the phoneme concept as such
provides a suitable basis for pedagogical considerations. Since the length of
the preceding vowel has been shown to be much more important to distin-
guish between word final plosives than the quality of voicing or non-
voicing in the plosives themselves, this ought to be reflected in the teaching
of pronunciation. In a different way, the convention of using /æ/ as a sym-
bol for the vowel in bank should not obscure the fact that in many British
accents (amongst them more recent forms of RP or BBC English) the pho-
netic realization of this phoneme comes very close to that of German /a/.119
Such facts underline the importance of a phonetic (rather than a phonologi-
cal) basis in the teaching of pronunciation.
A sensible starting point for weighing the goals of teaching pronuncia-
tion seems to be a perceptual approach, which does not focus on the differ-
ences between language systems, but on those features of a foreign accent
that native speakers might find disturbing. In this respect, pronouncing bad
with a short vowel might be more irritating than the voicelessness of the
final consonant. From a perceptual point of view, however, it is especially
suprasegmental aspects of pronunciation that will have to be considered.
This means that factors such as pitch glides, rhythm (which coincides with
the appropriate use of weak forms) or linking phenomena will have to be
attributed high importance. An experiment carried out by Dretzke (1985)
showed, for example, that English native speakers find the use of glottal
stops before word initial vowels (corresponding to a rule of German) ex-
tremely irritating.120

Recommended further reading:

− Gimson and Cruttenden (62001)

− Lass (1984)
− Gut (2009)
− Brown (1977)
− Burgschmidt und Götz (1974), König and Gast (2007)

119
For the question of /&/ and /a/ see also EPD17: ix.
120
For a justification of a perception-oriented approach towards the teaching of
pronunciation see Dretzke (1985) or Herbst (1992). For the importance of
pronunciation in the teaching of a foreign language see also Herbst (1994).
 Meaning-carrying units

8 Morphology

8.1 The concept of the morpheme

Chapter 8, 9 and 10 will be concerned with questions such as

− how words can be analysed into smaller meaningful units (morphology)
− how new words can be formed (word formation) and
− how meanings are expressed in multi-word units (phraseology).
The discussion of these questions will be based on the following text – a
short passage taken from a catalogue of an exhibition of work by the
painter Patrick Heron:121

(1) When he began to work again, he seemed to have listened to those 1

who had urged him to get ‘back in touch with the earth’. Justifiably, 2
Heron had always been irritated by those literal-minded critics who 3
insisted in seeking references to the Cornish landscape (which he 4
loved, and fought tirelessly to protect) in the shapes and contours of 5
his paintings. Nonetheless, as he would admit in his more philosophi- 6
cal moments, the power of the landscape to ‘transmit certain rhythms 7
straight up through the soles of my shoes every minute of the day’ was 8
unquestionable, and he frequently drew on it for the language he used 9
to describe the familiar elements of his compositions – the harbour 10
shapes, the boulder-like discs, the coastlines of colour-shapes. Now, 11
for the first time in quarter of a century, he picked up pencil and pa- 12
per and drew from the landscape. … Heron had decided, as he char- 13
acteristically put it, ‘to bugger things up and try something crazy’; 14
completing White, Pink and Scarlet : July 1983 (cat.18) he found he 15
could no longer bring himself to fill in its final unpainted shape and 16
scribbled across it with his brush. … 17

121
Source: James Beechey: Patrick Heron. Paintings 1970–1984, 25 February –
20 March 2004, Waddington Galleries London.
84 Meaning-carrying units

‘Looking at something – anything’, Heron once declared, ‘is more 18

interesting than doing anything else, ever, as a matter of fact.’ 19

After identifying the units that can distinguish meaning in a language (the
phonemes), the next step in a structuralist analysis is to identify the ele-
ments that carry meaning. These are not necessarily identical with words.
Thus the last clause of (1) could be divided up into the following units car-
rying meaning:
look | ing | at | some | thing | any | thing | Heron | once | declare | d | is | more |
interest | ing | than | do | ing | any | thing | else | as | a | matter | of | fact
In analogy to the terminology employed in phonology, such units can be
called morphs.122 As in phonology, where phones that show a great deal of
phonetic similarity and do not distinguish meaning are classified as allo-
phones of one phoneme, a distinction can be made between morphemes
and allomorphs in morphology:

A morpheme is the smallest abstract linguistic unit that serves to carry
meaning.123

An allomorph is one of several possible phonological realizations of a
morpheme.
For instance, in (1) declared was split up into two morphs, /dI"kle@/ and /d/
in the phonological representation. There are two further morphs in (1)
carrying the same meaning as /d/, ‘past tense’, which are rather similar in
form to /d/:
‘past tense’
/pIkt/ /t/ picked
/si;md/ /d/ declared, seemed, used, scribbled
/dI"saIdId/ /ɪd/ decided, insisted

122
Cf. Lyons (1968: 180–187).
123
Cf. the definition provided by Hockett (1958/2000: 394): “Morphemes are
the smallest individually meaningful elements in the utterances of a lan-
guage.” Bloomfield (1933/1935: 161) defines the morpheme as “a linguistic
form which bears no partial phonetic-semantic resemblance to any other
form.” See Anderson (1988: 151). Compare Lipka (32002: 85) for a critical
appraisal and interpretation of the term morpheme.
 8 Morphology 85

The phonological representations /t/, /d/ and /ɪd/ are thus called allomorphs
of the past tense morpheme {D}.124
A further parallel between phonology and morphology is that the allo-
morphs of a morpheme can be in complementary distribution. In the case of
/t/, /d/ and /ɪd/ the distributional rules can be stated as follows:

{D} ‘past tense’ examples

/t/ after voiceless consonants except /t/ worked, picked
/d/ after vowels allowed
after voiced consonants except /d/ seemed, loved, used, scribbled
/Id/ after /t, d/ insisted, decided

In very much the same way, /s/, /z/ and /Iz/ can be established as allo-
morphs of a plural morpheme {S} in words such as critics, paintings and
references (> 8.3.4).125

8.2 Types of morpheme

According to whether they can occur as words on their own or not, mor-
phemes can be classified as free morphemes or bound morphemes.
Thus, {when}, {he}, {work} or {again} are free morphemes, whereas {S}
‘plural’ or {D} ‘past tense’ are bound morphemes. A further, though less
clear-cut, distinction that is sometimes made is that between lexical mor-
phemes, which express lexical content which is not primarily of a gram-
matical nature, and grammatical or functional morphemes, 126 which

124
Curly brackets are used for morphemes.
125
/s/, /z/ and /Iz/ also occur as allomorphs of a morpheme expressing third per-
son singular with verbs (goes) and a morpheme expressing genitive singular
with nouns (University’s). For the rules of distribution of the allomorphs in
these cases see Schmid (2005: 63).
126
See Lipka (32002: 86) for the distinction, who amongst other differences
mentions as distinguishing criteria that lexical morphemes “denote (particu-
lar) extralinguistic objects & states of affairs: e.g. actions, events, situations,
relations” and belong to an open class, whereas grammatical morphemes “de-
note (general) grammatical functions, e.g. plural, tense. Syntactic relations:
e.g. concord of gender, number” and belong to a closed class. Compare also
Kastovsky (1982: 71–72) and Hansen et al. (1985: 15). See also Croft
86 Meaning-carrying units

serve to form particular grammatical forms or express grammatical rela-

tions:127
listed {list} (free, lexical) {D} (bound, grammatical)
composition {compose} (free, lexical) {-ition} (bound, lexical)
boulder-like {boulder} (bound, lexical) {like} (free, lexical)
The distinction between grammatical and lexical morphemes is relevant
with respect to two different types of morphological process: derivation and
inflection. Derivation is one way of forming new words, whereas inflection
distinguishes different grammatical forms of the same word. Inflectional
morphemes (or inflectional suffixes) are thus bound grammatical mor-
phemes, whereas derivational morphemes (or derivational affixes) are
bound lexical morphemes.128
The morpheme or combination of morphemes to which a derivational or
inflectional morpheme is attached is called the base. Thus in the case of
coastlines (line 11), the base is formed by coastline, which consists of two
free lexical morphemes, and questionable, which consists of the free lexical
morpheme question and the derivational suffix -able, is the base of unques-
tionable (line 9).
While a base can consist of several morphemes, the term root is used
for single morphemes that are not affixes such as question in unquestion-
able or coast and line in coastlines.129 Most roots in English consist of free

____________________________
(2000). Compare the list of criteria given by Plag, Braun, Lappe and Schramm
(2007: 88) to distinguish between inflectional and derivational affixes.
127
See Stein (2007) for a very useful account of English affixes.
128
As a further type of bound lexical morpheme one can identify so-called
bound roots (see below).
129
See Plag (2003: 11) and Schmid (2005: 48–49). Note that these terms are
notoriously difficult to define if one wants to avoid circularity of definitions.
For the distinction between root, stem and base see e.g. Bauer (1983: 20–21):
“A root is a form which is not further analysable, either in terms of deriva-
tional or inflectional morphology. It is that part of a word-form that remains
when all inflectional and derivational affixes have been removed. … A stem
is of concern only when dealing with inflectional morphology. It may be –
but need not be – complex … it is that part of the word-form which remains
when all inflectional affixes have been removed. … A base is any form to
which affixes of any kind can be added.” These definitions depend on the
definition of affixes as “bound morphs which do NOT realize unanalysable
lexemes” (Bauer 1983: 18).
 8 Morphology 87

morphemes, but there is also a case to be made out for bound roots130 such
as rasp- in raspberry and possibly Corn- in Cornish (line 4).

8.3 Problems of a static morpheme concept

8.3.1 The problem

The framework of morphology as outlined so far proves perfectly suitable

for the analysis of the examples discussed in the previous section. What
makes the concept of the morpheme rather attractive is the idea that one can
split up words into different segments with identifiable meanings. So a
word such as picked can be analysed nicely as consisting of a free lexical
morpheme {pick} and a bound grammatical morpheme {D} realized by the
allomorph /t/. Other words in text (1) turn out to be much more problem-
atic:131 in particular, began (line 1), fought (line 5), drew (line 9), found
(line 15) and also put (line 14) do not lend themselves to a straightforward
formal analysis into {begin}, {fight} etc. plus an allomorph of {D}. Simi-
larly, those (line 1) cannot easily be analysed as {that} + {S} etc. There are
different ways of accounting for these phenomena, all of which, however,
throw some doubt on the overall validity of the morpheme concept.132

8.3.2 Portmanteau morphs

One concept that has been suggested to account for the problem of forms
such as had, began or fought is that of a so-called portmanteau morph, i.e.
a morph that represents two different morphemes – {begin} and {past
tense} in the case of began. Similarly, teeth or those could be analysed as
morphs combining {tooth} or {that} with the plural morpheme.133

130
See Plag (2003: 10) for examples such as circul- (in circulate etc.) or hap- (in
hapless).
131
For a critical discussion of the morpheme see Anderson (1988: 153–154).
132
For a discussion of various problems of the traditional morpheme concept see
Palmer (1971: 110–124). A very important contribution to this discussion
can be seen in Hockett’s (1954) distinction between item and arrangement
and item and process. For this see Matthews (1974: 225–227), Anderson
(1988: 157–162) or Vennemann and Jakobs (1982/2000: 409–411).
133
For critical remarks on this concept introduced by Hockett see Palmer (1971:
116).
88 Meaning-carrying units

An alternative approach would be to treat the relationship between begin

and began or fight and fought in terms of a vowel change. Strictly speaking,
this is not in line with the definition of morphemes and allomorphs given
above because accounting for such relations in terms of processes “seri-
ously weakens the idea that the morpheme is a minimal sign” (Plag 2003:
23).134
A similar problem is presented by forms such as was (line 8) or went,
which bear no formal resemblance to forms of the “same” verbs such as be,
am, are and go or gone respectively. Schmid (2005: 43) accounts for such
cases in terms of suppletion.

8.3.3 Zero-morphs

The case of put is even more intriguing since there are good reasons for
saying that put in line 14 expresses the meaning of ‘past tense’ although
this is not explicitly marked (apart from the absence of third person singu-
lar {S}). The fact that some verbs in English such as put or cut have identi-
cal present tense and past tense forms (with the exception of the third per-
son singular) or some nouns identical singular and plural forms such as
sheep has given rise to the idea of introducing the concept of zero-morph
or zero-morpheme. The argument justifying such an analysis is the obvi-
ous analogy between examples of the following kind:
present tense past tense singular plural
{look} {look} + {D} {dog} {dog} + {S}
{put} {put} + {Ø} {sheep} {sheep} + {Ø}
Palmer (1971: 115) points out that there is a conflict between the concept
of a zero-element and the definition of the morpheme given by Bloom-
field:135

134
Cf. Palmer’s (1971: 116) critique of such approaches: “… it is immediately
obvious that these are very strange allomorphs; an instruction to replace one
item by another can hardly be regarded as in any sense consisting of pho-
nemes.” See also Matthews (1974: 122): “A process of replacement is no
more a ‘morph’ than zero is a ‘morph’”. Compare the discussion of this prob-
lem by Matthews (1974: 118–122).
135
Cf. Palmer (1971: 111 and 115): “It was thought reasonable that morphemes
could have zero allomorphs, though with the condition that not all the allo-
morphs can be zero. This is an important condition since otherwise we could
 8 Morphology 89

The notion of the zero allomorph in sheep and also in hit (past tense) is a
very useful one, but once again we are moving away from Bloomfield’s
conception. A zero element cannot really be said to have ‘no partial pho-
netic-semantic resemblance to any other form’.
Similar arguments against the concept of a zero-morph have been put for-
ward by Matthews (1974: 117–118):
The PLURAL morpheme of men and the PAST PARTICIPLE of come can
no longer be defined as ‘classes of allomorphs in complementary distribu-
tion’ – the reason being, quite simply, that one cannot examine one’s data
and determine the ‘distribution’ of ‘zero’. One cannot say that in some
forms ‘nothing’ is ‘there’ but in others it is not ‘there’, that the ‘presence’
of ‘nothing’ in one form ‘contrasts’ with the ‘absence’ of ‘nothing’ in an-
other, and so on. …
But let us turn to a more modern conception of the morpheme. Syntacti-
cally, the sailed of I have sailed is SAIL + PAST PARTICIPLE; we are
therefore justified in saying that the come of I have come is syntactically
COME + PAST PARTICIPLE. Our problem is merely to specify the word-
form, come, by which COME + PAST PARTICIPLE is represented or real-
ised. Can we not therefore say, taking it morpheme by morpheme, that
come is represented by its normal alternant come and PAST PARTICIPLE,
quite simply, has no representation or realisation at all? This is surely quite
coherent; there is all the difference in the world between saying that a
grammatical element ‘has zero realisation’ and saying that a word-form
‘contains a zero allomorph’.

8.3.4 Morphological and phonological conditioning

The discussion in the previous sections has shown that a category such as
‘past tense’ is expressed in different ways in English. The same is true of
the category ‘plural’ with nouns:
− plurals can be formed by adding inflectional suffixes (critic – critics,
reference – references but also ox – oxen),
____________________________
say that CAT has a zero allomorph in the singular. (In itself, this suggestion is
not altogether ridiculous: we might say it has zero in the singular and s in the
plural; but we have then established a morpheme (zero) where there is no lin-
guistic form anywhere in the language since English never has a formal mark
in the singular.)” Looking at words such as formula, analysis or curriculum,
even this statement might be subject to discussion.
90 Meaning-carrying units

− in the case of tooth – teeth, foot – feet the plural forms are distinguished
from the singular forms by a vowel change and can thus be analysed as
portmanteaux,
− children involves a vowel change and a suffix,136
− in sheep – sheep singular and plural are not distinguished by morpho-
logical means (or, if you prefer, by Ø) etc.
If we want to describe when these different forms occur, we are faced with
the problem that generalizable rules for the distribution of these forms can
only be given for the allomorphs /s/, /z/ and /ɪz/ – or in the case of the past
tense /t/, /d/ and /ɪd/. The occurrence of these allomorphs is seen as being
phonologically conditioned because the occurrence of a particular allo-
morph depends on the preceding phoneme:

{S} ‘plural’
/s/ after voiceless consonants except /s, ʃ, tʃ/ critics, shapes, moments,
elements, discs
/z/ after vowels contours, shoes
after voiced consonants except /z, ʒ, dʒ/ paintings, rhythms, soles,
compositions, coastlines
/ɪz/ after /s, z, ʃ, ʒ, tʃ, dʒ/ references

However, whether a noun takes an {S}-plural at all (or whether, corre-

spondingly, a verbs takes a {D}-past tense) is not dependent on the phono-
logical environment, but must be seen as a property of the individual word,
hence this is a matter of lexical conditioning. The term morphological
conditioning is sometimes also used in this sense, sometimes it is restricted
to cases where a particular derivational morpheme requires the presence of
a particular allomorph as in:
compose /k@m"p@Uz/ /k@mp@s/ + /IS@n/ composition
declare /dI"kle@r/ /dekl@r/ + /eIS@n/ declaration
refer /rI"f3;/ /"ref@r/ + /@nts/ reference

136
For the notion of “empty morph” in words such as children see Palmer (1971:
115).
 8 Morphology 91

8.4 Inflectional morphology: historical background

Although it is not relevant to a strictly synchronic analysis of present-day

English, it is interesting to see that quite a number of the aspects of present-
day English inflectional morphology that cause problems for a synchronic
analysis have their causes in earlier stages of the language and subsequent
historical developments.137
For instance, in the description of Old English verb morphology a dis-
tinction can be made between different classes of so-called weak verbs,
which form the past tense by a {D}-suffix and strong verbs, which show
a regular vowel change in their paradigm known as ablaut. According to
different types of gradation series, seven classes of strong verbs can be
distinguished, for example:138

present tense preterite preterite past participle

1/3 sg pl and 2 sg
I rīsan rās rison -risen rise
II scēōtan scēāt scuton -scoten shoot
III sincan sanc suncon -suncen sink
IV cuman cōm cōmon -cumen come
V sprecan spræc sprǣcon -sprecen speak
VI standan stōd stōdon -standen stand
VII lǣtan lēt lēton -lǣten let

Without wanting to go into the details of the historical development here, it

is easy to see that some of the vowel changes to be observed in present-day
English verbs (rise – rose – risen, sink – sank – sunk etc.) can be related to
the gradation series of different classes of strong verbs in Old English. Oth-

137
For example, the morphological irregularity of plural forms such as feet, teeth
or mice historically goes back to i-mutation (> 20.2.3.1), which leads to a
vowel change in the dative singular and nominative and accusative plural
forms in Old English (fōt – fēt).
138
The following examples are taken from Quirk and Wrenn (21957). The vari-
ous classes show considerable variation, which is not indicated here. Class
VII verbs showed reduplication in Germanic; cf. Hogg (1992: 156).
92 Meaning-carrying units

ers (such as meet – met – met), however, are caused by certain phonological
developments (> 20.2).
Since quite a number of Old English strong verbs have become weak or
regular and since a number of weak verbs have become irregular, the sys-
tem has become even more obscured. Thus, there is little justification for
talking about strong or weak verbs in present-day English.

8.5 Further problems of morphological analysis

The weaknesses of the morpheme concept outlined above concerned

mainly inflectional morphology. However, trying to split up words into
morphemes can cause a number of problems, too, particularly concerning
the question of into how many morphemes a particular word should be
analysed. Thus it is probably relatively uncontroversial to analyse degree,
colour or whole as representing a single morpheme each, boredom as being
made up of two, {bore} and {-dom} and unfilled as consisting of the three
morphemes {un-}, {fill} and {D}. Even applications and molecular are
relatively unproblematic if one postulates two allomorphs to account for the
pronunciation differences between apply /@"plaɪ/ and application /@plɪ/ and
molecule /"mɒlIkju;l/ and molecular /mə"lekjʊl@/ respectively. In the same
way it is possible to relate competitive /kəm"petItIv/ to compete /k@m"pi;t/.
Much more problematical is a synchronic analysis of words such as
suppose, ensure or require, where it seems debatable whether an analysis in
terms of one or two morphemes is preferable. On the one hand, one might
argue that these elements are able to form series of a certain type, as in:
sup- suppose, support, suppress
-pose expose, impose, propose
en- ensure, enforce, enfranchise, engage
-sure ensure, insure, assure
re- require, resist, retain
-quire require, acquire, enquire, inquire
On the other hand, one might doubt whether these elements carry any iden-
tifiable meaning in English, which would be the same for all the words in a
series. Furthermore, one might ask to what extent native speakers of Eng-
 8 Morphology 93

lish would consider such words as being made up of two morphemes.139

The problem posed by such words can be accounted for by considering the
historical background. In fact, all of these words are not of Germanic but of
Romance origin. While these forms were morphologically transparent in
Latin (e.g. expōnere, impōnere, prōpōnere), this transparency has become
obscured in English. In the light of de Saussure’s postulate that synchronic
and diachronic analyses should be kept apart, such historical considerations
cannot contribute to solving the analytical problem at the synchronic level.
Since one would probably want to argue that native speakers of English
probably are not aware of the historical origins of these words (at least not
when they first encounter them in the process of language acquisition),140 it
makes sense to look for a purely synchronic solution anyway. In this par-
ticular case, however, it is probably fair to say that arguments can be found
for either analysis.141 Coates (1987: 113), discussing the analysis of words
such as dissociate and associate,142 comes to a very similar conclusion:

139
One could consider analysing elements such as -quire in terms of bound root
morphemes. A related case is presented by so-called neo-classical compounds
such as democracy, microscope or photograph, which Schmid (2005: 42)
analyses in terms of “combining forms” (demo- + -cracy).
140
To what extent native speakers of English would consider such words related
probably also depends on the degree of their own language awareness and to
the extent to which they have been exposed to a classical education.
141
See also Marchand (21969: 5) or Lipka (32002: 97).
142
Coates (1987: 113) discusses the term formatives for recurrent patterns with-
out a clear semantic status in this context: “On the grounds that dis- is rele-
vantly and independently meaningful (cf. trust vs distrust), we could analyse
DISSOCIATE as DIS-SOCIATE, and by virtue of that analysis treat –
SOCIATE as a lexical element or formative recurring in ASSOCIATE and
therefore AS- as a formative too. Here we have to be most careful of the cri-
teria we use in our analysis. It is all too easy to let historical justifications
support us; something similar to the above analysis is good for the Latin an-
cestors of the words in question, which is no sort of an argument about Eng-
lish. We must not be bemused by patterns, for there is prima-facie evidence
that many real English speakers have not analysed the given forms in the way
shown. The form /dIs@"s@UsieIt/ is now frequently heard, demonstrating that
whilst the DIS-/AS-alternation may be perceivable it is not necessarily per-
ceived. ASSOCIATE is clearly used as an unanalysed whole in this non-
standard (but spreading) form. …”
94 Meaning-carrying units

What, then, would be the status of … AS- in a definitive analysis of Eng-

lish? Is there indeed to be a boundary between it and –SOCIATE? Answer:
for some, perhaps; therefore no definitive answer is possible for a language
as opposed to a speaker.
A related problem is presented by words such as blackberry, cranberry or
raspberry. While blackberry can obviously (or perhaps not quite so obvi-
ously since it is arguable whether black in blackberry means ‘black’) be
analysed as being made up of {black} and {berry}, a corresponding analy-
sis of cranberry or raspberry into two morphemes is problematical since
cran- or rasp- only occur in the words cranberry or raspberry. On the other
hand, an analysis of these words as consisting of one morpheme would
obscure the fact that they contain the morpheme {berry}, which for both
formal and semantic reasons would appear highly inappropriate. Such ele-
ments as cran- or rasp-, which occur only once in the language, are thus
often treated as a special class of morphemes, called unique morphemes,
blocked morphemes 143 or also cranberry morphemes.
Even if the introduction of the category of unique morphemes may be
seen as solving that particular problem one can probably argue, at least
from a sceptic’s point of view, that quite a few analytical problems remain,
especially when one believes that all occurrences of a particular morpheme
should express (at least roughly) the same meaning. Whereas in the case of
blackberry and blackbird, the analysis in terms of a morpheme {black}
seems plausible enough, a blackberry is not just a ‘black berry’ and a
blackbird not a ‘black bird’. Similarly, a word such as soap opera, which at
first sight easily lends itself to an analysis into two morphemes, turns out to
be rather problematical when one considers the meanings of the constitu-
ents. Although such phenomena can be accounted for within a theory of
word formation (> 9), they also throw light on the slightly problematical
nature of the notion of morpheme.144

143
See, for example, Kastovsky (1982: 70) or Lipka (32002: 87).
144
Compare the gradient established for typical and less typical members of the
category morpheme (prototypical morpheme – bound root – combining form
etc.) established by Schmid (2005: 41).
9 Word formation

9.1 Words

9.1.1 Words and lexemes

It is one of the ironical facts about language that in everyday discussions of

language the term “word” plays a central role. In fact one can safely assume
that it is one of the few terms an average speaker of a language will know
with respect to talking about language. People ask what a particular word
means, how a word is spelt, possibly whether an expression is spelt as one
or two words, how many words a text contains, how a word translates into
another language etc. Even in more professional contexts of looking at
language one will frequently refer to words: the parts of speech are often
referred to as word classes (> 11.4), dictionaries claim to contain so and so
many words, and didacticians outline how many words a learner of a for-
eign language should know at a particular stage of learning. The irony is
that linguists are slightly wary about the term and would be very hesitant to
use it without explaining what they mean by it beforehand.145
The problem can be illustrated by a short passage taken from David
Lodge’s novel Thinks:146

(1) […] I had three hours of crucifying boredom at the Senate working 1
party on interfaculty modular compatibility, meaning we’re supposed 2
to come up with a formula to ensure that a course in say the School of 3
Community Studies requires as much work and deserves the same 4
weight as a course in say Electrical Engineering, so that the Univer- 5
sity can put its new degree in Interdisciplinary Studies on the market . 6
. . our Pick’n’Mix degree as the Registrar likes to refer to it, which is 7
supposed to give us a competitive edge in the annual applications 8
bazaar and save the University’s fortunes . . . Market research has 9
apparently established that there is an unfilled niche for a degree 10

145
For a discussion of the term “word” see e.g. Matthews (1974: 20–26) or Mair
(1997: 22).
146
Source: David Lodge: Thinks (2001; Penguin edition 2002: 180–181): Har-
mondsworth: Penguin.
96 Meaning-carrying units

course that would allow the student to combine courses across the 11
whole university curriculum, nuclear physics with soap opera analy- 12
sis, molecular biology with medieval mystery plays . . . I daresay it 13
looks quite enticing on paper, especially the glossy, colour-illustrated 14
paper of the University’s brochure, but some of these subjects are 15
harder than others and most of them can’t be studied properly in 16
isolation, but I didn’t make myself any friends this afternoon by point- 17
ing this out. 18

The question how many words this text contains is by no means easy to
answer. While my computer program counts 198 words, I count 191.147
With some justification, one could also arrive at a figure of 177 (or even
174), if one takes working party, come up, Community Studies, Electrical
Engineering, market research, degree course, soap opera, pointing out as
representing one word rather than two (or three in cases such as Senate
working party or come up with). Furthermore, if one interprets the question
in a way so that it does not refer to word tokens, but to types, this reduces
the figure to something like 135: then there are 12 occurrences of the, 6
occurrences of to, 5 occurrences each of of, in and a etc. But even there, it
is impossible to arrive at a precise figure: is an to be regarded as a different
word from a or not? If so, what about the in line 1 [ðə] and the in line 9
[ðɪ]? Whether this in this afternoon (line 17) is the same word as the this in
line 18 may depend on whether one recognizes a separate word class de-
terminer or not; it may also be seen independently from that question if you
argue that a word can belong to different word classes. More importantly,
however, should University and University’s be considered instances of the
same word? And what about we and us? Or our? There are probably good
reasons supporting any decision regarding these questions depending on the
purposes of the analysis. What this otherwise relatively insignificant word
count shows, however, is that a number of terms are needed for the entities
that in everyday language can be referred to as words, otherwise there is
bound to be considerable confusion. For the purposes of linguistic analysis,
it seems necessary to restrict the readings of the term word. Much of the
discussion has centered around the definition of the word as a ‘minimum
free form’ by the American structuralist Leonard Bloomfield (1933/1935:

147
Microsoft Word counts … as one word and ... as three words; we’re as one
word.
 9 Word formation 97

178), which, however, is not totally sufficient.148 Thus, Cruse (1986: 35–
36) mentions two characteristics of the word.149
The first is that a word is typically the smallest element of a sentence which
has positional mobility – that is the smallest that can be moved around with-
out destroying the grammaticality of the sentence (ignoring any semantic
effects).
John saw Bill.
Bill saw John.
Bill, John saw.
By no means all words are equally mobile in this sense, but with very few
exceptions, the smallest mobile units are words. …
The second major characteristic is that they are typically the largest
units which resist ‘interruption’ by the insertion of new material between
their constituent parts.
Thus, in a sentence such as (2), no insertions are possible within the units
separated by empty spaces, whereas insertions in those spaces seem per-
fectly feasible:
(2) Some of these subjects are harder than others

  
academic much many
This is the reading of the term “word” in which text (1) above has 191
words. However, when it comes to talking about “different words” it seems
sensible to introduce terminology to be able to express the “sameness” and
the “differentness” of the words establish, establishes, establishing, estab-
lished or university, university’s and universities. This can be done by say-
ing that university, university’s and universities are different word forms,
which can be seen as representations of the same lexeme. Matthews (1974:
22) describes the lexeme as a “fundamental unit (compare other terms in ‘-
eme’ such as ‘phoneme’ or ‘morpheme’) of the lexicon of the language”.
Thus the lexeme is an abstract unit of the language. The concept is an ex-
tremely important one since it is clear that, for instance, when one looks up
a “word” such as university in a dictionary, one is neither interested at all in
the particular token of the word printed in the dictionary (in the this-text-
consists-of-191-words sense) nor just in the particular word form university

148
See, for instance, Palmer (1971: 50).
149
See also Burgschmidt (1978: 27).
98 Meaning-carrying units

but in the lexeme with all its formal and semantic properties (spelling, pro-
nunciation, meaning etc.). Furthermore, the term lexeme is of great use in
that it enables one to treat expressions such as come up with that consist of
several words, but have one sense, as so-called multi-word lexemes. (>
10.4).
A less established, but equally useful distinction is that between lexemes
and lexical units, which refers to the semantic or content side of the lex-
eme and bears witness to the fact that most words (or lexemes) have more
than one meaning or sense. Cruse (1986: 80) thus describes the lexeme in
the following way: “a lexeme is a family of lexical units; a lexical unit is
the union of a single sense with a lexical form; a lexical form is an abstrac-
tion from a set of word forms (or alternatively – it is a family of word
forms) which differ only in respect of inflections”. A lexeme, then, can be
seen as an abstract unit representing one or more word forms and one or
more lexical units.

9.1.2 New words

One of the obvious functions of the lexemes of a language is to enable the

speakers of a language to express certain concepts or ideas. Since the world
is changing constantly, both in terms of real objects and of ideas, languages
must be made to accommodate new concepts. This can be done, as Leech
(21981: 30) points out, by “lexical innovation, which may take the form of
NEOLOGISM (the invention of new words, or more precisely lexical items
…) or TRANSFER OF MEANING (the derivation of new senses of estab-
lished words)”. Both principles can be explained referring to the previous
paragraph, in which the technical terms lexeme and lexicon were used. The
former is an example of a neologism, a word which was coined, in this case
on the basis of existing language material, to express a certain concept,
whereas in the case of lexicon, a word that has existed in the English lan-
guage since the seventeenth century, has been used with a new sense spe-
cific to linguistics. This kind of process can also be related to the influence
of another language. This is what happens, for instance, when German
words take over meanings of English words which are similar in form: kon-
trollieren came to be used to mean ‘have power over’ alongside ‘check’ or,
more recently, meinen is increasingly being used in the sense of words
meaning something rather than someone having a certain opinion. It is ob-
vious that the advantage of both types of lexical innovation is to express
 9 Word formation 99

concepts in a more condensed and less circumspect way. Leech (21981: 31)
argues “that combined with this abbreviatory function, the word as a lexical
element has a concept-defining role”, which he illustrates with examples
such as defenestration or the use of new adjective compounds such as
ready-to-eat in advertising.150
There are several possibilities speakers have for the creation of new
words. They can
− use lexical material of their own language and combine or modify it in
some way,
− take over words from another language (which may be integrated and
modified to various degrees),
− invent new words without making use of the first two techniques.
The invention of words in this latter sense seems to be a relatively rare
phenomenon, in English at least, but it does happen: the brand name Kodak
and the technical term quark are relatively well-known examples. Lexical
borrowing (checken, talken and fighten are recent examples from German)
is a common language contact phenomenon and as such is studied widely
in the context of historical linguistics (> 20.3), while the combination or
modification of lexical material of the same language as in soap opera,
heliport, smog or temp is studied under the headings of morphology and
word formation.151
It is, of course, difficult to get precise figures as to which of these meth-
ods is employed how often. An estimation carried out by Bauer (1994: 35)
suggests that between 1880 and 1982 75% of the new words in English are
based on English, 1.6% are of unknown origin and the rest have other lan-
guages as their source (4.3% Latin, 4.0% French, 3.0% Greek, 2.5% Ger-
man etc.).152

150
Accepting Halliday’s (1970b: 141) view that “the nature of language is
closely related to the demands that we make on it”, the fact that languages are
constantly subject to lexical innovation might be taken as argument in support
of Leech’s claim. See also Schmid (2008).
151
The two latter phenomena are interrelated, however, in that the influence of a
donor language may result in new word material being taken over in the re-
ceptor language or in the formation of new words (or transfer of meaning) in
the receptor language under the influence of the donor language. Cf.
Kastovsky (1992: 299–300).
152
Bauer (1994: 34) states a “decrease in the number of loans during the twenti-
eth century”. Compare also the figures given by Ayto (1996: 184).
100 Meaning-carrying units

9.2 Word formation

9.2.1 Introduction

Word formation (or, more precisely, lexeme formation153) is concerned

with one of the most creative aspects of language. The field of word forma-
tion can be seen as part of morphology in the wider sense – thus, as pointed
out above, excluding the inflectional properties of a word. It was defined by
Marchand (21969: 2) in his influential book Categories and Types of Eng-
lish Word-Formation as “that branch of the science of language which stud-
ies the patterns on which a language forms new lexical units, i.e. words.”
In a more speaker-oriented view of language, it would probably be more
appropriate to say that word formation studies the patterns according to
which the speakers of a language can form new words; Lipka (2002: 95)
thus stresses a distinction between “the analytic aspect of word-formation,
i.e. the perspective of the hearer or reader who encounters already existing
complex lexemes” and “the synthetic aspect, namely the perspective of the
speaker/writer who creatively produces a new lexeme”. This is a particu-
larly important distinction because it is essential in any discussion of word
formation to be clear about what one intends to explain. There are at least
three different phenomena that one could try to describe:
− the inventory of word formations of a language at a particular point in
time and the rules and mechanisms that have made these word forma-
tions possible (the study of word formation in this sense necessarily in-
volves a diachronic component),
− the ability of native speakers to understand and interpret the word for-
mations they encounter, i.e. the analytic aspect,
− the ability of native speakers to create new lexemes, i.e. the synthetic
aspect.
The three phenomena are quite obviously related, but they are not identical,
as will become apparent from the discussion below. It is also clear that the
term word formation is ambiguous in that it refers to the processes by
which new words can be created and to the result of such a process.154 The
fact that word formations make up a considerable proportion of probably

153
Cf. Lipka (32002: 86).
154
See Burgschmidt’s (1973: 1) distinction between Prozeß and vorfindlicher
Zustand and Kastovsky’s (1982: 155) remarks on “Wortbildung als Prozeß
und Inventar”.
 9 Word formation 101

any randomly chosen text of English can be seen from the following pas-
sage taken from A.S. Byatt’s novel Possession:155

(3) When his DES grant ran out, Val became the breadwinner, whilst he 19
finished his PhD. She acquired an IBM golfball typewriter and did 20
academic typing at home in the evenings and various well-paid temp- 21
ing jobs during the day. She worked in the city and in teaching hospi- 22
tals, in shipping firms and art galleries. She would not be drawn out to 23
talk about her work, to which she almost never referred without the 24
adjective ‘menial’. ‘I must do a few more menial things before I go to 25
bed’ or, more oddly, ‘I was nearly run over on my menial way this 26
morning’. Her voice acquired a jeering note, not unfamiliar to Roland, 27
who wondered for the first time what his mother had been like before 28
her disappointment, which in her case was his father and to some ex- 29
tent himself. The typewriter clashed and harried him at night, never 30
rhythmical enough to be ignored. 31

Whilst it is obvious that words such as breadwinner (line 19), golfball (line
20), typewriter (line 20) or disappointment (line 29) can be regarded as
word formations, this is perhaps not quite so clear in cases such as run over
(line 26), drawn out (line 23) or golfball typewriter (line 20), academic
typing (line 21), shipping firms and art galleries (line 23). It is important to
note that in English spelling is not a reliable criterion, less so than in Ger-
man. It seems that in such cases the criteria listed in 9.1.1, positional mobil-
ity and uninterruptibility, are not entirely satisfactory: while they clearly
identify art galleries as a complex lexeme (in this case, a compound)
(4a) *art and other galleries
(4b) *The gallery is art
they are not quite so clear in other cases:
(5a) golfball and other typewriters
(5b) *The typewriter is golfball
(6a) academic and other typing
(6b) ?The typing is academic

155
Source: A. S. Byatt, Possession, London: Random House (1990; Vintage
edition 1991: 13–14).
102 Meaning-carrying units

(7a) ?shipping and other firms

(7b) *The firm is shipping
Thus, further criteria have been used to distinguish between compounds
and noun phrases consisting of an adjective as premodifier and a noun as
head:156 Schmid (2005: 133) points out that from a cognitive point of view
compounds are to be seen as new conceptual gestalts. Furthermore, com-
pounds are often characterized by additional semantic features (such as
‘habituality’) and by a tendency of compounds to have their main stress on
the first syllable.157 Under these criteria, art gallery and shipping firm
would qualify as compounds (although only the former can be found in
dictionaries such as the English Pronouncing Dictionary (EPD15) or the
New Shorter Oxford English Dictionary (NSOED). It is important to re-
member, however, that the distinction between complex lexemes (or ana-
lysable forms), which can be treated as word formations, and mere se-
quences of words that should not is likely to take the form of a gradient
rather than to be clear-cut.

9.2.2 Formal types of word formation: a survey

Depending on the kinds of morphemes involved in the analysis of a word

formation, different formal types of word formation can be distinguished.
The following types are often identified for English:158

Compounding: a process by which a new word is formed by combining
two constituents, which are both free morphemes or contain a free mor-
pheme159

156
For the terms premodifier and head in syntactic analysis see 11.3.1.
157
See Schmid (2005: 132–134), Kastovsky (1982: 152, 176–179) or Hansen,
Hansen, Neubert and Schentke (21985: 50–51). Compare also Lipka (32002:
99).
158
The following account is based on especially Kastovsky (1982) and CGEL
(1985: Appendix I). Compare also CamG (2002) and Plag (2003).
159
Cf. Schmid (2005: 121–122). For a detailed description of compounding in
English including spelling and stress patterns of compounds see Plag (2003:
132–163). See also Plag, Braun, Lappe and Schramm (2007: 94–95) and
Schmid (2005: 121–147). For a more precise definition of compounds see
Plag (2003: 135): “… a compound is a word that consists of two elements,
the first of which is either a root, a word or a phrase, the second of which is
 9 Word formation 103

breadwinner (l. 19), typewriter (l. 20), teaching hospital (l. 22–23) shipping
firm (l. 23), art gallery (l. 23), well-paid (l. 21), colour-illustrated (l. 14)
Besides compounding two types of word formation can be identified,
which are often subsumed under the term of affixation:160

Prefixation: a process by which a new word is formed by com-
bining two constituents so that a bound morpheme (prefix) pre-
cedes the second constituent
unfamiliar (l. 27), illegal, post-modern, prearranged, non-finite, non-smoker

Suffixation: a process by which a new word is formed by combin-
ing two constituents so that a bound morpheme (suffix) follows
the first constituent
disappointment (l. 29), boredom (l. 1), compatibility (l. 2), application (l. 8),
winner, writer
crucifying (l. 1), competitive (l. 8), enticing (l. 14)
The constituents to which the affix is added are often referred to as the
base (> 8.2).

Conversion: a process by which a new word is formed without
any formal change 161
work (verb, l. 22)  work (noun, l. 24), talk (verb, l. 24)  talk (noun), must
(verb, l. 26)  must (noun), bed (noun, l. 26)  bed (verb), clash (verb) 
clash (noun), this (determiner, l. 17)  this (pronoun, l. 17), point (verb, l. 17)
 point (noun), ship (verb)  ship (noun)
with modification of pronunciation (esp. stress shift):162 subject (noun, l. 15)
/"sʌbdʒɪkt/  subject (verb) /səb"dʒekt/, research (noun, l. 9) /"--/ or
/-"-/  research (verb)/-"-/.
While CGEL (1985: I.43) and CamG (2002: 1640–1644), for instance,
make use of the term conversion in this context, other scholars speak of

____________________________
either a root or a word.” Bauer (1983: 213–216) or Schmid (2005: 130–131)
identify a category of neoclassical compounds such as biography or Anglo-
phone, which are described as consisting of “combining forms”.
160
Cf. Lipka (32002: 96) or CGEL (1985: Appendix I).
161
The symbol  is used here as a device to signal a relationship that could be
interpreted in terms of zero-derivation. No commitment as to the direction of
the process is intended at this stage. See 9.3 below.
162
For formal modifications such as stress shift, voicing of final consonants
(advice – advise) or spelling see CGEL (I.57).
104 Meaning-carrying units

zero-derivation, which can be described as a word formation process by

which a new word is formed by the addition of a zero-morpheme.

Notice seen in Mousehole harbour

The next type of word formation can also be seen as involving affixes to
a certain extent:

Back-formation: a process by which a new word is formed by
taking off a sequence of phonemes resembling a suffix from an
existing word 163
laze (< lazy), burgle (< burglar); televise (< television)
What further distinguishes back-formation from compounding, prefixation
and suffixation is that it involves shortening, which is the main characteris-
tic of the following three types:

Clipping: a process of shortening a polysyllabic word, often to a
single syllable
Val (line 19), ad, demo, prof, phone, fridge

Blending: a process of forming a new word from two existing
words by maintaining elements of both constituents 164
heliport (helicopter + airport), smog (smoke + fog)

163
Cf. Kastovsky (1982: 174).
164
Cf. Schmid (2005: 224–225).
 9 Word formation 105

Acronymisation: a process of shortening a compound by using the

initial letters of the words it is made up of to form an acronym
which is pronounced as one word
NATO (North Atlantic Treaty Organization), CARE (Cooperative for American
Remittances to Europe), radar (radio detecting and ranging), LDOCE (Long-
man Dictionary of Contemporary English)

Abbreviation: a process of shortening a compound by using the
initial letters of the words which are pronounced separately 165
DES (l. 19), PhD (l. 19), IBM (l. 20), U.N., U.S.A., E.U., SOED
A further type of abbreviation is that of making a word shorter and combin-
ing it with an affix to produce a familiarity marker such as telly, comfy
(CGEL I.77).
One important distinguishing criterion between these word formation
processes is that only affixation, conversion and back-formation lead to a
change of word class.
Finally, word formation processes that are related to onomatopoetic or
aesthetic features of the words have to be mentioned: Marchand (21969:
429–439) speaks of “ablaut and rime combinations”, whereas CGEL (I.72)
identifies a category reduplicatives to cover cases such as see-saw, hig-
gledy-piggledy, wishy-washy, teeny-weeny etc.166

9.2.3 Semantic description of word formations

A semantic analysis of word formations (in the sense of lexemes formed by

a word formation process) can take different forms. For instance, since the
analysis of many complex lexemes involves not only one, but two or more
word formation processes, whose order is not irrelevant, it is important to
establish a hierarchy of immediate constituents. CamG (2002: 1626) uses
the following graphic representations to make this clear:

165
Sometimes abbreviations are subsumed under acronyms. Cf. however CamG
(2002: 1632–1633) where the term initialism is used to cover both types.
166
In terms of the distinctions made in 9.1.2 these processes might be seen as
cases of inventing words rather than of word formation. While CGEL (1985:
I.72) relates din-din to dinner, cases such as tick-tock, ha ha, ping pong seem
to be different in character.
106 Meaning-carrying units

This is closely related to the description of the relationship of the constitu-

ents to one another, where, following established terminology, a distinction
can be made between the determinant, which expresses a kind of specifica-
tion, and the determinatum, which can be characterized as representing a
given category.167 In typewriter, type specifies write and typewrite then
specifies the category represented by {ER} (‘machine or device’).
A common device to further characterize the semantic relationship be-
tween the constituents of a word formation is to classify it according to the
functional categories they have in an appropriate syntactic paraphrase:
CGEL (I.61–5) uses this method to characterize compounds and arrives at
types such as the following:168

167
For determinant and determinatum see Kastovsky (2006: 229–230), see also
Kastovsky (1982); for modifier and head see Plag, Braun, Lappe and
Schramm (2007: 97–98).
168
For a differentiation of these types see CGEL (I.61–5). For the historical
background of the use of such paraphrases, the justification of particular
paraphrases and the application of this method to various types of word for-
mation see Kastovsky (1982: 186–195). Compare also Burgschmidt (1973:
22–67).
 9 Word formation 107

type paraphrase further examples

subject and verb sunrise The sun rises. rainfall, sound change
verb and object tax-payer X pays cigar smoker, song-
tax(es). writer
verb and adverbial swimming X swims in living-room, walking
pool the pool. stick, washing machine
subject and object windmill The wind coal fire, steam engine
powers the
mill.
subject and complement girlfriend The friend is killer shark, oak tree
a girl.

A further and probably also more appropriate semantic characterization of

types of word formation can be achieved by drawing upon the model of
case grammar as developed by Fillmore (> 12.2). Lipka (32002: 107) em-
phasizes that the use of such case roles enables one to demonstrate the
complexities of the relationship between formal and semantic aspects in
word formation. Thus breadwinner could be described as ‘AGENT’, but
typewriter as ‘INSTRUMENT’.169
Despite such possibilities of a relatively general semantic description of
certain word formation processes, further aspects of meaning may have to
be taken into account. Thus, as Kastovsky (1982: 195) points out, an addi-
tional semantic feature ‘profesionally’ is required to characterize words
such as baker or shopkeeper. Furthermore it should be noted that while
such characterizations may be seen as typical of particular word formation
processes, individual lexemes that have been formed through such proc-
esses may be subject to considerable semantic modification. Thus, as has
already been pointed out in 8.5, blackbird does not mean ‘black bird’ but
refers to a particular species of birds (50 per cent of which are brown any-
way), and, similarly breadwinner can hardly satisfactorily be paraphrased
as someone who wins bread. Such semantic idiosyncracies will have to be
stated individually wherever they occur. They show that word formations

169
Compare also Lipka (1976) or Kastovsky (1982: 231–245) for a discussion of
case grammar and word formation. For the aspect of profiling in word forma-
tion see Schmid (2005: 105–109).
108 Meaning-carrying units

can be subject to semantic processes, which can be described in terms of

such concepts as lexicalization or idiomatisation (> 9.5).
It is important to point out that this kind of semantic analysis is only ap-
propriate to certain types of word formation processes. It is obvious that in
the case of acronymisation, blending and clipping, any semantic differences
that one might wish to claim between base and word formation are of an
entirely different character.

9.3 Word formation and morphology

9.3.1 The overlap between word formation and morphology

All of the processes outlined in 9.2.2 no doubt serve to create new words
and thus fall under the scope of word formation. However, only some of
them fall under the scope of morphology, if one follows the distinction as
made by Schmid (2005: 14):

morphology

inflectional derivational morphology: compounding other types of

morphology and affixation word formation

word formation

Although some linguists would subsume word formation under morphol-

ogy, it is certainly true to say, as Schmid (2005: 14) does, that morpheme
boundaries are not necessarily relevant to word formation processes such as
clipping or acronymisation.170 A further criterion that distinguishes com-
pounding and affixation from other types of word formation is that only the
former can be analysed in terms of a modifier-head or determinant-
determinatum structure. While art gallery can be analysed as a particular
kind of gallery and, with a certain sense of abstraction, disappointment as a
particular kind of -ment, no such analysis seems possible in cases such as

170
See Lipka (32002: 86) or Plag, Braun, Lappe and Schramm (2007: 89).
 9 Word formation 109

ad, LDOCE or smog.171 So there are good reasons for not including word
formation types such as acronymisation or clipping under morphology.
To what extent this also applies to cases such as the relationship be-
tween, say, clean (verb) and clean (adjective) is related to the question of
whether one sees this relationship as conversion or as zero-derivation.172
Lipka (32002: 102) points at parallels of the kind
legal : legal/ize clean : cleanØ‘make Adj’
atom : atom/ize cash : cashØ‘convert into N’
and points out:
The notion of zero-morpheme was not introduced arbitrarily in order to
complicate matters. It accounts for the fact that two homonymous lexemes,
which are superficially identical, are in a synchronically directed relation-
ship. One lexeme is the base, while the other one is derived from it by
means of a zero-suffix. Nevertheless, the two distinct lexemes are very
closely related semantically and morphologically.
While, within a particular theoretical framework, there may be very good
and well-argued reasons for postulating a unit such as zero-suffix, as was
pointed out in 8.3.3, there are also very good reasons for not accepting such
a concept as a sensible device in the analysis of language. Would one really
want to say that in a case such as work Ø, for instance, the zero element
represents a determinatum, which is specified by a determinant work?173
In any case, this is a good example of the way in which any statement
about a language is determined by the theoretical framework of the analy-
sis.174 If one is not convinced of the idea of a zero-morpheme, then the term
conversion is certainly more appropriate.
Funnily enough, although Quirk, Greenbaum, Leech and Svartvik
(CGEL I.43) avoid the term zero-derivation and speak of conversion in-
stead, they treat it as a “derivational process” within the appendix on word-
formation. What is more, they apply this term not only to cases such as ship

171
Compare Marchand (21969: 3) for a discussion of this. Back-formations and
clippings also contradict Kastovsky’s (1982: 153) view that only elements
that can be analysed into smaller units fall under the scope of word formation.
172
See Kastovsky (1982: 172). Cf. also Burgschmidt (1974) or Hansen et al.
(21985).
173
Furthermore, as pointed out in CamG (2002: 1641), the fact that conversion
involves a number of clearly distinct semantic properties raises the question
of how many zero-morphemes would have to be identified.
174
For a parallel view cf. Vennemann and Jakobs (1982/2000: 409).
110 Meaning-carrying units

(noun  verb) or answer (verb  noun) but also to changes of what they
refer to as “secondary word class” as in coffee (noncount noun  count
noun) or write (transitive  intransitive). But would we really want to ex-
plain the difference between the uses of write in (8a) and (8b) as a change
of secondary word class?
(8a)PH Writing for an American audience, he was, nevertheless, gently ironic
about the claims already being advanced for the American influence on
European art.
(8b)BNC A personal friend of an artist may have a real advantage in writing
criticism
In Chapter 12, a model of grammatical analysis will be introduced –
valency theory – which simply accounts for such differences between dif-
ferent uses of verbs by saying that certain elements they can occur with are
optional (> 12.3.5). From this point of view, a description of these phenom-
ena in terms of a word-formation process seems rather strange and counter-
intuitive.175
Again, we are faced with the theory-relatedness of statements about lan-
guage. To what extent do we want to account for the fact that words can be
used in different ways – “transitively” (write something) and “intransi-
tively” (write) or as adjectives and verbs in the case of clean – by saying
that they are different words and as such related by a word formation proc-
ess? Or would it not be more appropriate to say that (8a) and (8b) are just
examples of different uses of one verb write? But would this then also
mean that we are justified in talking about one word clean, which can be
used as an adjective or a verb?176

175
Equally, Bolinger (21975: 116) says that “there is a sort of ‘zero-derivation’
every time the meaning of a word is extended” and draws attention to the fact
that words affected by conversion are “new only in a grammatical sense”.
Compare Hansen et al. (21985: 125), who seriously argue that way. For this
problem see also Ayto’s (1996: 183) discussion of the word carer.
176
This issue is related to the question of which word classes one thinks appro-
priate to identify for English (> 11.4.5–8). It is interesting to see that CGEL
distinguishes between different uses of this in terms of the word classes pro-
noun and determiner, but that amongst the various categories of conversion
listed in CGEL, there is no category “pronoun  determiner” or vice versa
nor are such relationships between prepositions and adverbs (round, up) or
adverbs and adjectives (round, up) identified as categories of conversion. For
objections to this approach see Herbst and Schüller (2008) and Chapter
 9 Word formation 111

9.3.2 Explanatory value of the analysis

One major problem of a synchronic analysis of conversions (or zero-

derivations) is that of the direction of the word formation process. Since
there is no (or, if you like, only a zero) suffix, there is no morphological
criterion to go by. Furthermore, within a synchronic analysis reference to
the actual historical facts is ruled out for methodological reasons so that
decisions about the direction of the word formation process are usually
arrived at by a criterion which in CGEL (I.44) is referred to as the “seman-
tic dependence of one item upon another”. Indeed, referring to the exam-
ples from text (3) identified above, it seems more than plausible to assume
that the verbs bed, ship and point are secondary to the corresponding nouns
and not the other way round. But in cases such as work, research or love
(noun ‘feeling’; noun ‘person’; verb) etc., semantic evidence is not particu-
larly forcing in either direction. One might even ask whether there is any
need to arrive at such decisions at the synchronic level.177
With reference to back-formation, precisely this point is made by Mar-
chand (21969: 3):
The process called backderivation (backformation) has often diachronic
relevance only. That peddle vb is ‘derived’ from peddler sb through rein-
terpretation is of historical interest. For synchronic analysis, however, the
equation is peddle : peddler = write : writer, which means that the dia-
chronic process of backderivation does not affect the derivative correlation
for present-day speakers who do not feel any difference between the rela-
tionship write : writer on the one hand and peddle : peddler on the other.
The question is, however, if speakers of present-day English do not inter-
pret, say, edit and editor to be related in a different way from advise and
advisor, which seems plausible enough, is there then any point in identify-
ing backformation as a word formation process at the synchronic level?
And, considering how many native speakers might not be aware of the ori-
gins of fridge, flu, plane or stereo (clippings), bit (binary digit), moped or

____________________________
11.4.6–7. How problematic the notion of conversion is (or, for that matter,
that of zero-derivation) is also pointed out by Leisi (51969: 93), who rejects
the idea of conversion in a synchronic analysis and sees word classes as func-
tional categories in present-day English.
177
This could be seen as a consequence of Leisi’s rejection of conversion at the
synchronic level. See previous footnote.
112 Meaning-carrying units

telex (blends) and e.g., i.e., laser and radar (acronyms), could this argu-
ment not be extended to clipping, blending and acronymisation?178
The answer is probably: yes and no. Whether it is yes or no depends on
what one is trying to describe. As far as the analytic aspect of word forma-
tion as identified in 9.2.1 is concerned, it probably does not matter whether
speakers identify (or are able to identify) such words as results of word
formation processes once these words have become part of their lexicon.
With words they encounter for the first time, the situation may be different
in that speakers may well be able to interpret an acronym such as DES or
SCR if they are familiar with the words (or institutions) Department of
Education and Science or Senior Common Room. In the case of back-
formations or conversions, one is faced with a rather similar situation. A
form which could be analysed as the result of either process will presuma-
bly present no difficulty of interpretation as long as the speaker is aware of
the fact that such processes (back-formation, conversion from noun to verb,
conversion from verb to noun) are possible in the language. The situation is
probably even slightly more complicated in that there is no reason to as-
sume that speakers (in the process which one could call a lifelong process
of language acquisition) necessarily encounter the derived form after the
non-derived form. In that sense, speakers might well be able to correctly
interpret such items on the basis of patterns of relationship between words,
but possibly without any awareness of the directions the process of conver-
sion can take or the fact that back-formation can be identified as a separate
process from suffixation.
The point of this argument is that there are certain assumptions we can
make about the competence of speakers with respect to word formation,
both with regard to analytic and to synthetic operations.179 However, there
does not seem to be any convincing reason that a native speaker has to be
aware of the direction (suffixation or back-formation; verbnoun or
nounverb-conversion) of the process which produced a particular form.
Apart from cases where there is sufficient semantic evidence that would put
someone with a certain degree of linguistic awareness in a position to arrive
at certain conclusions about the likelihood of the direction of a particular
derivative process, there is absolutely no reason to assume that worrying

178
Examples taken from CGEL (I).
179
Cf. Burgschmidt’s (1978: 28) view of word formation competence as the
competence to analyse (known or new) word formations and produce new
word formations on the basis of productive rules.
 9 Word formation 113

about such questions is a concern of the speakers of a language at all. This,

however, raises the question of what the point of a synchronic analysis of
such word formations is. Why should a linguistic description aim to arrive
at any conclusions about the possible directedness of work and work or
possibly even edit and editor when such directedness need not be consid-
ered part of a native speaker’s competence at all?180

9.4 Productivity and restrictions

The situation is probably different with respect to the synthetic aspect of

word formation. In this respect, speakers of course have to be aware of the
fact that particular derivative processes can take place in two different di-
rections (i.e. that suffixation as well as back-formation or conversion from
verb to noun as well as from noun to verb etc. are possible in English.) As
far as the competence of speakers to produce new word formations is con-
cerned, it is obvious that it must comprise knowledge of the processes and,
in some cases, also the particular word formation elements (such as affixes)
that can be subject to such processes. This productive competence differs
from the receptive competence required for the interpretation of existing
word formations in that two factors which are irrelevant to the analytic
aspect of word formation are suddenly relevant. The first of these is, as
pointed out above, the questions of directedness, the second concerns the
factor of productivity.
Following Aronoff and Anshen (1998: 242), productivity can be de-
scribed as “the extent to which a particular affix” or word formation proc-
ess181 “is likely to be used in the production of new words in the language”.
Aronoff and Anshen (1998: 242–243) continue:
On this view, productivity is a probabilistic continuum that predicts the use
of potential words. At one end of the continuum are the dead or completely
unproductive affixes, which are not likely to be used at all in coining new
words. One example of this from English is the nominal suffix -th (as in
truth or growth), which has not been used successfully to form a new word

180
See also Coates’s (1987: 111–112) discussion of whether morphology should
be seen as “merely a historical discipline”.
181
Aronoff and Anshen (1998: 242) only use the term “affix” here, but produc-
tivity is generally seen as a feature of all word formation processes.
114 Meaning-carrying units

for 400 years, despite valiant attempts at terms like coolth182 (which is at-
tested sporadically, but which just never seems to be able to survive long).
At the other end in English are … highly productive derivational suffixes
like -ness and -ation. In the middle, we find the less productive derivational
suffixes like -ity. Some linguists treat morphological productivity as an ab-
solute notion – a pattern is either productive or unproductive – but there is a
good deal of evidence for the existence and utility of intermediate cases.
Productivity is of course extremely difficult to measure. Bauer (1994: 38),
in a survey of the lexical development of English between 1880 and 1982,
mentions a figure of 47.7% for suffixation, 18.5% for compounds and
10.4% for prefixation amongst the “types of formation in new words”.
These figures no doubt point at a high level of productivity of these word
formation processes. To what extent they can be seen as representative of
what happened in “the language” is of course a different matter since
Bauer’s data are based on lexicographical evidence, which, as Ayto (1996:
186) has convincingly demonstrated, can be misleading.183
To what extent productivity is related to frequency is also difficult to
say. Aronoff and Anshen (1998: 245) point at interesting correlations be-
tween word frequency and productivity:
… the less productive a morphological pattern is, the more frequent on av-
erage its individual members will be. But frequency is also important in the
selection of bases: a less productive affix is generally found attached to
higher frequency-base words than a more productive affix.
Productive word formation competence involves more than simply an in-
ventory of productive word formation processes and elements. It also com-
prises knowledge of a number of the distributional rules which apply to
certain word formation elements. These are often dealt with in terms of
restrictions: for English, Hansen, Hansen, Neubert and Schentke (1985: 33–
35) distinguish between the following types:184

182
Note, however, that coolth is listed in the OED and might become more and
more established as it is shown by the many hits of this word on the internet.
183
For a critical view see Ayto (1996: 186). For frequency and productivity see
also Plag (2003: 53–59).
184
For a detailed account of such restrictions see especially Hansen et al. (21985:
33–37). Plag (2003: 59–68) distinguishes between pragmatic restrictions,
structural restrictions and blocking. See also Burgschmidt (1978: 34–39) and
Kastovsky (1982: 160–182). Cf. especially Neuhaus (1971).
 9 Word formation 115

phonological restrictions of the kind that the morpheme {-ish} does not
combine with nouns ending in /ʃ/ or /tʃ/ (*rubbishish, *bitchish but:
rubbishy),

morphological restrictions of the kind that suffixations in {-ness} cannot
be followed by another suffix (*kindnessless, *awkwardnessful),

semantic restrictions of the kind that suffixations in {-ess} (stewardess)
only occur with nouns denoting human beings or higher animals,185

etymological restrictions of the kind that the negative prefixes {in-} and
{dis-} combine only with adjectives of Romance origin or that the pre-
fix {di-} only combines with words of Latin or Greek origin.186
It is obvious that such distributional rules must be seen as part of a native
speaker’s productive word formation competence.

9.5 Possible words – nonce formations – institutionalized

words

The discussion of word formation processes so far has in many ways shown
that many parallels can be, and indeed have been, drawn between word
formation and syntax. This is true in particular with reference to the notion
of a productive word formation competence, which shows obvious similari-
ties to how one could describe the competence of speakers to form sen-
tences. In very simplified terms one could say that in both cases there are
sets of rules which speakers can apply to what Chomsky (1957: 13) de-
scribed as a finite set of elements (in the sense of lexical material such as

185
Note that Hansen, Hansen, Neubert, and Schentke (21985: 35 and 103) use
the feature <+ HUM>.
186
As an example of a particular type of semantic restriction Kastovsky (1982:
161) mentions that the negative prefix {un-} does not occur with adjectives
with the suffix {-ish} as in (*uncleanish, *undeepish, *unplainish) although
the non-suffixed adjectives admit the prefix (unclean, undeep, unplain). The
fact that undeep and unplain are listed neither in NSOED nor in EPD15
throws some doubt on the importance of the phenomena described. Further-
more, Kastovsky (1982: 161) identifies a type ‘syntactic restriction’, where
the examples however also lend themselves to a semantic explanation.
116 Meaning-carrying units

morphemes or lexemes etc.) in order to produce words or sentences.187

However, there is a significant difference as to how the result or output of
this application of rules to items is judged by the language community. If,
for instance, native speakers of English were asked whether
(9)PH Several important painters of this generation, including William Scott,
Roger Hilton, Bryan Wynter, John Wells, Peter Lanyon, Alan Davie and
Terry Frost, all a few years older than Heron, were close friends, and in
the post-war period his associations with these and other artists were to
be consolidated by shared creative connections with West Penwith and
St Ives.
was a sentence of English, the answer would no doubt be yes. If, however,
native speakers of English were asked whether
(10)PH ceruleum
(11)PH juxtapositional
(12)QE psychophilatelic
(13)QE snow-cream188
were words of English, the answers would be more difficult to predict: in
the case of (12) and (13) it might well be no, whereas with (10) and (11)
answers might differ. People might even say “I don’t know” in a way that
they probably would not in the case of (9).
The point is, of course, that the two questions mean different things.
Judging something such as (9) as a sentence of English means that it is
grammatically well-formed according to the rules of English grammar,
whereas judging something such as (10–13) as a word of English would
probably be understood to mean ‘Does this word exist in English?’. If
speakers have not encountered a word before, they might be inclined to
judge it as non-existent in the language if it seems implausible to them for
some reason or other. However, the fact that it had not been encountered
before would not be used as a criterion to judge a sentence as not being a
sentence of English. In fact, it is relatively unlikely that any reader of this
text will ever have come across sentence (9) and even if someone had done,
in which case it is almost 100 per cent certain that they had read Mel Good-

187
While Chomsky actually refers to phonemes and letters in this passage, the
following discussion about grammaticalness allows the conclusion that this
view can be extended to lexical material (> 13.1.1).
188
Examples (12) and (13) from CGEL (I.13).
 9 Word formation 117

ing’s biography of Patrick Heron,189 they probably would not remember

that they had.190
A further difference in this respect between sentences and words is that
native speakers might well expect to find the answer to the question of
whether something is a word of English in a reference book and turn to a
dictionary in order to find it there or not, but, even if in doubt, they would
never turn to a grammar book and make their decision as to its Englishness
on the basis of whether they can find that particular sentence there.
In the case of (10) – (13), only one of these words, juxtapositional, can
be found in even a large dictionary such as NSOED. Is that sufficient evi-
dence to claim that ceruleum, psychophilatelic and snow-cream are not
words of English? Certainly not, since they all correspond to the rules of
English word formation in the way that (9) corresponds to the rules of Eng-
lish grammar. However, the fact that a word is not contained in a diction-
ary, at least in a dictionary of that kind, is generally taken as an indication
of its status in the language. In that respect, several distinctions can be
made:
− Words such as psychophilatelic and snow-cream, according to CGEL
(1985: I.13), or policeability, according to CamG (2002: 1624), are pos-
sible or potential words of English, i.e. words of English that corre-
spond to the rules of English word formation. At the same time, these
words are not used. There are different reasons why possible or potential
words of English have not come into existence – and even psychophi-
latelic and snow-cream paradoxically have only been formed in order to
demonstrate that they do not exist: one is that no one has seen the need
to conceptualise a certain sense, as in the case of psychophilatelic, of
which Quirk, Greenbaum, Leech and Svartvik (CGEL: I.13) say: “un-
used because psychological aspects of stamp collecting have not called
for lexicalization”. A second reason has to do with a phenomenon that
Burgschmidt (1973: 124) explains in terms of an occupied slot rule and
which is sometimes also referred to as blocking (> 9.6).191 Thus un-

189
Mel Gooding: Patrick Heron, London: Phaidon Press, 1994.
190
Compare also the very enlightening discussion of the differences between
words and sentences in this respect by Quirk et al. (CGEL I.7).
191
See Aronoff and Anshen (1998: 238): “Because blocking is a psychological
phenomenon, it is subject of the vagaries of the mind: if a person has tempo-
rarily forgotten the word fame, then that person may in fact use the word
*famousness, which fame would otherwise block. This seeming failure of
118 Meaning-carrying units

tall192 or Sinnvollizität may not be used because of the existence of the

words short and Sinn.
− Moving from “potential” to “established”, the next status to be identi-
fied is that of a nonce formation. 193 A nonce formation is a word that
has been formed and used by a speaker on a particular occasion but not
spread any (or much) further. This is the level at which the creative po-
tential of word formation resembles that of syntax. Numerically, nonce
formations may well be a much more important phenomenon of lan-
guage use than is often given credit for (especially in studies that ana-
lyse neologisms on the basis of dictionaries). Two very important areas
where nonce formations might feature prominently are cases where
speakers do not have a lexeme at their disposal which serves their im-
mediate communicative needs and the language of newspaper headlines.
The former can be imagined as being caused by “gaps” in the lexicon of
speakers, especially perhaps in the language of children,194 or indeed
gaps in the lexicon of the language as such, for example in the field of
terminology.195 In the language of the press, nonce-formations are used
consciously to achieve a certain, often humorous or ironical effect, as in
the case of Pharmaschinken, which the Süddeutsche Zeitung used as a
title of an article about one of the food scandals in 2002. Burgschmidt’s
extensive study of headlines in English papers include Confessions of a
chocaholic, In the bleak mid-winterval, Prof-lifers put MPs on the spot,
Would you Adam and Eve it? and England’s Knightmare.196
____________________________
blocking is especially common in children, who coin new words quite freely,
because their vocabulary is not as entrenched as that of adults. An articulate
child might use words like famousness and liquidize without hesitating.” So
might articulate adults, one might add, since famousness can be found in dic-
tionaries such as EPD15 or NSOED. Compare also Hansen, Hansen, Neubert
and Schentke (²1985: 35) for this.
192
Untall is analysed in CGEL (I.13) as “unused perhaps because alternatively
lexicalized”.
193
Bauer (1983: 45) defines a nonce formation as “a new complex word coined
by a speaker/writer on the spur of the moment to cover some immediate
need”.
194
See Kastovsky (1982: 163).
195
Cf. CGEL’s (I.15) discussion of The scribbler of this fould message should
be punished.
196
The Times (24/1/1998), The Guardian (23/12/1998), The Independent
(7/2/1997 referring to anti-abortionists), Sunday Times (19/1/1997) and Daily
 9 Word formation 119

− Nonce formations have to be distinguished from institutionalized

(CGEL) or established (CamG 2002) words.197 Quirk, Greenbaum,
Leech, and Svartvik (CGEL: I.7) express this distinction in terms of the
contrast between “the ‘made-to-measure’” and “the ‘ready-made’” and
say “we expect every lexical item we hear or read to be already ‘institu-
tionalized’ … and to have been selected from an existing stock of
words, complete with its form and meaning”. Consequently, it is the in-
stitutionalized words of a language that are the candidates for being en-
tered into a dictionary. (The opposite does not hold true, of course, one
could not expect a dictionary, however large, to list all the institutional-
ized words of the language.)
While the distinction between nonce formations and institutionalized words
is probably a rather useful one, it ought to be pointed out that institutionali-
zation is probably a matter of gradience in the sense that not all speakers
may perceive a word as being institutionalized or not in the same way.
A further distinction can be made between institutionalization and lexi-
calization – institutionalization referring to the socio-pragmatic aspect and
lexicalization to the structural aspect of word formations (Schmid 2005:
79–85). While the term institutionalization describes the fact that a word
formation is known198 and used by a large number of speakers, a word for-
mation can be called lexicalized if it can no longer be fully related to the
underlying word formation process – either at the formal or the semantic
level.199 The extreme case is to be found in so-called amalgamated com-
pounds such as cupboard (loss of accent on the second element), boatswain

____________________________
Mail (23/12/96 referring to batsman Knight). Examples collected by Ernst
Burgschmidt and presented in a paper at the Deutsch-Britische Gesellschaft
in Nürnberg in 2002.
197
Bauer (1983: 48): “The next stage in the history of a lexeme is when the
nonce formation starts to be accepted by other speakers as a known lexical
item.”
198
Bauer (1983: 48) takes it as a criterion of institutionalisation “that the poten-
tial ambiguity is ignored, and only some of the possible meanings of the form
are used (sometimes only one)”. Compare, however, CGEL’s (I.11) use of
the term lexicalization in this respect. For a distinction between different
types of lexicalization see Bauer (1983: 48–61).
199
For this definition of institutionalization and a more detailed discussion of the
terms lexicalization, institutionalization and concept formation see Schmid
(2005: 79–85).
120 Meaning-carrying units

/"b@Us@n/ or /"b@UtsweIn/, breakfast, daisy, husband, neighbour etc.200 Less

extreme cases are provided by words that have developed semantic proper-
ties that are not immediately predictable from those of their component
parts such as with blackboard (which need not be black)201 or bus driver
(with stress on the first element and a component ‘professionally’). Quirk,
Greenbaum, Leech and Svartvik (CGEL: I.11) discuss the example of the
verb paint, where “two verb senses have been separately lexicalized” –
‘decorate with paint’ and ‘make pictures with paint’.

9.6 Psychological aspects of morphology

What can be concluded from the outline of morphology and word forma-
tion given in Chapters 8 and 9 is that the concept of the morpheme as de-
fined in structuralist linguistics serves quite well to account for complex
word forms such as critics, paintings or coastlines but is less convincing in
other cases. With respect to psychological aspects of language use, it is the
aspect of item-specific knowledge that needs to be accounted for. This con-
cerns two slightly different phenomena. Firstly, speakers of a language
know that the past tense form of begin is not formed by adding /d/. In tradi-
tional accounts, this sort of phenomenon is treated as an irregularity or an
exception. Secondly, speakers know that words such as kindness and fury
exist and seem to avoid using kindity or furiousness, which is where the
aspect of institutionalization and lexicalization is relevant. We are talking

200
Examples taken from Götz (1971), who discusses their historical develop-
ment in detail. For a discussion of lexicalization in terms of “unrecoverabil-
ity” and “unpredictability” see Quirk et al. (CGEL: I.9–10), who also discuss
questions of terminology such as fissionable and fissile in this context. Com-
pare also Brinton and Arnovick (2006: 173).
201
See Lipka (32002: 115). Cf. Lipka’s (32002: 113) definition of lexicalization
as a “gradual, historical process, involving phonological and semantic
changes and the loss of motivation”. Lipka refers to semantic aspects of lexi-
calization as idiomatization (Lipka 32002: 114). Cf. also Coates (1987: 116).
For cognitive aspects of lexicalization see Ungerer and Schmid (22006: 268–
275). For lexicalization and grammaticalization see Hopper and Traugott
(22003: 133–135).
 9 Word formation 121

here about conventionalized or established ways of expressing certain

meanings in a language.202
In fact, the question of to what extent certain types of word formation
should be treated as a rule-driven or as an idiosyncractic phenomenon has
been subject of much discussion in linguistics.203
In more recent work, the discussion of this question has focussed more
on psychological issues. Coates (1987: 115) addresses the matter by asking
whether it is “possible to decide whether a form is rote-learned (stored
whole) or combined in usage from its constituent parts”. While Coates is
concerned with word formations such as toothbrush, there are also obvious
parallels between the kind of idiosyncracy observed in word formation and
irregularity in inflection. For this reason, the two aspects are often dealt
with together in discussions of the mental lexicon. Jackendoff (1997: 121)
translates the problem into psycholinguistic terms in the following way:204
The basic psycholinguistic question about morphology is whether complex
morphological forms are stored in long-term memory or composed “on-
line” in working memory from stored constituents.

202
This can be accounted for in terms of Coseriu’s (1973) distinction between
system and norm, which, as, Burgschmidt (1973 and 1977) shows, can fruit-
fully be applied to word formation (> 3.1.4). The rules underlying potential
and actual word formations of a language can be described in terms of the
system, but since not all lexical units the system would permit are actually re-
alized in the language, the inventory of word formations in the sense of insti-
tutionalized lexical units can be considered a norm phenomenon.
203
Within the strongly formalized models that became current in the years fol-
lowing the publication of Chomsky’s (1957) Syntactic Structures the discus-
sion took the form of asking whether certain linguistic facts ought to be
placed in the syntactic component of a model of grammar or be relegated to
the lexicon. It culminated in the formulation of the so-called transformational
hypothesis by Lees (1960/1968), who analysed nominalizations such as the
committee’s appointment of John or John’s appointment by the committee as
“noun-like versions of sentences” of the type The committee appoints John.
The alternative analysis was favoured by Chomsky (1970) in the form of the
lexicalist hypothesis. Compare Herbst (1983: 46–54) and Kastovsky (1982:
219–229).
204
Cf. Clark (1998: 388). See also the discussion of different approaches given
by Jackendoff (1997: 121–123). Compare also McQueen and Cutler (1998:
407), who talk about “the relative importance of rule-based processing and
rote-storage” in this context.
122 Meaning-carrying units

A lot of psycholinguistic research seems to be in line with the assumption

made by Coates above, that “individuals may both store particular complex
forms whole and have them available for analysis”.205 This both-and-
situation seems to manifest itself in different ways:206 one of these is repre-
sented by so-called dual-route (or dual processing) models, favoured, for
instance by McQueen and Cutler (1998: 423) “where some words are mor-
phologically parsed prior to access via their constituent morphemes and
others are accessed directly via whole-word representations”. This means
that this model permits different processing mechanisms to be at work.207
Similarly, Aronoff and Anshen (1998: 240) account for the blocking of
inflectional forms such as *goed or lexical forms such as furiosity in terms
of a “race between the mental lexicon and the morphology. Both operate
simultaneously, and the faster one wins”.
An alternative to such approaches has been developed by Bybee (1995:
428–429) in the form of the network model. She characterizes her model as
being different from “structuralist models containing rules”:
The basic proposal is that morphological properties of words, paradigms
and morphological patterns once described as rules emerge from associa-
tions made among related words in lexical representation. …
Words entered in the lexicon are related to other words via sets of lexi-
cal connections between identical and similar phonological and semantic
features. These connections among items have the effect of yielding an in-
ternal morphological analysis of complex words, as shown by Fig. 1.

205
McQueen and Cutler (1998: 420), discussing a number of experiments in the
field, come to the following conclusion: “In summary, the evidence for de-
composition reviewed so far is stronger for inflectional than for derivational
forms. Decomposition may be an optional strategy for derived words, avail-
able when normal access procedures fail. It would therefore appear that de-
rived words have independent whole-word access representations. Inflected
words may not have their own access representations, and access to the cen-
tral lexicon for inflected forms may be via decomposition, leading to access
representations of their component stems and affixes.”
206
For a discussion of the dual-processing model, connectionist models and the
network model see Bybee (1995).
207
Compare also Pinker (1991: 532) or Pinker and Prince (1994: 326).
 9 Word formation 123

Even though the words entered in the lexicon are not broken up into their
constituent morphemes, their morphological structure emerges from the
connections they make with other words in the lexicon. Parallel sets of pho-
nological and semantic connections, if they are repeated across multiple sets
of words, constitute morphological relations (in Fig. 1, these are repre-
sented by heavier lines.) Note that in Fig. 1 [see above] connections be-
tween base ([kæt]) and complex form ([kæts]) exist, as well as relations
among complex forms ([ræts], [mæts], [kæts], [kæps]).
Central to the model are the concepts of schemas and strength. Of the for-
mer Bybee (1995: 430) says: “Sets of words having similar patterns of se-
mantic and phonological connections reinforce one another and create
emergent generalisations describable as schemas.”
When we ask how many and which elements of morphology ought to be
seen as derived on the basis of morphological rules, and thus be attributed
to the morphological component of a model of language, and how many or
which elements ought to be treated in terms of the lexicon, the rejection of
the notion of rules in Bybee’s model has of course far-reaching repercus-
sions. Whereas other models would claim that, say, all regular past tense
forms are the result of applying a rule of some kind and irregular past tense
forms (and only these) are stored in the lexicon, according to Bybee (1995:
450) both irregular and highly frequent regular past tense forms are stored
in the lexicon. The past tense forms of low-frequency verbs are then arrived
at “by applying the strongest schema to base forms”.208 This approach is of
course very much in line with the finding that the verbs that have irregular
past tense and participle forms tend to be amongst the most frequent verbs

208
Bybee (1995: 450–451) takes phonological evidence concerning the duration
of the final consonants in words such as rapt or rapped to reflect “a differ-
ence in processing type”.
124 Meaning-carrying units

in English, and also that the verb be shows greater morphological diversity
than any other verb in the language.209
Bybee (1995: 452) argues that morphology can be accounted for solely
in terms of a “highly structured lexicon” and describes a functionalist view
of the lexicon as follows:
The distinction between storage of complex forms versus formation by
schema application are based on the availability of stored items, which is
determined by frequency of use, not by structural distinctions, such as the
classifications into regular and irregular patterns.
What makes such an approach very interesting is that it may also offer an
explanation for irregularities of the type curriculum – curricula, which are
awkward to account for in structuralist morphology, and, in particular, that
it can easily be extended to larger phraseological units.210

209
One particularly interesting aspect of this is the phenomenon that in a certain
stage of language acquisition children seem to overgeneralize -ed-past tense
forms. This is often explained in terms of children having acquired the rule of
past tense formation at this stage. Bybee (1995: 449) explains this in terms of
children having understood the obligatory nature of tense marking. See
Tomasello (2003: 232–235).
210
Cf. Bybee (2007: 281–292). Thus usage-based approaches play an increas-
ingly important role in cognitive linguistics – not only in morphology, but
also in other areas of linguistics such as syntax (> 13.2), and in particular in
models of language acquisition (Tomasello 2003). It is interesting to see that
the concept of pre-emption can be used for morphology as well as syntactic
constructions, see Tomasello (2003: 178–182 and 234).
10 Phraseology

10.1 Prefabs

Institutionalized lexemes can be seen as prefabricated items which speakers

can make use of when conceptualizing ideas and forming sentences. It
would be a mistake, however, to suppose that prefabrication stops at the
word boundary. While some aspects of this phenomenon have always
raised scholarly attention, especially as far as collocations (> 10.2–3) and
idioms (> 10.4) are concerned, the full dimension of this phenomenon has
only become apparent through the results of the machine-supported analy-
sis of large text corpora. These findings are related to statistical data and it
is thus of particular interest that frequency is also taken as a highly relevant
factor of language description in some recent models of the mental lexicon
in psycholinguistics (> 9.6).
There is a lot of experimental and corpus evidence to illustrate the role
prefabricated material plays in language production.
− For instance, sentence completion tests carried out by Greenbaum
(1974/1988: 118–119) demonstrate the high association between par-
ticular lexical items. The following table lists the percentages of the
verbs that were given frequently by British and American speakers for
the sentence beginnings in the left-hand column:211

I badly UK: need 65% – want 28%

US: need 48% – want 17%
Your friend very much UK: like 29% – want 18%
US: like 19% – want 10%
I entirely UK: agree 82%
US: agree 27% – forget 13%
I completely UK: forget 50%
US: forget 46%

211
For the results of these tests with German students of English cf. Herbst
(1996: 389–391).
126 Meaning-carrying units

− Mittmann (2004) points out that in two large corpora of American and
British spoken English, all occurrences of beg your were followed by
pardon, while all occurrences of your pardon were preceded by beg or
that should think is almost always preceded by I.
− Sinclair (2004: 45–47) has shown that if one finds the phrase naked eye
at the beginning of a sentence, it is very likely to be preceded by the,
which again is likely to be preceded by with or to, with is preceded by a
form of see, which in turn tends to be preceded by a modal verb, to
tends to be preceded by visible or invisible.
− The following extract from the British National Corpus illustrates that
the phrase would you is followed by the verb like in 25% of all occur-
rences.

What else would you expect after the rotten day I’d had?
“What would you know?” demanded Thomas, the under-footman,
jealous that it had been she who had found the
“You wouldn’t expect them to paint it would you ?”
pretty pink,
Would you like to come round for a cup of tea?”
Would you like some more tea?”
Would you mind desperately if just this once I paid at the
other end?”
Would you like some tea?”
I’ll check to see what’s on offer and in would you like a day’s pay here?”
the meantime, how
Would you like to come and sit out the back with me for a
minute?”
Would you mind if I bought you a coffee or something?”
“ Would you take the flat now, then?
“Where would you like to live?”
Do you want to check into a hostel just would you rather have another night in the park?”
in case it rains or
“Where would you like to eat?” she asked me.
I paused for a moment, then added, “ Would you mind if I had some pudding?”
Would you mind if I made a few housing enquiries for you?
Would you like some tea?”
If you were given “carte blanche” so to would you go for?”
speak, what
“ Would you like some newspapers to read on your journey?”
Would you like Jenny to go along as well?”
Which would you rather have in your garden --; sticky clay or thin,
hungry sand?
Would you do something for me?
If you did not burn chemicals like would you do with them?
PCBs, what else
“If the people in Hong Kong were would you be taking the same view?”
white,” he asked,”
 10 Phraseology 127

Would you care to hear my analogy?”

“Right,” said Marina, “Dionne, what would you be?”
kind of animal
And now, she thought in the blessed would you like to do?
silence, my dear Jay, what
“ Would you do that, Donald?
Would you Hide if you had to?”
“Well, where else would you go to meet up with the Duke and his followers?”
Donald came down two hours ago, and would you believe it, Bisset will not come to marry them.”
“Well --; Donald asked me to ask you -- would you do the honours?”
;
If you were Atholl, who would you come for first?
ands and said eagerly, “Don --; we would you like that, Don?”
could live here --; it is right for us --;
“If a deadly beast was coming for you, would you do?
what
Would you lie low, if it came to it?”
“ Would you talk to me for a few minutes?”
“ Would you sit closer?” he asked, smoothing the sand.
Would you , I wonder, would you walk with me?
Would you, I wonder, would you Walk with me?
Take my situation; you’d never be able would you ?
to handle that,

It is facts like these which point at the relevance of units higher than the
individual word or the individual lexeme in linguistic analysis. What is
particularly interesting to observe is that prefabrication does not necessarily
coincide with units of syntactic analysis such as phrase, clause or sentence
(> 11.1.2 and 11.3). This was demonstrated by Altenberg (1998: 120):

What is perhaps the most striking impression that emerges from the mate-
rial is the pervasive and varied character of conventionalized language in
spoken discourse. The use of routinized and more or less prefabricated ex-
pressions is evident at all levels of linguistic organization and affects all
kinds of structures, from entire utterances operating at discourse level to
smaller units acting as single words and phrases.

Altenberg (1998) identifies the following types of recurrent word-

combinations:

full independent clauses it’s all right, that’s all right, well I don’t know
full dependent clauses if you like, that is to say, I should think
multiple clause constituents and you know, there is, and then I, because I
mean
128 Meaning-carrying units

single clause constituents on the whole, at the moment, the whole thing, in
this country, and so on
incomplete phrases a sort of, a kind of, part of the, one of the

The enormous amount of variation reflected in this material demonstrates

the problems of classification in this field.212 Altenberg (1998: 121) con-
cludes:
The picture that emerges … emphasizes rather than clarifies the fuzzy char-
acter of phraseology. Speakers engaged in spontaneous interaction are in
constant need of easily retrieved expressions to convey their intentions and
reactions in discourse. At their disposal they have a large stock of recurrent
word-combinations that are seldom completely fixed but can be described
as ‘preferred’ ways of saying things – more or less conventionalized build-
ing blocks that are used as convenient routines in language production.

10.2 Statistically significant collocations

That certain affinities hold between particular words can be demonstrated

by looking at almost any text.213

(1) I shall never forget the immense sensation of space the first moment 1
we entered that room, at the end of our journey from London: it was 2
an October night and a full moon was rising over Godrevy. … I 3
probably painted more St Ives harbour window paintings in London 4
than in Cornwall. … Also established at the time I painted it was my 5
habit of applying only one coat of colour over the white ground, 6
which seems to be responsible for the brilliance of hue – there are one 7
or two areas, very noticeably, where I have applied more than one 8
colour … 9

This short passage contains quite a number of co-occurrences of words

which one might consider typical in one way or another: I shall never for-
get, enter – room, full – moon, moon – rise, apply – colour. Following the

212
See, for instance, Gläser (21990).
213
From a letter by Patrick Heron to the Tate Gallery, quoted in Mel Gooding:
Patrick Heron, London: Phaidon Press, 1994, 74.
 10 Phraseology 129

terminology established by the British contextualist John Rupert Firth, who

emphasized the “mutual expectancy of words”214 such combinations are
generally referred to as collocations. 215 Sinclair (1991: 170) defines collo-
cation in a very general way as “the occurrence of two or more words
within a short space of each other in a text”. This mutual expectancy can be
demonstrated very nicely by looking at the likelihood of co-occurrence on
the basis of corpus material (which, of course, has to be treated with some
caution but nevertheless can be taken as some indication of the facts). For
instance, the BNC contains 148 instances of the cluster I shall never. These
are followed by the following words:216
forget 48; know 17, be 13, see 9, go 5, marry 4, forgive 4, get 4, go 4, let 3,
understand 3, regret 2 and words which occur only once.
Corpus analysis enables us not only to find out the absolute frequencies
of words occurring together within a certain span (the number of words to
the left and to the right from a particular key word), but to determine so-
called statistically significant collocations, which are “collocations
that occur more frequently than would be expected on the basis of the fre-
quency of occurrence of individual items” (Herbst 1996: 382). The statisti-
cal significance of collocations can be represented in the form of log-
likelihood values, T-scores or Z-scores, which are different ways of relating
the frequency of occurrence of a collocation to the probability of the two
elements of the collocation co-occurring in that corpus. Thus a lemma
query (covering all morphological forms) of the verb paint in the BNC lists
the following among the 50 most frequent nouns occurring within a span of
+1 to +3, i.e. within the next three words:

214
Cf. Firth (1957/1968: 181).
215
For a discussion of Firth’s ideas on collocation and the distinction between a
text oriented, a statistically oriented and a significance oriented approach to
collocation see Herbst (1996). For discussions of different approaches to col-
location see Nesselhauf (2005: 11–40) or Siepmann (2007: 236–237 and
2008: 186–187).
216
All figures in this chapter are based on BNC World Edition queried through
BNCweb (described by Lehmann, Hoffmann, and Schneider 2000).
130 Meaning-carrying units

lemma collocate number of co- log-likelihood value

(noun) occurrences
paint (verb) picture 178 1824.9
portrait 39 420.9
pictures 38 321.1
colours 32 272.0
portraits 19 208.9
house 30 109.9
boats 7 49.6

Similarly, a lemma query for the noun moon within a span of -3 to -1 in-
cludes the following collocates:

lemma collocate number of co- log-likelihood value

occurrences
moon (noun) the 2074 8534.2
full 209 1951.2
sun 88 829.4
over 102 494.4
phases 23 265.7
blue 31 233.3

These figures show that there is not necessarily a direct correlation between
the absolute frequency and the statistical significance of a combination.
Thus I shall never see appears more significant in terms of its log-
likelihood value (56.1) than I shall never be (47.2).
Furthermore, statistically significant collocations may of course be the
result of a number of factors, not all of which are linguistically interesting.
If, for instance, it is found that house frequently collocates with sell in a
corpus of British English, this may be an indication of certain elements of
British culture (such as that British people tend to talk about selling houses
quite often), but it is not necessarily something that would affect the lin-
guistic description of house or sell. Nevertheless, the statistically oriented
 10 Phraseology 131

approach to collocation can make an important contribution to, for exam-

ple, foreign language teaching, for instance, as far as the selection of typical
and natural examples is concerned. Furthermore, irrespective of the degree
in which they are caused by cultural factors, the phenomenon of statisti-
cally significant collocations opens up an important field of psycholinguis-
tic research in that a tendency of co-occurrence of particular words can be
surely seen as an important factor in the planning and perception of speech.

10.3 Institutionalized collocations

There is a further type of collocations to be distinguished from statistically

significant collocations, which can also be characterized in terms of the
significance of the co-occurrence of two or more lexical items, which how-
ever is not related to frequency. Such collocations, as Sinclair (1991: 170)
puts it, “can be dramatic and interesting because unexpected”. Indeed, it is
the unexpectedness or unpredictability at the semantic level that character-
izes such collocations. Thus there is no particular reason why white wine
should be called white wine and not *yellow wine, why weak tea should not
be *feeble tea or why a conscience should be clear and not *clean. Collo-
cations of this kind are norm phenomena in very much the same way as
word formations, where there is also no reason why blackbird should not
be *brownbird or dishwasher should not be *washing-up machine.
Nevertheless, the term unpredictability is very problematical as a crite-
rion for determining the status of a collocation. Firstly, as Palmer (21981:
77) points out, whether a combination is semantically predictable or not
depends on the meanings one assigns to the items involved.
It would … be a mistake to draw a clear distinguishing line between those
collocations that are predictable from the meanings of the words that co-
occur and those that are not. … For one can, with varying degrees of plau-
sibility, provide a semantic explanation for even the more restricted colloca-
tions, by assigning very particular meanings to the individual words. … We
can thus explain white coffee, white wine … by suggesting that white means
something like ‘with the lightest of normal colours associated with the en-
tity’.
Indeed, if one argues that it is part of the (or one) meaning of white that it
can refer to wine or part of the meaning of the lexeme coat that it can refer
132 Meaning-carrying units

to paint, then the combinations white wine and coat of paint217 are highly
predictable. In fact, sentences of the type
(2) Do you prefer white or black?
strongly support such an analysis since (2), quite obviously, will be taken
to refer to coffee (as an extralinguistic entity), while – and that is the crucial
point – the use of white and black is not dependent on the presence of the
word coffee.218 Nevertheless, an element of unpredictability remains in that
within such an analysis of the meaning of white it is difficult to explain
why tea with milk is not generally referred to as *white tea (although tea
whitener seems to exist as a word).219
The second problem with the criterion of unpredictability is that unpre-
dictabiliy entails the question of “unpredictable for whom”. Thus, to take a
simple example, make coffee is a common collocation in English, whereas
*boil coffee or *cook coffee are not. From a semantic, intra-English point of
view, make coffee does not seem to be a case of great unpredictability. This
is different for users of English with a different language background, how-
ever. Germans who tend to say Kaffee kochen have to learn that it is make
coffee in English, Germans who say Kaffee machen are not confronted with
a collocational difficulty. Similarly, it may depend on a foreign user’s lan-
guage background whether apply colour and coat of colour present a prob-
lem or not. While such considerations lead to the conclusion that the ideal
place for dealing with collocations is the bilingual and not the monolingual
dictionary, as Klotz220 has argued, the contrastive approach can of course
only be seen as a methodological device for discovering collocations of this
kind in a language. Nevertheless, it highlights two characteristics of collo-
cations of this kind: firstly, that they are norm phenomena (> 3.1.4), and
secondly, that they have to do with choice. Thus one might argue that one
could postulate a theoretical choice between, say, good, clear, and clean in
the case of conscience, but only clear conscience seems to be an estab-

217
Coat of colour as used in text (1), however, seems to be rather unusual as a
collocation and restricted to the artistic context.
218
Cf. Herbst (1996: 386–389).
219
This may be an instance of language change: occasionally, the combination
white tea can be found in coffee shops etc.; the BNC has three instances of
white tea, one of which is an example of coordination a cup of white tea or
coffee, the other two referring to instant tea.
220
I owe these examples and this line of argument to Michael Klotz (personal
communication). See Herbst and Klotz (2003).
 10 Phraseology 133

lished collocation, and good is used in the phrase in good conscience. This
arbitrariness of the combination features prominently in Hausmann’s
(1985: 118) definition of collocations as “typical, specific and characteristic
relations between two words”.221
In fact, one could approach a definition of this type of collocation by
saying it occurs in cases where on semantic grounds there would seem to be
a lexical choice in the language but where the combination of two (or more)
particular lexemes is so established that it is the only combination possible
or at least the preferred combination. Nevertheless, the frequency factor
should not be overlooked. Thus, looking at three possible collocates of the
words gale(s), storm(s) and wind(s) one finds:

gale gales storm storms wind winds

heavy 1 0 2 4 3 2
severe 4 6 6 9 1 1
strong 4 0 0 0 85 144

As far as these figures can be regarded as representative at all, apart from

strong storm(s) all combinations have been used, while at the same time
there is a strong tendency for gale(s) and storm(s) to co-occur with severe
and for wind(s) to be used together with strong.222 This can be represented
in the following way:

heavy storm

severe gale

strong wind

221
German original: “typische, spezifische und charakteristische Zweierbezie-
hungen von Wörtern”.
222
Of course, one has to consider the overall number of occurrences of these
words in the corpus: storm (2271), storms (489), gale (443), gales (193),
wind (7357), winds (1617). Note, however, that these include noun and verb
uses of these word forms.
134 Meaning-carrying units

In a way, collocations of this type can be compared to institutionalized

word formations. They are institutionalized in the sense that they are pre-
fabricated items, “Halbfertigprodukte der Sprache” in Hausmann’s (1984:
398) words, and not productively created on the spur of the moment. Up to
a point, institutionalization in this case also entails limiting the choices
available to a speaker at the formal level.223 One important difference, how-
ever, is that in a collocation, as opposed to word formation, the semantic
independence of the constituents is not affected, even if this criterion
clearly admits for a considerable amount of gradience.224

10.4 Idioms

This transparency of meaning is also often taken to be the distinguishing

criterion between collocations and idioms. An idiom can be defined as “a
combination of two or more words which function as a unit of meaning”
(Cowie and Mackin 1974: viii–ix) or as “a lexical complex which is seman-
tically simplex” (Cruse 1986: 37).225 As such, idioms are multi-word lex-
emes.226
Idioms can take different forms, for instance:
− noun phrases: the wind of change, bad blood, a nervous wreck
− predicates: have a frog in one’s throat, have second thoughts, pull
strings, scream blue murder
− sentences: the early bird catches the worm, don’t cry over spilt milk

223
This type of collocation has, of course, received considerable attention in the
fields of foreign language teaching and lexicography. See Cowie (1981),
Hausmann (1984, 1985), Alexander (1992), Bahns (1996), Herbst (1996),
Siepmann (2005, 2007, 2008) and Lea (2007).
224
Cf. Cruse (1986: 40): “The term collocation will be used to refer to se-
quences of lexical items which habitually co-occur, but which are nonetheless
fully transparent in the sense that each lexical constituent is also a semantic
constituent.”
225
Cruse (1986: 37) criticizes definitions of idioms which characterize them as
expressions “whose meanings cannot be inferred from its parts” and arrives at
the following conclusion: “We shall require two things of an idiom: first, that
it be lexically complex – i.e. it should consist of more than one lexical con-
stituent; second, that it should be a single minimal semantic constituent.”
226
It could be argued that idioms are word formations and should have been
dealt with in Chapter 9.
 10 Phraseology 135

− phrasal verbs (combinations of verb and adverbial particle): look up,

come up with.
The area of idioms is subject to gradience in various respects. First of all,
idioms are lexically stable to various degrees. So have a screw loose is
paralleled by got a screw loose and with a screw loose, similarly one finds
get a good press as well as get a bad press, get a fair deal and get a square
deal etc. Secondly, idioms can be used in a shortened form, so, for instance,
the early bird catches the worm can be reduced to early bird etc. Thirdly,
the extent to which idioms allow syntactic modification also differs consid-
erably, so scratch one’s head can also be found in a sentence such as
(3)QE This caused a bit of head-scratching at the BBC.227
whereas passives
(4) ?*Blue murder was screamed
(5) ?*The wood was not seen for the trees
or correspondence to the general rule of concord are often subject to restric-
tions:228
(6) *They all had frogs in their throats
(7a) *They cannot have their cakes and eat them too.
(7b)BNC As it stands the districts seem to be wanting their cake and eat it in that
they would like er a policy restricting development in the open country-
side but they don’t want it to come with baggage that is specific which
says what the exceptions should be.
Furthermore, of course, different degrees of idiomaticity can be distin-
guished, as Cowie and Mackin (1974: x) also point out: “the boundary be-
tween highly idiomatic items and the rest is not sharply drawn but hazy and
imprecise. We shall do better to think in terms of a scale of idiomaticity”.
This can be illustrated by looking at examples containing the particle up:
(8a) They pulled the flag up.
(8b) They pulled the flag.

227
Example taken from the Oxford Dictionary of Current Idiomatic English
(ODCIE), volume 2.
228
Since idioms are often used in informal contexts, generalizations of this kind
are difficult to make since jocular uses might easily occur. See also Götz
(1976: 63–65).
136 Meaning-carrying units

(8c) The flag was up.

(9a) They ate the cake up.
(9b) They ate the cake.
(9c) *The cake was up.
(10a) They gave up smoking.
(10b) *They gave smoking.
(10c) *Smoking was up.
Pull up is a perfectly transparent combination of the verb pull and the parti-
cle up, which has no idiom status at all; in eat up the particle up does not
have a locative, but a terminative meaning (in the sense ‘bring to an end’ as
in finish up), whereas in give up neither the verb nor the particle contain the
meanings they have in other contexts, so give up is best analysed as one
unit of meaning.

10.5 The idiom principle and the mental lexicon

Whereas idioms, as has been pointed out in 10.4, clearly have lexeme status
and thus have to be stored in the mental lexicon, the status of other prefab-
ricated items such as collocations (in either interpretation of the term) is
less clear. Corpus linguists like Sinclair and linguists concerned with for-
eign language teaching certainly emphasize the importance of these ele-
ments for the production of sentences. Psychologically, this raises the issue
of how much weight ought to be given to rule-governed aspects and how
much to rote-learned aspects in the sense of prefabricated items.229
Sinclair (1991: 109–110) contrasts the open-choice and the idiom prin-
ciple in this context. Of the open-choice principle he says:
This is a way of seeing language text as the result of a very large number of
complex choices. At each point where a unit is completed (a word or a
phrase or a clause), a large range of choices opens up and the only restraint
is grammaticalness.
This is probably the normal way of seeing and describing language. It is
often called a ‘slot-and-filler’ model, envisaging texts as a series of slots
which have to be filled from a lexicon which satisfies local restraints. At

229
Compare the controversy between the lexicalist and transformationalist hy-
potheses in word formation (> 9.6).
 10 Phraseology 137

each slot, virtually any word can occur. Since language is believed to oper-
ate simultaneously on several levels, there is a very complex pattern of
choices in progress at any moment, but the underlying principle is simple
enough.
Any segmental approach to description which deals with progressive
choices is of this type. Any tree structure shows it clearly: the nodes on the
tree are the choice points. Virtually all grammars are constructed on the
open-choice principle.
This open-choice principle Sinclair (1991: 110) contrasts with the principle
of idiom (which does not refer to idioms in the sense identified above but to
idiomaticity):
The principle of idiom is that a language user has available to him or her a
large number of semi-preconstructed phrases that constitute single choices,
even though they might appear to be analysable into segments. To some ex-
tent, this may reflect the recurrence of similar situations in human affairs; it
may illustrate a natural tendency to economy of effort; or it may be moti-
vated in part by the exigencies of real-time conversation. However it arises,
it has been relegated to an inferior position in most current linguistics, be-
cause it does not fit the open-choice model.
One very obvious type of evidence emphasizing the importance of the id-
iom principle comes from advanced foreign language teaching, where a
common comment of native speakers marking translations or essays is of
the type “it is not wrong but it does not sound English”. Experimental evi-
dence confirms this: for instance, Greenbaum’s sentence completion ex-
periments described in 10.1 were repeated with German students of Eng-
lish. Whereas 82% of Greenbaum’s native speaker subjects continued a
sentence beginning I entirely using the verb agree, only 23% of the Ger-
man students did so; in the case of I badly need the figures are 65% versus
11% (Herbst 1996: 390). The same point can be made on the basis of the
study of learner corpora (> 3.3.2.2); Guilquin (2007: 286) found colloca-
tional errors such as the following in a corpus of French-speaking learners
of English:
(11)QE They wanted to make an end to these conflicts … (mettre fin)
(12)QE Children are often unable to make the difference between fiction and
reality (faire la différence).
Detailed research of the valency structures of English verbs can be taken as
another source of evidence in favour of the idiom principle. Thus Klotz
(2000: 210) points out that there can be a correlation between the lexical
138 Meaning-carrying units

realizations of a valency slot and the valency patterns in which the slot
occurs. This often takes the form of tendencies such as ride a bike being
more common than ride on a bike.230
If one takes these ideas to the extreme one could argue that the idiom
principle as described by Sinclair should not only be taken to cover the use
of prefabricated items such as idioms, institutionalized collocations or sig-
nificant collocations, but that it might also comprise the idea that certain
concepts are more likely to be formulated one way rather than another in a
particular language. One indication of this is the fact that bilingual diction-
aries very often give an equivalent for a word such as zerquetschen in the
case of grind and then list certain ‘phrases’ or contexts in which another
equivalent would be more appropriate:
grind …  v.t. ...
C (rub harshly) zerquetschen; ∼ a cigarette end
into the ground einen Zigarettenstummel austreten231
Such equivalents, for which Herbst and Klotz (2003) introduce the term
probabemes, are not necessarily identical with any of the phraseological
units above but also contribute to the idiomatic character of language.232

10.6 Phraseological units

It can be said that the importance of multi-word units has been recognized
not only by linguists working within the traditional discipline of phraseol-
ogy, which put a strong focus on idioms, but also by corpus linguists, for-
eign language linguists and construction grammarians.233 However, it is
relatively difficult to identify classes of units that are relevant in this con-
text since phraseological units are subject to gradience in several respects.
In some cases, semantic transparency (or, rather, lack of semantic transpar-
ency) may be the defining criterion, others may be recognized as units on
the basis of frequency, some can be seen as extremely fixed, while others

230
Compare also Klotz (2000: 220–230). For the correlation of patterns and
nouns see also the collostructional approach developed by Stefanowitsch and
Gries (2003).
231
Duden Oxford Großwörterbuch Englisch 21999.
232
For a definition of probabemes and a discussion of faintest (or foggiest) idea
and German keine Ahnung see Herbst and Klotz (2003: 145–146).
233
For the role of multi-word units in construction grammar see e.g. Fillmore,
Kay, and O’Connor (1988) and Croft and Cruse (2004: 226–256) (> 13.2).
 10 Phraseology 139

show a higher degree of variation.234 Thus it is not surprising that many

classifications should have been suggested in the literature. The following
is a list of the categories (and examples) given by Granger and Paquot
(2008: 43–44):

1. Categories of referential phrasemes:

− lexical collocations (in the sense of institutionalized collocations as de-
scribed above): heavy rain, closely linked, apologize profusely
− idioms (to spill the beans, to let the cat out of the bag, to bark up the
wrong tree)
− irreversible bi- and trinominals (bed and breakfast, kith and kin, left,
right and centre)
− similes (as old as the hills, to swear like a trooper)
− compounds (black hole, goldfish, blow-dry)
− grammatical collocations (depend on, cope with, a contribution to,
afraid of, angry at, interested in), to which other valency relations (such
as manage to_INF etc) could be added235
− phrasal verbs (blow up, make out, crop up)

2. Categories of textual phrasemes

− complex prepositions (with respect to, in addition to, apart from, irre-
spective of)
− complex conjunctions (so that, as if, even though, as soon as, given
that)
− linking adverbials (in other words, last but not least, more accurately,
what is more, to conclude)

234
Compare Handl (2008), who makes a distinction between a semantic, a lexi-
cal and a statistical dimension in the analysis of collocation. For a survey of
approaches to phraseology see Gries (2008), Granger and Paquot (2008) and
also Gläser (21990).
235
Granger and Paquot (2008: 43) use the term grammatical collocations. Within
a construction grammar approach, there is very little point in distinguishing
between lexical expressions of valency patterns and more abstract categories
such as infinitives or clauses so that a category valency patterns might seem
more appropriate.
140 Meaning-carrying units

− textual sentence stems (the final point is …, another thing is …, it will

be shown that …, I will discuss …)

3. Categories of communicative phrasemes

− speech act formulae (good morning!, take care!, happy birthday!, you’re
welcome, how do you do?)
− attitudinal formulae (in fact, to be honest, it is clear that, I think that)
− commonplaces (Enough is enough, We only live once, It’s a small
world)
− proverbs (A bird in the hand is worth two in the bush, When in Rome)
− slogans (Make love, not war).

Recommended further reading:

− Schmid (forthc.)
− Gries (2008), Granger and Paquot (2008)
− Sinclair (2004)
 Sentences – models of grammar

11 Syntax: traditional grammar

11.1 Syntax and grammar

11.1.1 Descriptive frameworks

Syntax, the analysis of the structure of sentences, has been one of the main
points of interest in the study of language since the ancient Greeks. As a
result, we are faced with an enormous number of different approaches to-
wards the description of sentences. The frameworks developed by the
American structuralists during the first half of the twentieth century, the
version of Chomsky’s transformational grammar which dominated linguis-
tics during the sixties and seventies, Chomsky’s (1995) Minimalist Pro-
gram, theories such as lexical functional grammar,236 Hudson’s (1984)
word grammar, construction grammar (> 13.2) or dependency and valency
theory (> 12.3), to name just a few, all propose different ideas for the de-
scription of sentence structure.
Alongside the development of such syntactic theories, there is also the
very strong tradition of describing the “rules” of a particular language
which is commonly associated with the word grammar. This is the kind of
description usually found in the grammar books used at schools or that is
referred to in statements about grammatical properties of words in diction-
aries. Much of this framework has emerged from the analysis and the teach-
ing of languages such as ancient Greek and Latin over the centuries. It is
obvious (to linguists) that many of the categories established by traditional
grammar – such as the well-known distinction between transitive and in-
transitive verbs, the terms gerund or future tense – really appear to be inap-
propriate for the description of present-day English or German (> 2.3).
Nevertheless, one should be aware of the fact that up to this day traditional

236
Cf. Kaplan and Bresnan (1982). See also Sells (1985: 132–190).
142 Sentences – models of grammar

grammar determines linguistic thought to a considerable degree: all the

theories mentioned above identify word classes such as noun and verb or
make use of a category subject and indeed most would use the term object.
This is not to say, of course, that all representatives of these theories would
have exactly the same concept of an object; quite the contrary.
While models of syntax such as the Minimalist Program or dependency
theory, for which the term grammar is also sometimes used (in the sense of
‘model’), are usually concerned with principles of syntactic organization in
general, grammars in the more traditional sense aim at a more or less com-
prehensive description of a particular language.237 Modern descriptive
grammars of this kind can be seen as being firmly rooted in the tradition of
grammar writing, while however incorporating insights, concepts and ter-
minology of various approaches of modern linguistics and avoiding the
inadequacies and pitfalls of traditional grammar.
The most important descriptive grammar of English at the beginning of
the twenty-first century is probably the Comprehensive Grammar of the
English Language, abbreviated as CGEL, which was published by
Randolph Quirk, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik in
1985. It is the largest and most detailed of a number of grammars by this
team of researchers, who were all involved in the Survey of English Usage
(> 3.3.1) and are sometimes referred to as the London School. A grammar
of comparable size and complexity with contributions of thirteen authors
was published in 2002 under the title of Cambridge Grammar of the Eng-
lish Language (CamG), edited by Rodney Huddleston and Geoffrey Pul-
lum.
Since the London School has influenced linguistic research and the
teaching of English all over the world, the terminology and the categories
employed in CGEL will be outlined in this chapter.

11.1.2 Sentence and clause

Syntax is concerned with the way in which words combine to form larger
units such as sentences. Although texts or utterances often consist of more

237
For a highly interesting discussion of different uses of the term grammar see
Greenbaum (1988: 20–26), who distinguishes senses such as: (i) “a general
theory of language description”, (ii) “a theory for describing one language”,
(iii) “a book about English grammar”, (iv) “the contents of the books (‘a
grammar is a book about grammar’)” etc.
 11 Syntax: traditional grammar 143

than one sentence, it makes sense, for certain purposes of analysis at least,
to consider the sentence the largest unit of description.238 Basically, all
theories of syntax, i.e. traditional grammar, structuralist syntax, or genera-
tive approaches, although recognizing the fact that sentences can link with
each other to form texts, treat the sentence as the basic unit of investigation.
One reason that is often given to justify the sentence as the highest unit of
analysis is that the formation of sentences is rule-governed in a way that the
formation of texts is not (> 18).
However, the situation with the term sentence is not unlike that with the
term word described in 9.1 in that sentence is a linguistic term used in eve-
ryday language but linguists do not necessarily find it easy to define. While
in the written language sentences are divided by full stops, the notion of
sentence is very difficult to define when we consider spoken language (cf.
CGEL 2.11), as can be seen from the following extract from a corpus of
spoken language:239
(1)QE Well I had some people to lunch on Sunday and – they turned up half an
hour early – ... – I mean you know what [g] getting up Sunday’s like
anyway and – I’d – I was behind in any case – and I’d said to them one
o’clock – and I almost phoned them up and said come a bit later – and
then I thought oh they’ve probably left by now – so I didn’t – and –
twelve thirty – now that can’t be them – and it was – ....
As will become clear in this chapter, sentences can be described as consist-
ing of clauses and phrases: if we proceed from top to bottom, a hierarchy of
the following kind can be established:
sentence ⊇ clause / phrase ⊇ word ⊇ morpheme
This means: a sentence can consist of one or more clauses, a clause can
consist of one or more phrases, a phrase can consist of one or more words,
and a word can consist of one or more morphemes.
As is pointed out by Quirk, Greenbaum, Leech and Svartvik (CGEL
2.11), the clause is a much more clearly definable unit of linguistic analysis
than the sentence. A clause can occur as

a dependent (or subordinate) clause if it forms a constituent of a larger
clause as in
(2)PW Barbara Hepworth had also begun to work in metal.

238
For problems connected with the definition of sentence see CGEL 2.11.
239
Cf. Leech and Svartvik (21994: 11).
144 Sentences – models of grammar

an independent clause if it is not a constituent of a larger clause as in

(3)PH We shall now watch New York as eagerly as Paris for new develop-
ments.
The term sentence will be used

for a single independent clause that is not co-ordinated with another
clause as in (3).

for a combination of two or more independent clauses in co-ordination
as in
(4)PH It was an October night and a full moon was rising over Godrevy.

11.1.3 Subject and predicate

When it comes to analysing clauses, the most immediate distinction that

can be made is that between subject and predicate. In a way, one can
consider those two elements as essential parts of an active declarative
clause such as
(2)PW Barbara Hepworth had also begun to work in metal.
The distinction between subject and predicate was already established in
traditional grammar. However, it is also perfectly in line with structuralist
principles: firstly, Barbara Hepworth and had also begun to work in metal
represent the immediate constituents of the clause, i.e. if asked to divide
the clause into two parts one would draw the boundary between Hepworth
and had and not anywhere else:240
Barbara Hepworth  had also begun to work in metal
not: BarbaraHepworth had also begun to work in metal
not: Barbara Hepworth had  also begun to work in metal
not: Barbara Hepworth had also  begun to work in metal
etc.
Secondly, from a purely formal point of view, i.e. disregarding aspects of
meaning, both Barbara Hepworth and had also begun to work in metal can
be exchanged for other elements (with which they stand in the relation of

240
See Matthews (1981: 73–74). See also Hockett (1958).
 11 Syntax: traditional grammar 145

commutation) without affecting the acceptability of the sentence. Struc-

turally, (2) can be reduced to
(2a) She  worked.
or expanded to
(2b) Barbara Hepworth, whose former studio in St. Ives is now open to the
public, had also begun to work in metal.
In a way, the distinction between subject and predicate reflects the fact that
subject and predicate represent the structural minimum of active declarative
clauses, and, indeed, many other types of clause in languages such as Eng-
lish or German.
Of course, the analysis is not always as straightforward as in the case of
(2). For instance, a sentence such as
(5)PW At this time Ben Nicholson and Barbara Hepworth identified completely
with the European avantgarde.
does not lend itself so easily to a division into two parts: one would obvi-
ously hesitate to take at this time as a part of the subject since At this time
Ben Nicholson and Barbara Hepworth do not commute with any other
single element. At the same time, for positional reasons alone it would not
be particularly satisfactory to regard it as a part of the predicate. Thus, one
might find it preferable to separate (5) into three parts:
At this time  Ben Nicholson and Barbara Hepworth  identified com-
pletely with the European avantgarde.
Unfortunately, there is no general agreement about the use of the term
predicate in linguistics: while Robins (21971: 240) says that “predicate may
be used to refer to the rest of the sentence apart from the subject”, Lyons
(1968: 334) takes a narrower view of the predicate and arrives at a threefold
distinction of the parts of the sentence:241
It is a fundamental principle of traditional grammar, and also of much mod-
ern syntactic theory, that every simple, declarative sentence consists of two

241
Note that this definition is in line with the use of the term predicate in CGEL
(2.47), where buys her vegetables in the market and arrived late today are
given as examples of predicates. However, Matthews (1981: 97–99) identi-
fies three types of predicate: those consisting of a verb and an object, those
consisting of a copula and a predicative, and those consisting merely of an in-
transitive verb.
146 Sentences – models of grammar

obligatory major constituents, a subject and a predicate; and that it may

contain, in addition, one or more adjuncts. Adjuncts (of place, time, man-
ner, reason, etc. …) are optional, or structurally dispensable, constituents of
the sentence: they may be removed without affecting the remainder of the
sentence.
A distinction between what Lyons (1968: 334) calls the “nucleus” of the
sentence, the part of the sentence that consists of subject and predicate, and
the extranuclear part or parts made up of adjuncts, seems more useful for
many purposes of analysis. This approach to the analysis of clauses is taken
by Aarts and Aarts (1982/21988) and in certain versions of valency the-
ory.242 Within this terminology, the following elements of a clause can be
identified:243
Subject: a clause constituent

which is obligatory in an active declarative clause

which is realized by a noun phrase (> 11.3.1) or can be replaced by a
noun phrase

which shows concord (i.e. the same value with respect to singular or
plural) with the verb of the predicate.
Predicate: a clause constituent

which is obligatory in an active declarative clause

contains a verb (and possibly further elements) with the verb showing
concord (i.e. the same value with respect to singular or plural) with the
subject.
Adjunct: a clause constituent which is optional from a structural point of
view.
Within this framework, which will be expanded below, the following
clause constituents can be identified for sentences (6), (2), and (7):

242
Cf. Herbst et al. (2004) and 12.3.
243
Note that these definitions are to some extent prototypical in character, see
Herbst and Schüller (2008: 4–9 and 18–21). For a more precise characteriza-
tion of adjuncts see 11.2.2 and 12.3.2.
 11 Syntax: traditional grammar 147

(6)PH We were actually in St. Ives when I

painted them
SUBJECT PREDICATE ADJUNCT PREDICATE ADJUNCT
 1  2

(2)PW Barbara had also begun to work in

Hepworth metal
SUBJECT PREDICATE ADJUNCT  PREDICATE

(7)TS Following Ben Nicholson’s he painted a number of land-

first visit to St. Ives scapes and sea paintings of
Porthmeor beach.
ADJUNCT SUBJECT PREDICATE

11.2 The elements of clause structure in CGEL

11.2.1 Elements of clause structure as functional units

Despite the advantages of an analysis of clauses in terms of subject, predi-

cate and adjuncts as outlined above, CGEL employs a slightly different
framework. In this grammar, clauses are analysed in terms of the following
five elements of clause structure:

subject

verb (which here is a functional term not identical with the word class
verb)

object: direct object and indirect object

complement: subject complement and object complement

adverbial
Since, with the exception of the subject, these elements of clause structure
can be used to further analyse the predicate into smaller constituents, for
example, the two approaches are highly compatible.
148 Sentences – models of grammar

What is important to realize about these elements of clause structure is

that they are functional categories. This can easily be demonstrated by
showing that the same formal element – such as colour in
(8a)PH Colour exists in itself, has its own beauty.
and
(8b)PH I used colour as a means of expressing my emotion and not as a tran-
scription of nature.
can fulfil different functions in different clauses (such as that of subject in
(8a) and object in (8b)). Looking at the word colour in isolation (i.e. taken
out of context) it could not be attributed any function, whereas its formal
properties such as that it is a noun consisting of six letters can be deter-
mined irrespective of any particular occurrence.

11.2.2 Criteria for the distinction between different elements of

clause structure

Whereas in a language such as German case plays a very important role in

determining the elements of clause structure, this is a very problematic
criterion for a language such as modern English, where most inflexional
endings have vanished. This is why other criteria have to be used for dis-
tinguishing between different elements of clause structure, as is outlined in
detail by Quirk, Greenbaum, Leech and Svartvik (CGEL 10.6).
CGEL, in each case, draws upon criteria from different levels of linguis-
tic analysis. In the case of the subject, these are, for instance:244

morphological in that pronouns functioning as subjects take subjective
case (he, she, they etc.)

positional in that subjects normally precede the verb

semantic in that subject status often coincides with a semantic role such
as AGENTIVE
In contrast, objects and complements (called predicatives in CamG
2002: 251–253) usually follow the verb and, if realized by pronouns, take

244
See CamG (2002: 236–239).
 11 Syntax: traditional grammar 149

objective case (him, her, them etc.).245 The distinction between objects and
complements is made on the basis of two main criteria:

objects can become the subjects of passive clauses, complements cannot

complements refer in some way to the subject or object of the clause
(hence the distinction between subject complements and object
complements), objects do not.
Thus, one of the leading artists of his generation in (9a) refers to the sub-
ject of the clause, Patrick Heron, and thus is a subject complement.
(9a)PH Patrick Heron is one of the leading artists of his generation.
It cannot become the subject of a passive clause such as
(9b) *One of the leading artists of his generation was been by Patrick Heron.
This is different in the case of an object such as a number of landscapes
and sea paintings of Porthmeor beach in
(7a)TS Following Ben Nicholson’s first visit to St. Ives, he painted a number of
landscapes and sea paintings of Porthmeor beach.
where a passive is possible:
(7b)PH They are painted in a variety of methods.
Furthermore, complements can also be realized by adjectives, whereas ob-
jects cannot:
(10)PH That period at St Ives was idyllic.
The two criteria distinguishing objects and complements can also be illus-
trated by looking at a sentence such as (11), which contains an object (him)
and an object complement (the best art critic to have emerged in London
since Roger Fry):
(11a)PH I considered him then the best art critic to have emerged in London
since Roger Fry.
Only the object can be made a subject in a passive clause such as
(11b) He was considered the best art critic to have emerged in London since
Roger Fry,

245
In CamG (2002: 455–456) the traditional terms nominative and accusative
are used for forms such as I and me.
150 Sentences – models of grammar

but only the object complement can be replaced by an adjective:

(11c) I considered him excellent.
If two objects occur, the first is the indirect object, the second the direct
object. Both are also characterized semantically in CGEL (10.7). Direct
objects are described as typically referring to an ‘entity that is affected by
the action denoted in the clause’, whereas indirect objects are characterized
as realizing ‘an animate being that is the recipient of the action’.
(12)PH Mrs. van der Straeten has written me (indirect object) a very nice letter
… (direct object).
Finally, adverbials, which in CGEL (2.15) are described as “a heterogene-
ous category”, are defined by a number of criteria, such as a tendency to-
wards general mobility within the sentence, a tendency to provide informa-
tion about temporal, spatial or other circumstances etc.:
(13)PH A full moon was rising over Godrevy.
(14)PH I shall never forget the immense sensation of space the first moment we
entered that room, at the end of our journey from London.
(15)PH Also established at the time I painted it was my habit of applying one
coat of colour over the white ground.
(16)PH Ben Nicholson and Barbara Hepworth had been living in Carbis Bay, a
little to the East of the town ...
It is important to understand that the classification provided by CGEL is
very much in line with the categories and also the principles of traditional
grammar. In particular, one should note that the distinctions are not made
on the basis of a single level of linguistic analysis but that both formal and
semantic criteria are being used. As a result, the categories established must
be seen as exemplifying prototypes and not as clear-cut entities separated
by rigorous boundaries. Thus, for instance, a long tradition as an artists’
community in
(17)PW By 1918 St Ives had a long tradition as an artists’ community
would be analysed as an object according to CGEL, since, although the
sentence cannot be passivized, the semantic relationship between tradition
and have meets the description of an object and not that of a comple-
ment.246

246
For a very similar treatment see CamG (2002: 246). See also 11.2.4.
 11 Syntax: traditional grammar 151

11.2.3 CGEL’s clause types

Quirk, Greenbaum, Leech and Svartvik (CGEL 2.16) draw upon the ele-
ments of clause structure identified in order to establish seven clause types
in English.247

clause verb class (CGEL) valency

type (> 12.3)
SV She is reading intransitive 1
SVC This is an island copula 2
SVA She was there copula 2
SVO She is reading a book monotransitive 2
SVOO He gave her the book ditransitive 3
SVOC She called him a fool complex transitive 3
SVOA She put the book on the table complex transitive 3

These clause types are then taken as the basis for classifying verbs as

intransitive

transitive
monotransitive

ditransitive

complex transitive or

copular.
This classification is based on a traditional and well-established distinction:
the term transitive indicates that a verb takes an object. It is subclassified
here into three different types of transitivity, depending on whether it oc-
curs in an SVO, a SVOO or an SVOA structure.248

247
See CamG (2002: 218–219) for the distinction between intransitive, complex
intransitive, monotransitive, ditransitive and complex (mono)transitive and
their relation to valency.
248
It is worth noting, however, that the distinction between the seven clause
types results in a classification of only five types of verb, since copular verbs
represent clause types SVA and SVC and complex transitive verbs represent
types SVOC and SVOA respectively.
152 Sentences – models of grammar

It is important to note that Quirk, Greenbaum, Leech and Svartvik

(CGEL 2.16) only subsume a clause containing an adverbial under the SVA
or SVOA type if that adverbial is obligatory. As a consequence,
(13)PH A full moon was rising over Godrevy
belongs to clause type SV since the adverbial is optional, whereas
(16)PH Ben Nicholson and Barbara Hepworth had been living in Carbis Bay, a
little to the East of the town
is an example of SVA since the adverbial cannot be left out (at least not
without changing the meaning of the verb).249

11.2.4 Problems of traditional terminology

Categories such as object or subject complement or transitive and intransi-

tive verb are widely used in traditional grammar, lexicography and also
more recent theories. Nevertheless, it should be pointed out that these cate-
gories are by no means unproblematic because their use varies considera-
bly. For instance, Aarts and Aarts (1988: 131–132, 138) take passivization
as a defining criterion of the category direct object, which, however, con-
tradicts the use of the term in CGEL (16.26–27), where
(18)QE They have a nice house.
is given as an example of a “monotransitive verb” with “a noun phrase as
direct object” although it does not “normally” have a passive.250 The diffi-
culty of applying the term object to such cases arises from the fact that in
CGEL these categories are defined by formal and semantic criteria (>
11.2.2), which do not always coincide.
A further example of this is presented by sentences such as
(19)BNC He’s in a good mood.

249
Making this distinction between obligatory and optional adverbials introduces
a distinction which has the same effect as that between predicate and adjuncts
established in 11.1.3. What is referred to as an optional adverbial in CGEL is
classified as an adjunct if one wishes to make the predicate-adjunct distinc-
tion, whereas CGEL’s obligatory adverbials are part of the predicate.
250
For a discussion of so-called middle verbs such as have, possess, resemble
see also CGEL 10.14. See also 11.2.2.
 11 Syntax: traditional grammar 153

Semantically, in a good mood meets the description of subject comple-

ments given in CGEL (10.8), but it hardly fits the formal characteristics
listed there: “normally a noun phrase or an adjective phrase, but it may also
be a nominal clause”.251 It is for this reason that some approaches avoid the
use of terms such as object (cf. Halliday 1994: 26, Herbst and Schüller
2008) and replace them by terminology which keeps the levels of form and
meaning more strictly apart (> 12.4).252

11.3 Phrases

11.3.1 Types of phrase

The next question to address is which formal elements of a sentence can

actually function as an element of clause structure, i.e. occur as subjects,
objects, etc. In the terminology employed in traditional approaches such as
CGEL, this can either be another clause, as in:
(20)PW Peter Lanyon felt strongly that figuration and abstraction were not
incompatible [that-clause as direct object]
(21)PW When he moved to St Ives [wh-clause as adverbial], Nicholson acquired
one of the large Porthmeor studios.
(22)PW Still living in Carbis Bay [-ing-clause functioning as adverbial] ... for a
time Barbara Hepworth came each day to the studio by taxi, and then
she started to live there during the week [infinitive clause as direct ob-
ject]
or a unit generally identified as a phrase:
(23)PW The town of St Ives [noun phrase] stands [verb phrase] on the Atlantic
coast of the Penwith district of west Cornwall [prepositional phrase]
(24)AE During the summer [prepositional phrase as adverbial] St. Ives [noun
phrase as subject] has attracted [verb phrase as verb] many visitors
[noun phrase as direct object]

251
Compare, however, CGEL (10.11)’s discussion of “prepositional phrases
[that] are semantically similar to adjective or noun phrases functioning as
complement” and of cases such as They are in love as subject complements.
252
Compare also the classification proposed by Meyer (2009).
154 Sentences – models of grammar

A phrase is characterized by the fact that it can take a particular slot in the
structure of a sentence. Most types of phrase, the so-called endocentric
phrases, can be realized by a single word, which on its own can fulfil the
same syntactic function in a sentence:
(25a) Painters
(25b) Many painters felt attracted to the place
(25c) The painters who moved to St Ives after the
war
A word which can stand for a phrase on its own is called its head, endocen-
tric phrases are also called headed phrases. Such headed phrases can con-
tain a number of other elements, which are structurally dependent on the
head, and which are known as modifiers.
For English, the following types of phrase are often identified, the first
four are headed, the fifth type is non-headed (or exocentric).253

1. Noun phrases

determi- premodifier head: noun or postmodifier

native pronoun
the town of St Ives
many very interesting paintings showing St Ives Bay
a spectacular view over the bay
the one that I mean
St Ives which is an artists’ community
they
this fascinating book on linguistics

253
It should be noted that the account of phrase structure terminology employed
here is not always generally accepted. Within a dependency framework, for
instance, it is more convincing to see the relations holding between the ele-
ments making up a phrase as dependency relations. Also, the distinction be-
tween headed and non-headed phrases is by no means uncontroversial. See,
for instance, Hudson (1993) and 12.4.2.1.
 11 Syntax: traditional grammar 155

The head of a noun phrase is a noun or a pronoun. It can be modified by

adjective phrases, prepositional phrases, relative clauses, adverb phrases,
etc. Apart from modifiers, noun phrases can also contain determinatives
(for which the term determiner is also often used). Although there is a cer-
tain overlap between modification and determination, the typical function
of modifiers is to further describe what is expressed by the head, whereas
the sole function of determiners is to establish reference, i.e. to identify the
elements in the real world to which the phrase can be applied. Thus, in the
case of a phrase such as this fascinating book on linguistics, the function of
this can be seen as pointing out or identifying which of all the possible
books is referred to in a particular context; this thus fulfils a determinative
function, whereas fascinating and on linguistics describe the book more
closely and are thus classified as modifiers.

2. Verb phrases
The term verb phrase is used in different ways in linguistics (sometimes to
refer to the whole predicate). In the more traditional approach of CGEL, for
instance, what is meant by verb phrase is the complex of a verb and the
auxiliary verbs that are used to express such grammatical categories as
aspect (simple/progressive; non-perfective/perfective), modality (possibil-
ity, obligation, etc.) or voice (active/passive). In CGEL, the terms main
verb and auxiliary verb are used to describe the functions of head and
premodifier in the case of the verb phrase:

auxiliary verbs main verb

paints
is painting
has been painted
may have been painted
156 Sentences – models of grammar

3. Adjective phrases

premodifier head: adjective postmodifier

beautiful
very nice
friendly to him
too far to walk

In some cases, as in too far to walk, pre- and postmodifier have to be seen
as one complex modifier, which can then be called a discontinuous modi-
fier (Aarts and Aarts 21988: 63).

4. Adverb phrases

premodifier head: adverb postmodifier

beautifully indeed
very well
much more slowly than they had expected

5. Prepositional phrases
Prepositional phrases consist of a preposition and a prepositional comple-
ment. Both these elements are obligatory in the sense that neither can repre-
sent a prepositional phrase on its own. Prepositional phrases are thus non-
headed.

preposition prepositional complement

to the island
on going there

Except for the rather special use of the term verb phrase, these categories
outlined above or similar categories are fairly established in linguistics,
although a number of alternative views have also been suggested (>
12.4.2).
 11 Syntax: traditional grammar 157

11.3.2 The role of the phrase

While categories such as sentence or word also play an important role in

general and non-academic discussions of language, the phrase does not.
Nevertheless, the phrase is to be seen as a central unit of linguistic descrip-
tion. In particular, it must be borne in mind that it is through noun phrases –
and not through nouns as words – that speakers can refer to concrete people
or objects of the real world (> 14.2). A word such as book in itself evokes a
certain idea about the kind of object a speaker is talking about in the
hearer’s mind, but it does not enable them to identify a concrete object in
the particular situation of communication. This is quite clearly different in
the case of a noun phrase such as this book, which in the present context
enables you to identify the object referred to as the one you are just read-
ing. Thus the notion of the phrase is highly relevant to syntactic and seman-
tic analysis.
There is also psycholinguistic evidence to suggest that phrases are an
important unit of language perception, for instance. Experiments were car-
ried out in which people were asked to listen to recorded sentences, into
which distorting signals called clicks were inserted. The test subjects were
then asked to locate the clicks in the sentence.254 Interestingly, subjects
were able to identify the location of all clicks at phrase boundaries cor-
rectly, while there was a tendency to move the location of clicks within a
phrase towards a phrase boundary.

11.4 Word classes

11.4.1 Criteria for the establishment of word classes

If one aims at describing the way words combine to form sentences it is

desirable to be able to make generalizations. Thus, little is gained by saying
that a word such as the can be followed by the word professor or the word
book or the word scheme or the word island etc. It seems more satisfactory
to state that a word such as the can be followed by words belonging to a
particular class, namely nouns.
Such labels also play a major role in language teaching and in dictionar-
ies: if, for example, a word such as flight attendant is classified as a noun,
this can be taken as a shortcut for listing a number of properties such as the

254
See Clark and Clark (1977: 53–55) and Wimmer and Perner (1979: 126).
158 Sentences – models of grammar

fact that it can be preceded (a) by words such as the or (b) by an adjective
and (c) that it can take a plural form:
(26a) the flight attendant
(26b) a friendly flight attendant
(26c) flight attendants
However, it is important to realize that some features commonly associated
with a particular word class do not apply to all members. Thus, for exam-
ple, words such as advice or psychiatry are marked as uncountable nouns in
the Longman Dictionary of Contemporary English (LDOCE5) or the Ox-
ford Advanced Learners’ Dictionary (OALD7) to indicate that they do not
have plural forms. It is thus essential to know how a word class label is to
be interpreted in a particular framework: it can either mean – in a kind of
prototype approach – to indicate that a particular word shares more features
with other members of this class than with members of any other class, or it
can mean that the word has all the features listed for this particular class.
In the history of linguistics, different types of criteria have been used for
the establishment of different word classes. Semantic criteria, which were
often used in traditional approaches, have turned out not to be particularly
suitable.255 If one tries to distinguish between nouns and verbs, for instance,
by saying that nouns refer to ‘persons, things or ideas’ and verbs denote
‘actions’, one is immediately faced with problem cases such as nouns de-
noting actions such as arrival, birth, development or verbs not denoting
actions such as be, belong, resemble.256 It is obvious that even a modifica-

255
Cf. CGEL (2.43), Palmer (1971), Huddleston (1984). For problems of the
traditional classification of word classes see Robins (1971: 218–219). Com-
pare Herbst and Schüller (2008: 30–36). Compare also Behrens (2005: esp.
182–183), also with respect to the role of word classes in cognitive linguis-
tics.
256
Cf. Allerton (1990: 89): “... the noun hesitation, the verb hesitate and the
adjective hesitant show clearly that different word classes can be endowed
with the same basic meaning; and yet in traditional grammar we are given
mainly semantic definitions of the ‘parts of speech’ (= Latin ‘partes ora-
tionis’, as word classes are usually termed). This is mainly due to the influ-
ence of the Roman grammarians Palaemon and Priscian, who (unlike their
more enlightened predecessor Varro) not only insisted on finding eight word
classes, just because Greek had eight ... but also defined their classes on a
purely semantic basis. This is unfortunate, because, for instance, although
verbs, e.g. hesitate, are meant to be the words that designate an activity,
 11 Syntax: traditional grammar 159

tion of the semantic descriptions provided for nouns or verbs will not solve
the problem but only result in different borderline cases.
It thus seems reasonable to draw the distinction between different word
classes on formal criteria such as the distribution of words or their morpho-
logical properties. In the case of present-day English, morphological crite-
ria are less useful in that respect than in the case of inflected languages
such as Latin, Russian, Old English or even German. Nevertheless, the
criterion of which kinds of endings particular words take can be usefully
applied in present-day English to the definition of verbs, nouns, and also
adjectives.257 It is interesting to see that different models tend to agree with
respect to the identification of these three classes, whereas there are re-
markable differences as far as number and character of the other word
classes are concerned.

____________________________
nouns like hesitation do so just as much, and even if the adjective hesitant
does, as required, denote a quality, the derived noun hesitancy does so
equally. It is true that prototypical nouns (such as author, town, word) desig-
nate a person, place or thing (or more generally an ‘entity’), that verbs (like
abscond, write and sit) designate an event, process or state, and that most ad-
jectives denote qualities. The difficulty is with the large number of abstract
nouns, verbs and adjectives.” Cf. also Robins (1966/2000) and Pullum
(2009).
257
Interestingly enough, in attempts at automatic classification of words with
respect to their word classes in computational linguistics, the morphological
criterion can be extended to cover certain orthographic regularities. Thus,
Garside (1987: 37), describing the probabilistic approach to parsing devel-
oped at the University of Lancaster, describes as one factor relevant to ma-
chine-based classification of word classes the so-called suffixlist: “The suf-
fixlist contains sequences of up to five letters, including ‘suffixes’ in the or-
dinary sense, such as -ness (noun), but also any word endings which are asso-
ciated invariably (or at least with high frequency) with certain word classes,
for example -mp (noun or verb) – the letters -mp do not constitute a morpho-
logical suffix but it is a fact that almost all words ending with these letters are
either nouns or verbs (the few exceptions such as damp or limp are listed in
the lexicon). The suffixlist is searched for the longest matching word-ending.
Thus there are entries in the list for -able (adjective), -ble (noun or verb), and
-le (noun), and these will be tested for in that order; exceptions (such as cable
or enable) are in the lexicon.”
160 Sentences – models of grammar

11.4.2 CGEL’s word classes

One established model of word class categorization is that of the Compre-

hensive Grammar of the English Language by Quirk, Greenbaum, Leech
and Svartvik (1985: 2.34), where the following word classes – or parts of
speech – are distinguished:258

I NOUN John, room, answer, play (> 11.4.1)

II ADJECTIVE happy, steady, new, large, round (> 11.4.4)
III FULL VERB search, grow, play (> 11.4.3)
IV ADVERB steadily, completely, really
V PREPOSITION of, at, in, without, in spite of (> 11.4.7)
VI PRONOUN he, they, anybody, one, which (> 11.4.6)
VII DETERMINER the, a, that, every, some (> 11.4.6.)
VIII CONJUNCTION and, that, when, although (> 11.4.7)
IX MODAL VERB can, must, will, would (>11.4.3)
X PRIMARY VERB be, have, do (> 11.4.3)
XI NUMERAL one, two, three; first, second, third
XII INTERJECTION oh, ah, ugh, phew
words of unique function not (negative particle), to (infinitive marker)

The first four of these word classes – nouns, full verbs, adjectives and ad-
verbs – are often referred to as open classes because the number of words
they comprise is not restricted and new members can constantly be added.
The number of prepositions, pronouns, determiners, conjunctions, modal
and primary verbs, however, is limited and not easily subject to change.
These word classes are thus called closed classes.259

258
Criteria for the definition of these word classes are discussed in the sections
listed in the table. See also the relevant chapters in CGEL or CamG. For a de-
tailed discussion of the properties of different word classes see Herbst and
Schüller (2008: 30–75).
259
For the distinction between autosemantic and synsemantic words see Mair
(2008: 73).
 11 Syntax: traditional grammar 161

11.4.3 Verbs

Morphological criteria serve rather well to define a word class full verbs
(CGEL 1985) – or lexical verbs (CamG 2002; Herbst and Schüller 2008) –
in present-day English. A full verb or lexical verb can be characterized by
the following properties:

It has a base form (speak) which can occur on its own (used in the in-
finitive and all present tense forms with the exception of the third person
singular).

It has a specific form for third person singular present tense; marked by
the suffix {S} (speaks).

It is tensed in that it is morphologically marked for present tense or past
tense and shows tense distinctions, i.e. it possesses a past tense form
(spoke).

It can form a present participle with the morpheme {-ing} (speaking).

It can form a past participle (spoken).
For some lexical verbs, some of these forms coincide: thus, in the case of
paint, the past tense form (painted) is identical with the past participle form
(painted); similarly, verbs such as put or cut do not show a contrast be-
tween the two tenses except in the third person singular.
The following forms can thus be identified for lexical verbs:260

base form paint, put, speak, write

{S}-form paints, puts, speaks, writes finite
past tense form painted, put, spoke, wrote finite
-ing-participle/present part. painting, putting, speaking, writing non-finite
-ed-participle/past participle painted, put, spoken, written non-finite

260
See Herbst and Schüller (2008: 39). For a discussion of finite and non-finite
see CamG (2002: 1173) and CGEL (1985: 3.52). The base form is often ana-
lysed as a finite form in declarative-‘statement’-constructions such as This
does not mean that they put themselves into the hands of an absolute author-
ity.BNC and as non-finite when used in imperative-‘directive’-constructions
such as Put them in now, before autumn gales and winter blasts make the
plants rock at their roots.BNC This is not entirely unproblematic since there is
no morphological difference between these forms. Compare also: The argu-
ments put forward can be interesting in themselves.BNC
162 Sentences – models of grammar

It seems appropriate to distinguish two further types of verb from the class
of lexical (or full) verbs (CGEL 3.21), namely

modal verbs (can, may, shall, will, could, might, should, would and
must), which are characterized by the fact that they do not admit suffixa-
tion by {S} and do not have participle forms, and

the primary verbs be (which, by the way, has eight different morpho-
logical forms), do and have, which can occur as auxiliaries or main
verbs in the verb phrase.

11.4.4 Central and peripheral members of word classes – word

classes as prototypes

It is important to realize that no single criterion seems to suffice to define a

word class. Although the {ing}-suffix can be used for the characterization
of verbs, this does not mean that any form containing –ing could or should
necessarily be classified as a verb:
(27a)BNC That bloody bird has been annoying me for days.
(27b)BNC And she knew she was annoying them whenever she questioned their
assumptions.
(27c)BNC This was extremely annoying.
(27d)BNC The most annoying fact was that it could and should have been a mas-
terpiece.
In (27c) and (27d) annoying should be classified as an adjective rather than
a verb because it occurs in positions that are typical of adjectives (such as
clever):
(27e)BNC Coleridge had developed an apparently relaxed, but in fact extremely
clever style of blank verse.
Adjectives can generally be characterized by the following four criteria:261

attributive use (I) annoying questions, clever girl

predicative use (after seem etc.) (II) This was extremely annoying,
this seems clever ...

261
See Herbst and Schüller (2008: 58), CGEL (1985: 7.3) and CamG (2002:
526–542).
 11 Syntax: traditional grammar 163

premodification by adverbs such as extremely annoying, very clever

very (III)

comparative and superlative forms (IV) darker, most annoying
Again, it is important to point out that these four criteria do not apply to all
adjectives alike, as can be seen from the following examples:
(28) *an afraid person/state
(29) *The fool/failure seems utter
(30) *These two expressions are very synonymous
(31) *These two words are more homophonous than those two.
CGEL (7.3) establishes a gradient of the following kind to illustrate this
phenomenon:

I II III IV
hungry + + + +
central adjectives
infinite + + - -
old (as in old friend) + - + +
afraid ? + + +
peripheral adjectives
utter + - - -
asleep - + - -
soon - - + +
adverbs
abroad - - - -

Adjectives that can occur in attributive and predicative position are consid-
ered central adjectives in CGEL; adjectives that occur only in one of these
positions peripheral adjectives. In terms of prototype theory (> 16.4), one
could say that the prototypical adjective shows all four criteria whereas
more marginal members of the category do not have all these features. If a
word does not meet all the criteria that characterize the prototypical mem-
bers of that word class, it may be necessary to determine its word class with
respect to the greatest similarity to any prototype it may show.
164 Sentences – models of grammar

11.4.5 Multiple-class membership

In other cases, of course, it may also be convenient to attribute a word or, to

be precise, a particular word form, dual or multiple class membership.
Thus, there would be little controversy about the question that the form
painting in
(32a)PH … his paintings of the room and window at St Ives had been done from
the experience of a sea-level interior in a town
can be regarded as a noun, whereas in
(32b)PW It came as a great shock, so much so that I actually stopped painting
myself
it can be considered a verb. In fact, it is quite common to assign dual word
class membership to particular word forms:
(27a)BNC That bloody bird has been annoying me for days. verb
(27b)BNC This was extremely annoying. adjective
(33a)BNC It was the car that annoyed her. verb
(33b)BNC We are very annoyed. adjective
(8a)PH Colour exists in itself, has its own beauty. noun
(8b)BNC Brushes may also be used to colour larger areas such verb
as the blue ribbon of this design.
(34a)BNC Come in! adverb
(34b)BNC the Barbara Hepworth Museum in St Ives preposition
A particularly interesting example of multiple class membership is pre-
sented by round, which can be analysed as belonging to no less than five
word classes:
(35a)BNC Would you like to come round for a cup of tea? adverb
(35b)BNC It’s just round the corner. preposition
(35c)BNC They’d have their round hat boxes… adjective
(35d)BNC He enjoyed a round of golf and playing cricket … noun
(35e)BNC You round the Point to make your way south to Douglas verb
— first on beaches, then on cliff and meadow paths.
 11 Syntax: traditional grammar 165

11.4.6 The distinction between determiners and pronouns

How many word classes a word belongs to is not just a question of the
properties of the word as such but also depends on how one analyses the
word or the language in question. Thus, while it is probably relatively un-
controversial that a word such as interest has two clearly distinct uses – one
corresponding to a prototypical verb and one corresponding to a prototypi-
cal noun – this is not quite so obvious with words such as this, these, those,
each or all.
(36a)WG This was followed in the 1960s by a series of paintings …
(37a)WG These were necessary to ensure that there was no overlap where two
colour-areas touched …
(38a)WG It is impossible to discuss Heron’s paintings, particularly those of the
1970s, without drawing on his own writings about them.
(39a)WG Each new period … seemed to involve an abandonment of all that had
gone before, so that each could be said to mark a true beginning again.
In these uses, the words underlined fall under the category of pronoun, even
in terms of the most traditional definition of a pronoun “substituting a
noun”. This in (36a), like these in (37a) and the second each in (39a), rep-
resent a whole noun phrase, those in (38a) and all in (39a) function as
heads of noun phrases. This is different with the following uses of the same
words:
(36b)PW The prehistoric standing stones of the Penwith peninsula were the cata-
lyst for this transformation.
(37b)PW These years brought artistic success for Wood.
(38b)WG … Heron had always been irritated by those literal-minded critics who
insisted in seeking references to the Cornish landscape (…) in the
shapes and contours of his paintings.
(39a)WG Each new period … seemed to involve an abandonment of all that had
gone before, so that each could be said to mark a true beginning again.
(39b)WG All Heron’s canvasses of the 1970s were painted using small, delicate
Japanese watercolour brushes.
(39c)WG His work of all periods retains a remarkable freshness.
Here, this, these etc. do not represent the heads of noun phrases (as defined
above) but rather they have a determinative function within those noun
166 Sentences – models of grammar

phrases in the sense that it is these which determines the range of referents
of the noun phrase these years. This function is identical to that of the
words that in traditional grammar were called articles:
(36b)PW The prehistoric standing stones of the Penwith peninsula were the cata-
lyst for this transformation.
Words such as all, each or this can thus function as heads of noun phrases,
as in the first group of examples, or occur as determinatives within a noun
phrase, as in the second group. There are two ways of dealing with this:
either one takes the fact that these words can occur in different functions in
a noun phrase as a property of the word classes under which these words
were traditionally treated or one establishes a separate word class deter-
miner and argues that when a word such as these occurs in a sentence such
as (37a), it is a demonstrative pronoun, whereas in a use of the type exem-
plified by (37b), it is a determiner. Many modern grammars of English,
amongst them CGEL, have opted for this second approach262 and estab-
lished determiners as a word class. Allerton (1990: 90) justifies this on the
following grounds:
In languages like English, it is worth distinguishing a class of ‘determiners’,
embracing the articles and words such as any and its possible replacements
..., i.e. words such as the, a, any, some, all, my, your, this, that. The words
my, your, etc. are traditionally called either pronouns or adjectives: but they
differ from true pronouns like mine, yours (which stand for a whole noun
phrase) in that they occur with a noun: equally they differ from adjectives ...
This means that most words which fall under the category of determiner
actually have to be assigned dual class membership, as indicated above.263
The alternative option in terms of a classification of these words is not
to assign multiple class membership to such words. Thus Aarts and Aarts
(1982/1988: 51) classify this, that, these and those as demonstrative pro-
nouns and say that they “function as constituents of the sentence or in the
structure of the noun phrase”. Herbst and Schüller (2008) go a step further
and make a three-fold distinction between

262
Compare, however, the approach taken by Aarts and Aarts (21988), who use
the term determiner to refer to the function within the noun phrase that CGEL
calls determinative.
263
For a subclassification of determiners see, for instance: Leech and Svartvik
(21994: 269).
 11 Syntax: traditional grammar 167

pure pronouns (which comprise all words which are classified as pro-
nouns but not as determiners in CGEL, i.e. words such as someone, eve-
rybody or you that can function as the head of a noun phrase)

pure determiners (which comprise the words a, an, little, no, which can
only occur as determiners but not as pronouns) and

determiner-pronouns (which like this, that, each show both uses).

11.4.7 The distinction between prepositions and subordinating

conjunctions

The concept of multiple class membership is used in CGEL and many other
models to account for the different uses of words such as before in exam-
ples of the following kind:
(40a)PH Several years before this Heron had adopted the practice of giving his
paintings titles almost factually descriptive ...
(40b)PW St Just is the last town in England before you reach Land’s End.
(40c)WG In 1967, two years before he embarked on this series, Heron broke a leg
in an accident …
(40d)WG He was using a fuller palette of colours, too, with the introduction of
paler tints … that he had never before dared attempt.
Traditionally, these three uses of the word before are analysed as represent-
ing three different word classes:
− in (40a) before is classified as a preposition because it is followed by a
noun phrase,
− in (40b) and (40c) it is seen as a subordinating conjunction because it
introduces a clause, and
− in (40d), in which it is neither followed by a noun phrase nor a clause, it
is classified as an adverb.264
Again, the alternative is to simply see (40a–d) as different uses of one word
and assign it to one word class. In fact, one can argue that the differences
displayed by these examples find a direct parallel in verb complementation,
where quite a large number of verbs show transitive and intransitive uses.

264
Note that before also allows premodification, e.g. by two years in (40c) and
never in (40d).
168 Sentences – models of grammar

For these reasons, CamG (2002: 612–617) subsumes all uses of words such
as since and before under one word class label, namely that of prepositions.
This means that most subordinating conjunctions and a number of words
that have traditionally been categorized as adverbs are classified as preposi-
tions in CamG.265 Following a similar line of argument, Herbst and
Schüller (2008) introduce a word class particle as a cover term for all words
that traditionally are classified as prepositions or subordinating conjunc-
tions as well as some adverbs.266 Particles are seen as valency carriers (>
12.3.4) in this model so that the different uses exemplified by (40a–d) are
described as valency properties of these particles.

11.4.8 Word classes in English

It follows from what was said above that there is no obvious and not only
one answer to the question of how many word classes a language has and
how they ought to be distinguished. This is caused in part by the prototypi-
cal nature of word classes, i.e. the fact that it is not always possible to draw
clear dividing lines between different classes, but it also has to do with the
fact that different analysts come to different conclusions as to what is to be
considered the most appropriate way of analysing particular facts.

265
See CamG (2002: 600) for the parallels between She was eating an apple. –
She was eating. and I haven’t seen her since the war. – I haven’t seen her
since. For a discussion of this cf. also Pullum (2009).
266
Note that in this approach words such as up and here are excluded from the
adverb category because they can function as valency complements (>
12.3.2) of verbs such as be, place or put if they cannot function as premodifi-
ers of adjectives or adverbs (Herbst and Schüller 2008: 59). This is based on
the approach taken in CamG (2002: 602, 612–617). See Herbst and Schüller
(2008: 59–60) for a definition of adverbs in terms of criteria such as that they
can function as premodifiers of adjectives (very interested), other adverbs
(however slightly), that they can occur immediately before verbs (I was just
explaining …), etc. Furthermore, words derived from adjectives by {-ly}
count as adverbs.
 11 Syntax: traditional grammar 169

CGEL CamG Herbst and

(1985) (2002) Schüller (2008)

full verb lexical verb lexical verb write, say ...

auxiliary verb:
primary verb primary verb be, have, do
non-modal
can, could, may, might,
auxiliary verb:
modal verb modal verb shall, should, will,
modal
would, must
conference, idea, pan-
noun noun lexical noun
cake ...
determiner/ determinative/
the, a
article article determiner
every
determiner determinative some (as pre-head of
determiner-
NP)
pronoun
pronoun pronoun some (as head of NP)
clever, intelligent,
adjectives adjectives adjectives
friendly, ...
adverb adverb furiously, very, well
pronoun pronoun yesterday, today
adverb
here, somewhere
since
preposition
preposition since + noun phrase

particle since + finite clause

(subordinating)
conjunction that/whether + finite
subordinator
clause
infinitive infinitival
to + infinitive clause
marker subordinator
(coordinating) (coordinating)
coordinator and, but, or, nor
conjunction conjunction
interjection interjection interjection oh, ...

It is important to remember that word classes provide useful labels for some
purposes of linguistic description, but that they “tend in fact to be rather
170 Sentences – models of grammar

heterogeneous, if not problematic categories” (CGEL 2.41). In fact, it may

be doubted whether the relevant aspects of the use of all words can be de-
scribed appropriately by assigning them membership of a particular word
class.
Nevertheless, word class labels are widely used in linguistic description.
In many dictionaries, for instance, word classes are taken as one of the
prime principles of structure because it is assumed that it speeds up the
looking-up process considerably if the noun uses of a word such as interest
are kept apart from the verb uses of interest, for example. Again, however,
it is debatable whether this applies to all word class distinctions in the same
way. It is interesting to see that with respect to words such as each or this
the more recent editions of English learners’ dictionaries such as Longman
Dictionary of Contemporary English (LDOCE5) or the Oxford Advanced
Learner’s Dictionary (OALD7) tend not to make a systematic distinction
between e.g. a pronoun this and a determiner this but simply give both
word class labels without indicating which word class label applies to
which example.267
This of course raises the question to what extent word classes represent
concepts that are psychologically relevant. Garman (1990: 140) comes to
the following tentative conclusion:268
... if word class turns out to be a determinant of lexical accessibility, it will
be likely to prove a highly indirect one – a conclusion which in no way runs
from the apparent fact that words like a are very different from those like
match. One respect in which they are different is frequency of occurrence:
is this likely to be a factor? ... If frequency of a word has anything to do
with its accessibility, it may completely obscure the distinction between
open and closed classes.

267
LDOCE5 gives “determiner, pron, adv” for each and distinguishes between
this1 (“determiner, pron”) and this2 (adverb). This is treated similarly in
OALD7, where, however, They lost 40 each is subsumed under “det, pron”.
268
For aspects of language acquisition see Behrens (1999) and (2005). Compare
in this context Tomasello (2003: 45): “… virtually no one believes that adult
part-of-speech categories are relevant to children just beginning to learn lan-
guage”. It would be interesting to see to what extent this is also true of some
of the distinctions discussed in 11.4.6. and 11.4.7.
12 Valency theory and case grammar

12.1 Two types of hierarchy

12.1.1 Constituency

One way of analysing a sentence is to establish a hierarchy of elements that

takes the complete sentence or clause as the highest element and finishes at
the level of the word. For a sentence such as
(1)PW By 1918 St Ives had a long tradition as an artists’ community
this kind of a hierarchy can be represented in the following form of a struc-
tural tree:

This kind of tree structure, which represents the analysis of a sentence into
its constituents, represents a part-whole relationship. Grammars based on
this kind of analysis are known as phrase structure grammars or con-
stituency grammars. In various forms, mostly based on the principle of
172 Sentences – models of grammar

binary divisions, such phrase structure analyses are characteristic of Ameri-

can structuralism, but also of early phases of generative transformational
grammar.269 For instance, in Syntactic Structures, Chomsky (1957: 26)
formulates a phrase structure grammar in terms of so-called rewrite rules:

(i) Sentence
NP + VP
(ii) NP
T + N
(iii) VP
Verb + NP
(iv) T
the
(v) N
man, ball, etc.
(vi) Verb
hit, took, etc.

This can be seen as a description of the sentence

(2)QE The man hit the ball

which can also be represented in the form of a tree diagram:270

269
For the principles of phrase structure trees cf. Matthews (1981: 73–78) and
Vennemann (1977).
270
Cf. Chomsky (1957: 27) and Allen and van Buren (1971: 24–25). For a dis-
cussion of the limitations of pure phrase structure grammars and their status
 12 Valency theory and case grammar 173

12.1.2 Dependency

The principle of constituency is not the only one that can be employed in
the analysis of sentences, however. An alternative is to establish a hierarchy
of elements that are at the same level of a constituency analysis by looking
at the way in which they depend on one another.271 Matthews (1981: 78)
describes the principles underlying dependency theory by referring to the
traditional notion of government:
In the traditional language of grammarians, many constructions are de-
scribed in terms of a subordination of one element to another. A verb is said
to ‘govern’ its object … A preposition is also said to govern the noun which
follows.
This notion of government can perhaps more convincingly be demonstrated
in languages where it manifests itself in terms of case marking. Thus, in the
case of
(3)NF Am 17. Mai 1797 beschlossen die ostfriesischen Stände unter ihrem
Vorsitzenden, dem Grafen Edzard Mauritz zu Inn- und Knyphausen, die
Errichtung einer Seebadeanstalt auf Norderney.272
one could argue that the object die Errichtung einer Seebadeanstalt auf
Norderney is governed by the verb beschließen in that it determines its
form, in this case requiring the object to be in the accusative case. A verb
such as zustimmen would govern a dative, for example:
(3a) Am 17. Mai 1797 stimmten die ostfriesischen Stände unter ihrem Vor-
sitzenden, dem Grafen Edzard Mauritz zu Inn- und Knyphausen, der Er-
richtung einer Seebadeanstalt auf Norderney zu.

____________________________
within early transformational models see Chomsky (1957). See also Lyons
(31991: 56–65).
271
Hudson (1976: 197) points out that the constituency approach is “typically
American” and “underlies all the main American schools of syntactic theory
(so-called immediate constituent analysis, transformational grammar, stratifi-
cational grammar and tagmemics) – and, also, incidentally, other versions of
‘systemic’ grammar”, whereas the dependency approach is “typically Euro-
pean”.
272
Translation: (3) On the 17th of May 1797, the East Frisian estates, under the
leadership of Count Edzard Mauritz zu Inn- und Knyphausen decided to es-
tablish a seaside spa on Norderney./ (3a) … agreed to establishing …
174 Sentences – models of grammar

In the same way one could see ihrem Vorsitzenden as governed by the
preposition unter, which requires a dative. Despite the limited amount of
case marking in present-day English, the notion of government is appropri-
ate to describe whether a verb can be followed by a noun phrase, an -ing-
clause or a to-infinitive-clause or to describe the relationship between
nouns, adjectives and verbs and the prepositions which they allow for their
complementation patterns:
(4)PH This is the true significance of his preference for the earlier of the
Rothkos in the ICA show.
(5)PH In spite of misgivings about his ‘muted’ colour, he now seemed to
Heron to be ‘the best of living Americans …’.
Such “forms of subordination”, as Matthews (1981: 79) points out, are of-
ten dealt with under the name of dependency, because this kind of analy-
sis reveals how the elements of a sentence depend on one another.273 The
first main representative of a model of grammar based on dependency rela-
tions, dependency grammar or dependency theory, Lucien Tesnière
(1959), introduced so-called stemmata to illustrate these relationships:274

A dependency stemma is different from a phrase structure tree in that it

does not represent a part-whole relationship and does not attribute the high-
est rank to the whole clause or sentence. Rather, it is the governing word
that takes the highest place in the hierarchy. One important characteristic of
many dependency approaches distinguishing them from a constituency type
of analysis is that they do not attribute any special status to the subject.
Since the verb is generally seen as the highest element of the sentence, the
value of the traditional distinction of subject and predicate, which is also
reflected in an immediate constituents analysis, is rejected, as, by the way,

273
For the relationship between Rektion and Dependenz see also Ágel (2000:
47–79).
274
Cf. Tesnière (1959/22000: 526).
 12 Valency theory and case grammar 175

is the principle of binary division underlying the principles of immediate

constituent analysis (> 11.1.3).275
There has been considerable discussion in linguistics as to whether con-
stituency analysis and dependency analysis are to be seen as competing or
complementary models of analysis. On the one hand, a dependency stemma
is superior to a constituency tree because it reveals relationships between
the elements of a clause which are not brought out at all in the constituency
analysis. On the other hand, it must not be overlooked that a dependency
analysis presupposes an analysis into constituents because it is precisely
those constituents whose relationships with each other it describes. For this
reason, many linguists today would argue that the two approaches comple-
ment rather than exclude each other.276
If one accepts the view that dependency relations hold between elements
which belong to the same hierarchical level of a constituency analysis, the
dependency relationships holding between the phrases of a sentence such as
(1)PW By 1918 St Ives had a long tradition as an artists’ community.
could be represented as follows:

275
Cf., for instance, Tesnière (1959/22000: 525–526).
276
Cf. Matthews (1981: 84–93), Herbst, Heath, and Dederding (1980: 44–45),
Brinker (1977: 91), Vennemann (1977), Hudson (1976: 197–206), Baum-
gärtner (1970), Heringer (1970 and 1993).
176 Sentences – models of grammar

12.1.3 Case grammar and valency theory

This chapter will deal with two approaches to syntax which combine ele-
ments of dependency and constituency and which attribute a central role to
the verb in the structure of the sentence – case grammar and valency the-
ory. Whereas case grammar was originally formulated by Charles Fillmore
in a famous article called ‘The case for case’ (1968) in the general frame-
work of generative linguistics, valency theory, which features very strongly
in the analysis of German, is strongly related to dependency theory and the
ideas of Tesnière (1959). Despite the different theoretical backgrounds of
the two theories there are many parallels between them: in fact, they deal
with the same aspect of linguistic structure from different perspectives in
that the main concern of case grammar is semantic, whereas the starting
point of (most approaches to) valency theory is formal. In a way, both theo-
ries can be related to Tesnière’s (1959/2000: 522) comparison of the verbal
core to a play on the stage, where he makes a distinction between verb,
actants and circonstants – the verb corresponding to the action, actants to
the actors and circonstants to the circumstances.277

12.2 Case grammar: semantic roles

12.2.1 Basic principles of case grammar

The basic idea of case grammar as proposed by Fillmore (1968) can be

illustrated by considering sentences such as
(6a)QE John opened the door.
(6b)QE John opened the door with the key.
(6c)QE The key opened the door.
(6d)QE The door opened.
What is interesting about these sentences is that it can be argued that the
elements showing the same type of underlining make the same semantic
contribution to the overall meaning of the sentence irrespective of whether
they occur as the subject, the object or as a prepositional phrase. This has

277
For a discussion of different types of valency such as semantic or syntactic
valency see Helbig (1992: 7–10).
 12 Valency theory and case grammar 177

led Fillmore (1968: 24) to establish a system of what he (misleadingly)278

called case relationships:
The case notions comprise a set of universal, presumably innate, concepts
which identify certain types of judgments human beings are capable of
making about the events that are going on around them, judgments about
such matters as who did it, who it happened to, and what got changed.
In ‘The Case for Case’, Fillmore (1968: 24–25) identifies six such cases:
AGENTIVE (A) the case of the typically animate perceived insti-
gator of the action identified by the verb
INSTRUMENTAL (I) the case of the inanimate force or object causally
involved in the action or state identified by the
verb
DATIVE (D) the case of the animate being affected by the state
or action identified by the verb
FACTITIVE (F) the case of the object or being resulting from the
action or state identified by the verb, or under-
stood as a part of the meaning of the verb
LOCATIVE (L) the case identifies the location or spatial orienta-
tion of the state or action identified by the verb
OBJECTIVE (O) the semantically most neutral case, the case of
anything representable by a noun whose role in
the action or state identified by the verb is identi-
fied by the semantic interpretation of the verb it-
self; conceivably the concept should be limited to
things which are affected by the action or state
identified by the verb.
In (6ab), John could be described as realizing the semantic case of
AGENT, the door that of OBJECTIVE and the key that of INSTRUMENT.
A verb such as open can thus be characterized as having a case frame of the
following kind (Fillmore 1968: 27):
+ [____ O (I)(A)]
This means that if only the OBJECTIVE case is realized, it becomes the
subject, if the INSTRUMENTAL is also realized, it becomes the subject
and if the AGENTIVE is realized, it becomes the subject of the clause.279

278
The term case is usually used for morphological case such as nominative,
accusative etc.
279
Compare Fillmore’s (1968: 33) rule for “‘unmarked’ subject choice”.
178 Sentences – models of grammar

12.2.2 Advantages and drawbacks of case grammar

Although case grammar was originally developed within the framework of

Chomsky’s generative transformational grammar (> 13.1), the idea of deep
cases was adopted in many other models, for instance in valency theory (>
12.3 and 12.4), often under the more appropriate names of semantic
cases, semantic roles or participant roles (as well as theta roles/θ-
roles in generative work). This is not surprising since the model is rather
attractive in a number of respects:
− Semantic roles provide a relatively satisfactory labelling system for the
description of certain semantic properties of valency complements.
− The model is able to formally relate sentences such as (7a – 7d):
(7a)BNC I (A) nearly broke a chair (O) with my hands (I) as I listened.
(7b)BNC If we (A) broke a window (O), it was accidental.
(7c)BNC One tankard hit the ceiling, another (I) broke a window (O).
(7d)BNC The window (O) broke because John threw a ball at it.
− The model can account for restrictions in coordination of the type:
(6e) *Mr Davey (A) and the key (I) opened the case.
(7e) *Mary Yellan (A) and the rock (I) broke the window.
− The meanings of different verbs can be related:
(8a) He (A) showed them (D) Jamaica Inn (O).
(8b) They (D) saw Jamaica Inn (O).
On the other hand, the model also suffers from a number of problems. First
of all, the correspondences outlined above cannot necessarily be general-
ized. Thus
(7f)BNC ... they were breaking the law and prepared to face the consequences.
(7g)BNC … these people broke the agreement …
are not paralleled by
(7h) *The law broke.
(7i) *The agreement broke.
 12 Valency theory and case grammar 179

Furthermore, not all restrictions on coordination can be explained by means

of different semantic roles. Palmer (21981: 148) points out that
(9)QE *I saw Helen and a football match.
“is a very strange sentence, yet both Helen and a football match are here in
the object case.” On the other hand, coordination of noun phrases repre-
senting different semantic roles can be found:
(7j)BNC Someone (or something) has broken the kitchen window.
What is perhaps more important is that it is by no means always clear
which semantic role to assign to a particular valency complement in a
clause. Thus while the with-complements in
(10a)BNC Severe forearm bruising may be covered with both a crêpe bandage and
a shin pad ....
(10b)BNC Once dug and left to dry, potatoes can be sorted and stored in teachests,
thick paper sacks or shallow boxes covered with sacking to keep the
light from turning them green.
can clearly be assigned the semantic role of INSTRUMENT, this is far less
clear in cases such as
(10c)BNC The walls are covered with posters illustrative of various radical causes
– nuclear disarmament, women’s liberation, the protection of whales ....
(10d)BNC ... her kitchen cupboard is covered with postcards, love letters from old
boyfriends and local newspaper articles.
(10e)BNC Dorothy said her lawn was covered with snow when she arrived yester-
day morning.
The main problem of any case grammar or semantic role approach is the
difficulty of clearly delimiting semantic categories, which is also described
by Palmer (21981: 148–149):
… case grammar runs into the familiar difficulty of the vagueness of se-
mantic categories. Often it will be difficult to decide, on semantic grounds,
what is the case of a particular noun phrase. Fillmore sees the smoke as ob-
ject in The smoke rose, and the same would be true of the wind in The wind
blew. But what, then, shall we say of The smoke rose and blotted out the
sun, The wind blew and opened the door?
How justified Palmer’s objections are can be seen by the fact that even
almost forty years after the publication of Fillmore’s (1968: 25) article, in
180 Sentences – models of grammar

which he said that “additional cases will surely be needed”, various models
with varying numbers and kinds of semantic role have been suggested.280 It
would certainly be naive to expect that there could ever be anything such as
a finite inventory of such roles. Despite these apparent difficulties one must
not overlook the fact that semantic roles, especially when taken to be proto-
types rather than clearly distinct categories, present a useful concept for the
analysis of certain syntactic problems.

12.2.3 Some useful participant roles

One way of avoiding the problems that inevitably arise with general seman-
tic roles is to make use of much more specific roles which only apply to
one particular verb or a limited number of verbs. Thus within Fillmore’s
FrameNet project verbs are attributed to certain frames, which are charac-
terized by frame elements resembling semantic roles. FrameNet (frame-
net.icsi.berkeley.edu) includes a verb such as paint in a frame Cre-
ate_physical_artwork:281
(11)PH I probably painted more St Ives harbour window paintings in London
than in Cornwall.
This frame is then described as having as core frame elements Creator (I)
and Representation (more harbour window paintings). While such specific
descriptions may be useful for certain purposes of linguistic description, it
has to be said that they do not provide the kind of generalization which is
desirable for other purposes. It may thus be advisable to apply rather spe-
cific role labels whenever general role labels do not seem to be appropriate
but to make use of more general labels whenever they seem unproblematic.
Thus in the case of example (11), a case can be made out for using a label
such as ‘AGENT’ and ‘EFFECTED’. Herbst and Schüller (2008) list a number
of participant roles which can be applied fairly generally, for instance:

‘BENREC’: someone or something that benefits from the action ex-
pressed by the verb or is the intended recipient of the action described
by the verb:

280
See for instance CGEL (1985), CamG (2002), Dirven and Radden (1977) or
Haegeman (1991).
281
Compare Fillmore (2007) for an outline of the FrameNet project, in which
much more specific frame elements are used.
 12 Valency theory and case grammar 181

(12)PW It gave Ben (‘BENREC’) a reasonable studio and Barbara (‘BENREC’)

one near the kitchen, filled with light.
Whenever this seems appropriate for the purposes of the description, a dis-
tinction can be made between ‘BENEFICIARY’ and ‘RECIPIENT’:
(13)NW I’ll write you (‘BENEFICIARY’) a glowing reference …
(14)PH Zennor, he wrote to Middleton Murray and Katherine Mansfield
(‘RECIPIENT’) in the spring of 1916, ‘is a most beautiful place: ...’.

‘ÆFFECTED’: something (or someone) that is affected by or is the out-
come or result of the action described by the verb
(15)PH Once there, Heron had immediately started work on a series of small
vertical paintings ...
If appropriate, the more specific labels ‘AFFECTED’ and ‘EFFECTED’ can
be employed:
(16)PH At Zennor one sees infinite Atlantic (‘AFFECTED’)...
(17)PH The house (‘AFFECTED’) has been neglected ...
(11)PH I probably painted more St Ives harbour window paintings (‘EFFECTED’)
in London than in Cornwall.
(18)PH Red Ground (‘EFFECTED’) was painted in May 1957 ...

‘PREDICATIVE’: identification, equation, characterization or description
(which could be seen as subroles of ‘PREDICATIVE’):
(19)PH Heron has never been a programmatic artist ...

‘LOCATIVE: STATIVE’:
(16)PH At Zennor one sees infinite Atlantic …

‘LOCATIVE: GOAL’:
(20)PH ... Heron moved beyond the rapid summary strokes and washes of the
‘stripe’ and ‘horizon’ paintings towards what he described as a ‘true
spontaneity’ ...
(21)PH Exactly forty years before D.H. Lawrence had moved into Higher Tre-
gerthen, the tiny group of cottages at the bottom of the hill below Eagles
Nest ...
182 Sentences – models of grammar

‘LOCATIVE: SOURCE’:
(22)PH ... Heron moved beyond the rapid summary strokes and washes of the
‘stripe’ and ‘horizon’ paintings towards what he described as a ‘true
spontaneity’ ...
(23)PH ... the change from London to Zennor was registered immediately in the
bright colour, loose structures and atmospheric translucencies of the
‘garden’ paintings ...

‘TIME’:
(21)PH Exactly forty years before D.H. Lawrence had moved into Higher Tre-
gerthen, the tiny group of cottages at the bottom of the hill below Eagles
Nest ...

‘PURPOSE’:
(24)PW In order to realize the subject emotionally and to further his under-
standing of the composition, Lanyon made use of his practical skills in a
number of three-dimensional constructions.

‘TOPIC’:
(25)PH Heron had, of course, been thinking and writing about the nature of
abstraction since his very first published article in 1945 ...

‘AGENT’: someone that carries out an action282
(25)PH Heron had, of course, been thinking and writing about the nature of
abstraction since his very first published article in 1945 ...
While these participant roles can be used in the description of quite a few
cases, this list is not intended to provide a complete framework for all par-
ticipant roles required for the description of English verbs. Rather, the idea
is that these roles must be supplemented by other possibly more specific
roles in many cases.

282
The notion of AGENT is rather problematic. CamG (2002: 230–231) treats
‘AGENT’ as a “subtype of causer”. Compare Palmer (1994: 25) and also Gold-
berg (2006: 184–186).
 12 Valency theory and case grammar 183

12.3 The basic principles of valency theory

12.3.1 Introduction

While valency theory, as was pointed out above, is probably the most
widely used approach in the analysis of German (see Helbig 1992 or Ágel
2000), there are only relatively few attempts to use it as a descriptive
framework for English. It has to be said that no generally accepted version
of the theory has emerged as yet as the first two models that offer a fairly
comprehensive treatment of English verb valency – Emons (1974, 1978)
and Allerton (1982) – differ in many respects. For this reason, the follow-
ing outline will be mainly based on the version of valency theory that pro-
vides the basis of the Valency Dictionary of English (2004) (VDE) and the
approach taken by Herbst and Schüller (2008).

12.3.2 Complements and adjuncts

Like case grammar, valency theory attributes a central role to the verb in
determining the structure of clauses but takes formal considerations as its
starting point. Valency analysis concentrates on the relationships that hold
between a valency carrier (sometimes also called predicator) and those
elements whose occurrence in a sentence is related to the presence of that
valency carrier, i.e. those elements which can be considered to be governed
by it. With reference to a sentence such as
(1)PW By 1918 St Ives had a long tradition as an artists’ community.
this means that the verb have is to be seen as the valency carrier because
the occurrence of St Ives and a long tradition as an artists’ community can
be regarded as being dependent on the verb. This can be shown by a dele-
tion test:
(1a) *By 1918 St Ives had.
(1b) *By 1918 had a long tradition as an artists’ community.
By 1918, however, is independent of the verb have and can be deleted:
(1c) St Ives had a long tradition as an artists’ community.
This difference is reflected terminologically in the distinction between
complements and adjuncts:
184 Sentences – models of grammar

Complements (Ergänzungen) 283 are constituents of a clause

that are either required by the valency of the governing predicator

or determined in their form by the predicator.
Adjuncts (freie Angaben) are constituents of a clause

that are neither required by the valency of the governing predicator

nor determined by it in their form.
To indicate the different kinds of dependency relation holding between
complements and adjuncts and the governing element, different types of
line might be used in the stemma representing (1) (cf. p. 175):

While the notion of adjunct largely corresponds to that of adjunct in other

frameworks (or to that of “optional adverbials” in CGEL), the term com-
plement is used in valency theory to cover all constituents that are depend-
ent on a valency carrier (i.e., for instance, CGEL’s subject, object, obliga-
tory adverbial etc.).
Finding tests to establish the distinction between complements and ad-
juncts has played a major role in the development of German valency the-

283
The German terms are given here because the German terminology is much
more established than the English. Matthews (1981: 123) draws a distinction
between complements and peripheral elements while Allerton (1982) uses the
term verb elaborator for Ergänzung. The term complement in valency theory
must not be confused with complements in other theories, especially the sub-
ject and object complements of CGEL.
 12 Valency theory and case grammar 185

ory.284 On the whole, it has to be said that there is no one single criterion on
which the distinction can be based and that it must probably be seen in
terms of gradience. However, adjuncts typically meet the following criteria:

Adjuncts tend to show greater positional mobility than complements.
(1d) St Ives had a long tradition as an artists’ community by 1918.

Adjuncts can often be left out without the remaining sentence being
ungrammatical.
(1e) St Ives had a long tradition as an artists’ community.
It is important to note that deletability is not a reliable criterion for adjunct
status since adjuncts, like complements, can be necessary from a communi-
cative point of view. Thus it would not be possible to delete the adjuncts in
(16) or (18) in the contexts of (16a) or (18a) (> 12.3.5):
(16)PH At Zennor one sees infinite Atlantic...
(16a) Where does one see infinite Atlantic?
(18)PH Red Ground was painted in May 1957 ...
(18a) When was Red Ground painted?
A further criterion of adjunct status that is sometimes discussed is that of
free addibility, meaning that (as long as they are semantically compatible)
adjuncts can be added to sentences irrespective of the valency of the
verb.285

12.3.3 Qualitative and quantitative aspects of valency

One of the reasons why valency theory (and dependency theory in general)
attributes a central role to the verb in syntactic analysis is that the form of a
clause is largely determined by the verb in that it depends on the verb as to
how many and which complements can occur in a clause. Thus a verb such
as enjoy can be followed by a noun phrase complement and a [V-ing]-
clause-complement but not by a [to_INF]-complement:

284
For a discussion of tests to distinguish between complements and adjuncts
see e.g. Helbig and Schenkel (21973: 31–40), Helbig (1992: 72–87), Emons
(1974: 65–105), Somers (1987: 11–18) and Herbst and Schüller (2008: 113–
116).
285
See Heringer (1996: 195).
186 Sentences – models of grammar

(26a) She enjoyed the book.

(26b) She enjoyed going to the theatre.
(26c) *She enjoyed to go to the theatre.
In this respect, which can be referred to as qualitative valency, enjoy differs
from a verb such as want, for example:
(27a) She wanted the book.
(27b) *She wanted going to the theatre.
(27c) She wanted to go to the theatre.
Which type of complement can be used with a particular verb can thus be
seen as an item-specific property of the verb. There are similar differences
with respect to the number of complements that can occur with a particular
verb are concerned. One can make a distinction between

monovalent uses
(28a)PH Christmas Eve was painted at the invitation of the Arts Council for
the Festival of Britain Exhibition 60 Paintings for 51.
(28b)PH Abstract art never interested me. I have always painted realistically.

divalent uses
(11)PH I probably painted more St Ives harbour window paintings in Lon-
don than in Cornwall.
(29)PH Heron has said that he stopped making the ‘stripe’ paintings when
he realized that they resembled the sunset skies over the sea at Zen-
nor.

trivalent uses
(30)PW For two years this gave him studio facilities and a small stipend.

tetravalent uses
(31)SP I’ll bet you all the expenses of the experiment you cant do it.
Abstracting from such uses, statements can be made about the maximum
number of complements a lexical unit can occur with.286 Thus verbs such as

286
VDE furthermore contains statements about the minimum valency of verbs in
active and passive uses.
 12 Valency theory and case grammar 187

give and show can be said to have a maximum valency of 3 as illustrated by

(30) or
(32)PW Mondrian showed us how completely possible it was to find a personal
equation.
whereas realize has a maximum valency of 2:
(29)PH Heron has said that he stopped making the ‘stripe’ paintings when he
realized that they resembled the sunset skies over the sea at Zennor.
and malfunction one of 1:
(33)BNC But the rocket malfunctioned.
This aspect of valency, quantitative valency, makes clear why the term
valency was chosen to describe this linguistic phenomenon since the role of
predicators in (nuclear) sentences can be compared to the capability of an
atom to combine with one or more other atoms to form a molecule.

12.3.4 Valency carriers

The property of valency cannot only be attributed to verbs, but also to other
word classes such as adjectives or nouns: in
(34)PH This is true of his writings about his own paintings.
the phrase of his writings about his own paintings can be analysed as a
complement of the adjective true, and about his own paintings as a com-
plement of writings.
The elements underlined can be regarded as complements of the words
in bold type because the possibility of the occurrence of these phrases can
be seen as item-specific properties of these words. This is not the case with
modifiers, which are not determined by the valency of the governing word:
(35)PH This, as we have seen, is not entirely true of Heron’s production in that
period ...
(36)PH It goes without saying that the immediacy of sensational impact of
which he is writing is only possible in the actual relation of spectator to
painting.
The notion of verb, adjective and noun valency is generally accepted.
Herbst and Schüller (2008), following the extension of the word class
188 Sentences – models of grammar

preposition in CamG (2002), also apply it to the word class particle (>
11.4.7):
(37a)PW St Just is the last town in England before you reach Land’s End.
Irrespective of word class it is important to realize that valency is not a
property of words, but a property of lexical units, i.e. combinations of word
forms and senses (> 9.1.1): thus, in the case of deny, for instance, VDE
makes a distinction between two lexical units ‘say that something is not
true’ and ‘refuse’, which differ in the kinds of complements they allow.
(38)VDE He denied that the country was facing an economic crisis. (‘not true’)
(39)VDE This means courts can obtain important evidence which previously
would have been denied to them. (‘refuse’)

12.3.5 Components of a valency description

On the basis of these considerations, valency can be defined as follows

(Herbst and Schüller 2008: 108):
A lexical unit has the property of valency if it opens up one or more
valency slots which can or must be realised by a complement.
A valency description of a lexical unit can then comprise the following
components:

level of form: specification of the complements that can fill a valency
slot – a complement being any formal realization of a valency slot (e.g.
a phrase or a clause)

level of semantics: specification of the participants or participant roles,
which characterize the semantic function of the complement in the
clause (> 12.2.3)

level of optionality: specification of whether (or under which contextual
conditions) a valency slot must or can be realized by a complement
Like Fillmore’s (1968) case frames, valency descriptions of this kind are to
be seen as an account of the syntactic potential of a lexical unit. For in-
stance, (in a simplified account of the verb’s valency structures) one can
identify two valency slots for a verb such as paint, which can be attributed
the participant roles of I ‘AGENT’ and II ‘ÆFFECTED’.
As far as the description of the complements is concerned, one would
have to say that valency slot I can be expressed by a noun phrase [NP] in
 12 Valency theory and case grammar 189

active clauses and by a [by_NP]-complement in passive clauses, whereas

valency slot II can be realized by an [NP] in the predicate in active clauses
and a subject [NP] in passives:
(11)PH I probably painted more St Ives harbour window paintings in London
than in Cornwall.
(28a)PH Christmas Eve was painted at the invitation of the Arts Council for the
Festival of Britain Exhibition 60 Paintings for 51.
In the case of watch, valency slot II (‘AFFECTED’) can be realized by a
range of different complements such as
[NP] During the week I can watch all the soaps on TVVDE
[how-CL] If you can spare the time, spend a day watching how the light changes
in strength and direction.VDE
[NP INF] Today you can watch the catch come in (approx 7:30–8 am) at Hayle, St
Ives, Newlyn and Mousehole, as fishing boats bring back dover sole and
haddock.VDE
in active clauses and by [NP]s in passives:
[NP] We are being watched.VDE
It is remarkable that in English the fact whether a complement can function
as the subject of a clause or is realized in the predicate is an important part
of its description.287 In a language such as German, in which morphological
case plays a much more prominent role than in English, terms such as Ak-
kusativergänzung or Nominativergänzung can be used, which implicitly
contains information about possible subjects. It is for this reason that the
description of complements provided in the complement inventories of
verbs in the VDE combine information about morphological form such as
[NP], [to-INF] with information about the possibility of occurrence as a
subject in active and passive clauses.288

287
Herbst and Schüller (2008: 117) use a subscript act-subj for complements that
occur as subjects of active clauses and a subscript pass-subj for complements
that occur as subjects of passive clauses.
288
For a more detailed description of the approach indicated here see VDE
(2004: xxiii–xxxiii) or Herbst and Schüller (2008). For an alternative account
of the description of different types of complement in English see Emons
(1974 and 1978). See in particular Emons (1978: 26–27) for the principle of
Kommutationsklassen: A man/He sees a child.
190 Sentences – models of grammar

With respect to the level of optionality, a distinction can be made be-

tween

obligatory valency slots, which have to be realized by a complement
whenever the valency carrier is used,

contextually optional valency slots, which can but need not be realized
by a complement as long as the referent of the corresponding participant
can be identified from the context,

optional valency slots, which need not be realized by a complement.289
In this terminology, both valency slots of a verb such as paint realized in
(11) can be classified as optional
(11)PH I probably painted more St Ives harbour window paintings in London
than in Cornwall.
(28a)PH Christmas Eve was painted at the invitation of the Arts Council for the
Festival of Britain Exhibition 60 Paintings for 51.
(28b)PH Abstract art never interested me. I have always painted realistically.
Valency slot II of watch, however, is contextually optional since sentences
such as
(40)VDE The men just watched and laughed.
(41)AE I’ve always wanted to know.
only occur if it is clear what is being watched or known. Finally, valency
slot II of put – realized by her soiled breakfast things is an example of an
obligatory valency slot:
(42)NW She puts her soiled breakfast things into the sink ...
It is important to realize, however, that the fact that an element is optional
from the point of view of valency does not mean that it could be deleted
from a sentence. The level of valency is only one factor with respect to the
optionality of an element in a sentence. It can be supplemented by the lev-

289
Allerton (1982: 68–69) makes a distinction between indefinite and contextual
deletion on which the distinction between optional and contextually optional
complements is based. For a discussion of very similar concepts in the Fra-
meNet context see Fillmore (2007: 144–151). Compare also Goldberg’s
(2006: 39) description of lexically profiled roles as those which are “obligato-
rily expressed, or, if unexpressed, must receive a definite interpretation” – as
argument roles.
 12 Valency theory and case grammar 191

els of structural and communicative necessity. Structural necessity refers

to the influence of sentence structure in this respect: thus imperative con-
structions do not have subjects, whereas declarative and interrogative con-
structions usually have subjects.290 Communicative necessity finally refers
to the fact that in a given situation any element of a sentence can be neces-
sary from a communicative point of view, as pointed out in 12.3.2.

12.3.6 Valency patterns

The complementation properties of lexical units can either be described in

terms of valency slots that are realized by complements expressing particu-
lar participant roles or in terms of patterns in which a valency carrier can
occur.291 This is the approach taken in the model of Pattern Grammar de-
veloped by Hunston and Francis (2000)292 or the Erlangen Valency Pat-
ternbank (www.patternbank.uni-erlangen.de), for example.
Within a valency-driven approach, it may make sense to distinguish be-
tween three types of patterning:

valency patterns as patterns of formal valency complements

participant patterns as patterns of the same participant roles

valency constructions as patterns of valency complements expressing
the same participant roles.
Thus, at a purely formal level, be and reach in (37a) can be seen as repre-
senting the same valency pattern: NP – VHC2 – NP.293
(37a)PW St Just is the last town in England before you reach Land’s End.

290
See Herbst and Schüller (2008: 148–156) for a description of different sen-
tence types such as the imperative-‘directive’-construction or the declarative-
‘statement’-construction. It must be recognized that there is a certain overlap
between the levels of structural necessity and valency because the ability of a
complement to occur as subject entails certain facts about its optionality.
291
For reasons to regard these two perspectives as complementary see Herbst
(2007). Compare also Herbst and Faulhaber (forthcoming). For the role of
patterns in German valency theory and a distinction between Satzmuster and
Satzbauplan see VALBU (2004: 46–47) and Engel (1977: 180–181).
292
For parallels between pattern grammar and construction grammar see Stubbs
(2009: 27).
293
VHC stands for verbal head complex, 2 for the valency of the verb (> 12.4).
192 Sentences – models of grammar

They represent different valency constructions, however, because the NPs

have different semantic roles – such as ‘PREDICATIVE’ (the last town in
England) and ‘ÆFFECTED’ (Land’s End). (32) and (43), on the other hand,
display the same cluster of participant roles (possibly to be described as
‘AGENT’ – ‘BENREC’ – ‘ÆFFECTED’) with different formal realizations:
(32)PW Mondrian showed us how completely possible it was to find a personal
equation.
(43)TI This is the first work by Barbara Hepworth to which she gave a title
referring directly to landscape.
Examples (11), (44) and (45) can be attributed to the same valency con-
struction in that they represent the same valency pattern NP – VHC2 – NP
and the same participant roles (‘AGENT’ – ‘EFFECTED’):294
(11)PH I probably painted more St Ives harbour window paintings in London
than in Cornwall.
(44)TS In the early 1950s, Nicholson produced a series of paintings which
show him at the height of his powers.
(45)TI Lanyon made numerous versions of this composition …

12.4 A valency based approach to English syntax

12.4.1 Combining aspects of clause structure and valency

Although valency properties of individual words, in particular verbs, have a

determining influence on the structure of a sentence, a number of other
factors will also have to be taken into account.
In what follows we will introduce an approach towards the analysis of
English sentences that combines the aspect of clause structure and valency.
This seems appropriate since
(a) valency properties of governing verbs have an influence on the structure
of the clause, in particular the predicate,

294
It would be interesting to find out to what extent these different types of pat-
terning are cognitively relevant. The concept of valency constructions bears
close resemblance to Goldberg’s (2006) argument structure constructions (>
13.3.2), cf. also Herbst (2010).
 12 Valency theory and case grammar 193

(b) properties of the clause (active, passive or declarative, interrogative,

imperative) have an influence on the form and degree of optionality of
complements.
The approach developed by Herbst and Schüller (2008) combines the ter-
minologies of traditional grammar and valency theory and introduces the
following terms for different kinds of unit:

subject complement unit (SCU), which consists of the subject of the
clause, which is realized by a complement of the governing verb

predicate complement unit (PCU), which is any complement of the
governing verb which occurs in the predicate, i.e. any complement other
than the subject

predicate head unit (PHU), which contains the governing verb of the
clause plus any other elements in the predicate (auxiliaries, modifiers)
that are not complements

adjunct unit (AU)

linking unit (LU).295
(1)AE By 1918 St Ives had a long tradition as an artists’ community.

By 1918 St Ives had a long tradition as an artists’

community
AU SCU PHU PCU

(30)AE For two years this gave him studio facilities and a small stipend.

For two years this gave him studio facilities and a

small stipend
AU SCU PHU PCU1 PCU2

295
An example of a linking unit would be and in … it was an October night and
a full moon was rising over GodrevyPH, which links two independent clauses
(> 11.1.2).
194 Sentences – models of grammar

12.4.2 A modified view of phrase structure

12.4.2.1 Head complexes

One problem of the traditional definition of phrases as given in CGEL

(1985: 2.26), for example, is that in the case of endocentric phrases (>
11.3.1) the head of a phrase is defined as that element of a phrase that can
stand for the phrase on its own. This criterion works in cases such as:
(46)PH Once there Heron had immediately started work on a series of small
vertical paintings that carried forward a programme of radically ab-
stract experimentation that had begun in earnest in late 1955 …
(46a) Once there Heron had immediately started work on a series of paintings
…
However, in many cases, a noun phrase cannot be replaced by the head in
this way: 296
(47)TS Barbara Hepworth’s situation was very different at this time.
(47a) *Situation was different at this time.
(48)PH For Heron it was a dream realized …
(48a) *For Heron it was dream realized …
One possible way of tackling this problem is to identify a nominal head-
complex within the noun phrase, which can consist of a number of so-
called pre-heads and a head: the right-most constituent of the head-complex
is the head. The head can be preceded by up to three pre-heads.

296
It is for this reason that some theories consider it more appropriate to speak of
determiner phrases (with determiners as heads) than of noun phrases. While
this works in the case of a dream in (48), an analysis in terms of determiner
phrases is problematic in cases where there is no determiner as with Heron or
paintings in (46). See Hudson (1984: 90–92) and Matthews (2007). Cf. also
Radford’s (1993) proposal of co-headedness.
 12 Valency theory and case grammar 195

pre-head(s) head
book
a book
his second book
all these books

In the terminology introduced by Herbst and Schüller (2008) (> 11.4.6),

determiners (the), determiner-pronouns (this) or genitives of pronouns
(their) or nouns (John’s) can function as pre-heads of noun phrases, and

determiner-pronouns (this), pronouns (they) or nouns (John, book) can
function as heads of noun phrases.
In very much the same way, the distinction of pre-head and head can be
used to describe the verbal head-complex:

pre-head(s) head
was going
went
has been laughing

In the verbal head-complex,

modal verbs and primary verbs can function as pre-heads or (if they are
the right-most constituent of the verbal head-complex) as heads,
whereas

lexical verbs always function as heads.

12.4.2.2 Noun phrases, adjective phrases and adverb phrases

A further difference in the account of phrase structure in a valency-based

model is that a distinction is made between the valency complements of the
head and modifiers, which are not dependent on the valency of the head.
The structure of the noun phrase can then be described as follows (pre-
heads and head forming the nominal head-complex):
196 Sentences – models of grammar

Noun phrase
pre- premodifier(s) head postmodifier(s) complement(s)
head(s)
they
all the boys
books on linguistics
these lovely paintings made in St Ives
no idea where to go

Adjective and adverb phrases only have single heads:

Adjective phrase
premodifier(s) head postmodifier(s) complement(s)
very clever
extremely good at mathematics
interesting enough

Adverb phrase
premodifier(s) head postmodifier(s) complement(s)
cleverly
rather stupidly
independently of the outcome
very well indeed

Apart from pre- and postmodifiers, there are also discontinuous modifiers
(Aarts and Aarts 1988), where the second constituent is dependent on the
first constituent:
(49)PH I thought I was making the most extreme paintings in the world.
 12 Valency theory and case grammar 197

12.4.2.3 Particle phrases

If one follows the word class distinctions suggested by Herbst and Schüller
(2008) and subsumes subordinating conjunctions, prepositions and some
adverbs under a single category of particles which has the property of
valency (> 11.4.7), then one can also identify a category particle phrase.
Particle phrases can be analysed as being headed because the minimum
realization of any particle phrase consists of the particle and its valency
complements.

Particle phrase
premodifier(s) head complement(s)
1 here
2 not long after Robyn’s arrival
on the table
3 whether she wanted to stay
to go
three years after they first met

In traditional accounts, these phrases would be classified as adverb phrases

(1), prepositional phrases (2) and subordinate clauses (3).

12.4.2.4 Clauses as verb phrases

Within such a valency-oriented model of phrase structure, it does not really

make sense to distinguish between clauses and verb phrases, which is a
point of view also taken by Fillmore (1988: 43), for example. CGEL’s
(1985: 2.27) verb phrases correspond to the verbal head-complex as de-
scribed here. The verbal head-complex can be seen as being central to the
clause in very much the same way as the nominal head-complex is central
to the noun phrase.
In the description of the clause, the verbal head-complex will be speci-
fied in two ways:

as active or passive

in terms of the number of complements dependent on its head in the
sentence.
198 Sentences – models of grammar

(50) She received [VHCact: 3] a letter from him.

(51) She was given [VHCpass: 2] a letter yesterday.

12.4.3 Description of units

In the model proposed by Herbst and Schüller (2008), the different types of
unit identified are further specified as follows:

predicate head units (PHUs) are described in terms of the type of ver-
bal head-complex by which they are realized

subject complement units (SCUs), predicate complement units
(PCUs) and adjunct units (AUs) are described in terms of their formal
realizations, i.e. in terms of phrases or clauses

predicate complement units realized by noun phrases are assigned a
semantic role (> 12.2.3)

linking units are specified in terms of word class only.
(52) She had given him a book yesterday
SCU: PHU: PCU1: PCU2: AU: AdvP
NP VHCact:3 NPBENREC NPÆFFECTED
It is obvious that a comprehensive description of the sentence should com-
prise an indication of the semantic roles for all predicate complement units
as well as for SCUs and AUs. For practical purposes, this can be restricted
to noun phrase PCUs, where roles serve to make clear the difference be-
tween sentences such as (52) and (53):
(53) She had called him a fool yesterday
SCU: PHU: PCU1: PCU2: AU: AdvP
NP VHCact:3 NPÆFFECTED NPPREDICATIVE

12.4.4 Example

This model can be used to analyse a sentence such as

(52)PH I shall never forget the immense sensation of space the first moment we
entered that room …: it was an October night and a full moon was ris-
ing over Godrevy.
 12 Valency theory and case grammar 199

in the following way:297

297
For an approach towards the analysis of English sentences in eight steps and
more examples see Herbst and Schüller (2008: 173–193).
13 Theories of grammar and language
acquisition

13.1 Chomsky’s approach

13.1.1 Basic assumptions

A radically different approach towards the description of syntactic struc-

tures is taken by Noam Chomsky, whose ideas can no doubt be said to have
dominated much of the linguistic and particularly syntactic research in the
second half of the twentieth century all over the world. It is probably fair to
say that for many researchers in the field his theories presented not only a
new and in many ways interesting way of looking at language, but the only
conceivable framework for linguistic analysis. From a slightly more cau-
tious perspective, this phenomenon is difficult to explain. The enormous
appeal of Chomsky’s theories may have a number of reasons, such as
− the high degree of formalization characteristic of his theories,
− the precise formulation of the aims of linguistic theory connected with a
clearly defined view of language,
− his interest in language universals, i.e. properties that are common to all
languages,
− certain psychological implications of his theories.
For instance, the crucial distinctions introduced by Chomsky between com-
petence and performance (> 3.1.2) and his own concentration on compe-
tence as the object of research have resulted in a view of language and sen-
tence analysis that to a very large extent abstracts from (and thus ignores)
sociolinguistic and situational factors. This becomes particularly apparent
in Chomsky’s (1965: 3) notion of the ideal speaker/hearer, which forms the
basis of his theory:
Linguistic theory is concerned primarily with an ideal speaker-listener, in a
completely homogeneous speech community, who knows its language per-
fectly and is unaffected by such grammatically irrelevant conditions as
memory limitations, distractions, shifts of attention and interest, and errors
 13 Theories of grammar and language acquisition 201

(random or characteristic) in applying his knowledge of the language in ac-

tual performance.
This abstraction from social factors of communication also shows in the
following definition of language provided by Chomsky in Syntactic Struc-
tures (1957: 13):
I will consider a language to be a set (finite or infinite) of sentences, each
finite in length and constructed out of a finite set of elements. All natural
languages in their spoken or written form are languages in this sense, since
each natural language has a finite number of phonemes ... and each sen-
tence is representable as a finite sequence of these phonemes ..., though
there are infinitely many sentences.
The task of linguistics then is to describe which combinations of elements
can be sentences of a language. This is strongly linked with the notion of
generative grammar, which Chomsky (1965: 8) describes as follows:298
… by a generative grammar I mean simply a system of rules that in some
explicit and well-defined way assigns structural descriptions to sentences.
Obviously, every speaker of a language has mastered and internalized a
generative grammar that expresses his knowledge of the language. This is
not to say that he is aware of the rules of the grammar or even that he can
become aware of them, or that his statements about his intuitive knowledge
of the language are necessarily accurate. Any interesting generative gram-
mar will be dealing, for the most part, with mental processes that are far be-
yond the level of actual or even potential consciousness.
Within such a generative grammar, it is not only the rules that operate in
any particular language that are of interest to the linguist but rules or prin-
ciples which apply to all languages and can thus be subsumed under the
heading of universal grammar. In Aspects of the Theory of Syntax, Chom-
sky (1965: 6) emphasizes the need for research into the nature of universal
grammar, which has characterized his approach ever since:

298
While the use of the phrase “mental processes” might be taken as an indica-
tion of certain psychological or cognitive aims of the theory, Chomsky (1965:
9) also says: “When we speak of a grammar as generating a sentence with a
certain structural description, we mean simply that the grammar assigns this
structural description to the sentence. When we say that a sentence has a cer-
tain derivation with respect to a particular generative grammar, we say noth-
ing about how the speaker or hearer might proceed, in some practical or effi-
cient way, to construct such a derivation.”
202 Sentences – models of grammar

Within traditional linguistic theory … it was clearly understood that one of

the qualities that all languages have in common is their “creative” aspect.
Thus an essential property of language is that it provides the means for ex-
pressing indefinitely many thoughts and for reacting appropriately in an in-
definite range of new situations (…). The grammar of a particular language,
then, is to be supplemented by a universal grammar that accommodates the
creative aspect of language use and expresses the deep-seated regularities
which, being universal, are omitted from the grammar itself.
Thus one could argue that what made Chomsky’s approach towards the
study of language so attractive to many scholars was that he emphasized
different perspectives in the study of language from, say, the ones ad-
dressed by structuralism. The model of generative grammar as promoted by
Chomsky has given rise to an enormous number of different theories based
to a greater or lesser degree on the ideas outlined by Chomsky. Chomsky’s
own version of generative grammar has also gone through a number of
rather radical modifications, represented by publications such as Syntactic
Structures (1957), Aspects of the Theory of Syntax (1965), Lectures on
Government and Binding (1981) and The Minimalist Program (1995).

13.1.2 Transformations – deep structures and surface structures

One of the key notions of the early versions of Chomsky’s generative

grammar is that of transformation. It can be described as an operation be-
tween two levels of syntactic structure identified – deep structure and sur-
face structure. The distinction between two such levels of structure can be
justified through examples of syntactic ambiguity in sentences such as299
(1)QE Flying planes can be dangerous.
It is argued that the surface structure of (1) represents two deep structures
representing the different readings of these sentences as
(1a) The flying of planes can be dangerous.
(1b) Planes that fly can be dangerous.

299
For a discussion of constructional homonymy see Matthews (1993:149–153)
and Palmer (1971: 141–150).
 13 Theories of grammar and language acquisition 203

On the other hand, there are cases which can be explained as representing
different surface structures such as actives and passives that are derived
from the same deep structure.300
(2a)QE John plays golf.
(2b)QE Golf is played by John.
The deep structure as envisaged by Chomsky in Aspects can be imagined as
an unambiguous conceptual syntactic structure, in which the elements are
ordered according to certain principles. The deep structure as such is based
on the constituency or phrase structure analysis frequently employed in
American structuralism (> 12.1.1). In early versions of the model such as
Syntactic Structures, such deep structures are then subject to a process of
transformations, which maps them into surface structures, i.e. grammatical
sequences of the language.301 The passive transformation can be repre-
sented as follows: 302
NP1 + Aux + V + NP2 → NP2 + Aux + be + en + V + by NP1
which then leads to the corresponding surface structure. In Aspects of the
Theory of Syntax and related models, particular transformations required
for the generation of sentences in a language – such as the passive trans-
formation, deletion, substitution, relative transformation – were specified.
In later versions of the theory, this is no longer the case. In the 1981 Gov-

300
Cf. Chomsky (1965: 21). For the passive transformation see Chomsky (1957:
77–81). Similarly, declarative and different types of interrogative sentences
(John ate an apple – Did John eat an apple) are seen as being derived by dif-
ferent transformations from one underlying string “John – C – eat + an +
apple”; see also Chomsky (1957: 91). See also Palmer (1971: 135–141).
301
Chomsky (1965: 141) describes the Aspects model as follows: “The syntactic
component consists of a base and a transformational component. The base, in
turn, consists of a categorical subcomponent and a lexicon. The base gener-
ates deep structures. A deep structure enters the semantic component and re-
ceives a semantic interpretation; it is mapped by the transformational rules
into a surface structure, which is then given a phonetic interpretation by the
rules of the phonological component. … The categorical subcomponent of
the base consists of context-free rewriting rules.”
302
Compare Chomsky (1957: 43) and Chomsky (1965: esp. 131–132). For a
description of active and passive in terms of transformations see Harris
(1957/1981: 194–197). For the contribution of Harris see Matthews (1993:
160–162). For a discussion of the inadequacy of phrase structure grammars
see Chomsky (1957: 34–48).
204 Sentences – models of grammar

ernment and Binding model303 a general operation called move α is intro-

duced instead, which relates the levels of s- and d-structure (as they are
called in this model), and restrictions on move α are defined.304 Apart from
the levels of s-structure and d-structure, Chomsky identifies the levels of
phonetic form and logical form referring to output in terms of sound and
meaning. In the Minimalist Program, Chomsky (1995: 219) abolishes the
levels of d-structure and s-structure.
Looking at the overall development of generative grammar one aspect to
be noted is the increasing specification of syntactically relevant lexical
properties in determining sentence structure, which results in a closer simi-
larity to valency and dependency approaches. Thus, integrating notions of
case grammar (> 12.2), the lexical entry for a verb such as defeat is speci-
fied in the following form:
defeat [__NP] <Agent, Patient>
This description contains a formal statement that the verb defeat needs to
be followed by a noun phrase (as its object, in this case) and that subject
and object represent the semantic roles (called theta-roles in this model) of
AGENT and PATIENT.305
Despite considerable changes in the formalizations proposed, two basic
convictions can be identified as remaining relatively constant in Chomsky’s
approach. First of all, there is a strong focus on general principles, which is
also reflected in his view of the lexicon given in the Minimalist Program
(1995: 235):306

303
For introductory accounts of the model see Cook (1988), Haegeman (1991)
and Sells (1985). See also Matthews (1990), Lyons (1991) or Maclay (1971).
304
Cf. Chomsky (1995: 20): “We assume that the language (the generative pro-
cedure, the I-language) has two components: a computational system and a
lexicon. The first generates the form of SDs; the second characterizes the
lexical items that appear in them. ... We will assume that one aspect of an SD
is a system of representation, called D-Structure, at which lexical items are
inserted. D-Structure expresses lexical properties in a form accessible to the
computational system.” (SD = structural description) In addition to d-
structure and s-structure, Government-Binding Theory introduces phonetic
form and logical form (referring to meaning), cf. Sells (1985: 19–20).
305
See Cook (1988: 122) for this and the notions of s-selection, c-selection and
the Projection Principle. For the passive compare also Sells (1985: esp. 43).
For a discussion of s-selection and c-selection cf. Chomsky (1986: 86–90).
306
See Chomsky (1965: 164–170) for an earlier description of the lexicon in-
volving strict subcategorization features and selectional restrictions.
 13 Theories of grammar and language acquisition 205

I understand the lexicon in a rather traditional sense: as a list of “excep-

tions”, whatever does not follow from general principles. These principles
fall into two categories: those of UG, and those of a specific language. The
latter cover aspects of phonology and morphology, choice of parametric op-
tions, and whatever else may enter into language variation. Assume further
that the lexicon provides an “optimal coding” of such idiosyncracies.
A second principle underlying the various versions of the theory is the idea
of deriving syntactic structures, as is obvious from Chomsky’s (1995: 219)
description of language: “The language L determines a set of derivations”.
Both of these ideas have been challenged by construction grammar (>
13.2).

13.1.3 Claims and evidence

It was said in 13.1.1 that generative transformational grammar constitutes

one of the most important research paradigms in the history of linguistics.
Indeed it would be difficult to deny the great intellectual appeal of the
model and the value of an enormous number of contributions to descriptive
linguistics carried out within the Chomskyan framework. Furthermore, it is
an indication of the outstanding importance of Chomsky’s work that it is
not only referred to by people working within the framework or frame-
works outlined by him, but that many who work within alternative frame-
works also refer to Chomsky’s work in order to demonstrate in what way
their approach differs from his.
One of the main points of criticism raised against Chomsky’s approach
concerns the question of evidence. Thus Sampson (2001: 141) speaks of
“Noam Chomsky’s fairly unempirical theory of ‘Transformational Gram-
mar’” and criticizes that307
… the existence of linguistic universals is, for Chomsky and his followers,
not so much a finding that has emerged from their research despite their ex-
pectations, but rather a guiding assumption which determines the nature of
the hypotheses they propose in order to account for data. (Sampson 1980:
148)
Similarly, as far as data are concerned, the fact that judgments about the
grammaticality or ambiguity of sentences were largely based on the intui-
tion of the linguist, is highly problematic, as Matthews (1990: 124) points

307
For a similar view see Matthews (1990: 16).
206 Sentences – models of grammar

out.308 The focus on the rather abstract notion of an ideal speaker-hearer

and the question of what is possible in a language provides good reasons
for considering introspection as superior to the analysis of corpora: firstly,
corpora represent instances of performance, and, secondly, the fact that a
particular form does not occur in a corpus does not mean that it does not
exist in the language. However, the total rejection of the use of corpora and
the methodology of corpus linguistics by Chomsky – culminating in the
statement that corpus linguistics “doesn’t exist” (Aarts 2000: 5) – means a
neglect of certain aspects of language use for which (given the size of to-
day’s corpora) there is less justification today than there was in the 1950s
and 1960s.309 A similar type of limitation can be seen in the focus of atten-
tion to what Chomsky (1995: 19–20) calls core language in the exploration
of universal grammar (UG):310
For working purposes (and nothing more than that), we may make a rough
and tentative distinction between the core of a language and its periphery,
where the core consists of what we tentatively assume to be pure instantia-
tions of UG and the periphery consists of marked exceptions (irregular
verbs, etc.). Note that the periphery will also exhibit properties of UG (e.g.,
ablaut phenomena), though less transparently. A reasonable approach
would be to focus attention on the core system, putting aside phenomena
that result from historical accident, dialect mixture, personal idiosyncracies,
and the like.
Such an approach entails two dangers: from a methodological point of view
one could argue that it is relatively easy to observe regularity in core
grammar when all elements of irregularity can be relegated to the periph-
ery. Furthermore, one might ask whether it is not an essential part of lan-
guage, and thus of any comprehensive description of language, that it is the
result of a historical process and whether the amount of irregularity to be
observed in language is something that can be relegated to the periphery so
easily. In fact, study of such phenomena as collocation and valency might

308
For a more conciliatory view compare Matthews (1993: 33).
309
See 3.2.3. Of course, the (by today’s standards) small size of corpora avail-
able during the early stages of transformational grammar meant considerable
limitations. Cf. Leech (1991: 13) for this: “... Chomsky, in his turn, could not
have conceived, in the 1950s, of a corpus of 500 million words capable of be-
ing searched in a matter of minutes or hours. … it is unlikely that foreknowl-
edge of such a phenomenon would have changed Chomsky’s view of corpora
at that time.”
310
Compare also Matthews (1990: 136).
 13 Theories of grammar and language acquisition 207

lead one to assume that the opposite is the case and that a considerable
amount of item-specific information will have to be accounted for in any
theory of language, as has been done by the formulation of the idiom prin-
ciple by Sinclair (> 10.5) and in particular in construction grammar (>
13.2).
Critics might also object to the general tenor of some of the claims made
within the theory.311 Thus, there is a certain discrepancy between Chom-
sky’s (1995: 7) describing the principles and parameters model outlined in
the Minimalist Program as “in part a bold speculation rather than a specific
hypothesis” and at the same time claiming that it “constituted a radical
break from the rich tradition of thousands of years of linguistic enquiry, far
more so than early generative grammar, which could be seen as a revival of
traditional concerns and approaches to them” (1995: 5). Similarly, a state-
ment of the type
Generative grammar can be regarded as a kind of confluence of long-
forgotten concerns of the study of language and mind, and new understand-
ing provided by the formal sciences.
The first efforts to approach these problems quickly revealed that tradi-
tional grammatical and lexical studies do not begin to describe, let alone
explain, the most elementary facts about even the best-studied of languages.
(1995: 4)
is hardly reconcilable in tone with saying
The end result is a picture of language that differs considerably from even
its immediate precursors. Whether these steps are on the right track or not,
of course, only time will tell. (1995: 10)
However, in this respect Chomsky seems to agree with one of his critics,
Peter Matthews (1993: 252), who says:

311
Matthews (1982: 22–23) deplores “never-ending reformulation” and speaks
of the challenge of “demystification”. Compare Chomsky’s introduction to
the Minimalist Program (1995: 10): “The field is changing rapidly under the
impact of new empirical materials and theoretical ideas. What looks reason-
able today is likely to take a different form tomorrow.” Compare also Chom-
sky’s (2004: 152) statement in an interview given in 2002: “… I think the
minimalist critique of the past ten years has given substantial reasons to sup-
pose that none of these things exist: d-structure, s-structure, and LF just don’t
exist. There is no X-bar Theory forming d-structure.”
208 Sentences – models of grammar

Faith is a beautiful thing, which a non-believer can only regard with awe. ...
There are many who share Chomsky’s faith, and for them ‘The research
program of modern linguistics’ (Chomsky 1988: Ch. 2) must without arro-
gance be his programme. Others cannot bring themselves to accept that it
has empirical content. ‘C’est magnifique, mais peut être n’est pas la linguis-
tique’. But all they can do is watch and wait.

13.1.4 Language acquisition

13.1.4.1 The language acquisition device

One of the main attractions of Chomsky’s approach is perhaps to be seen in

the fact that right from the beginning it had been linked with questions of
language acquisition. In his famous review of Skinner’s book Verbal Be-
havior, Chomsky (1959: 57) rejects the behaviourist approach to language
acquisition and argues for a mentalistic approach:
The fact that all normal children acquire essentially comparable grammars
of great complexity with remarkable rapidity suggests that human beings
are somehow specially designed to do this, with data-handling or ‘hypothe-
sis-formulating’ ability of unknown character and complexity.
The key points in this statement are “great complexity” and “remarkable
rapidity”. The idea is that since language acquisition is such a highly com-
plex task and takes place relatively quickly, humans must be equipped with
special genetically determined properties which enable them to acquire
language.312 In the 1965 Aspects model, this takes the form of an innate
language acquisition device (LAD), which means that “the child has an
innate theory of potential structural descriptions” (Chomsky 1965: 32). The
job of the LAD is to select the appropriate type of grammar on the basis of
the data the child is exposed to during the language acquisition process:
This device must search through the set of possible hypotheses G1, G2 ...,
which are available to it ... and must select grammars that are compatible

312
Note, however, that Steinberg (1982: 100) points out that one “basic underly-
ing premise on which this argument rests is that four or five years (or what-
ever length of time it takes the child to acquire a grammar) is not a long
time.” Cf. Steinberg (1982: 96–100). See also Klann-Delius (1999: 51–52),
who criticizes the concept of the LAD as remaining rather vague.
 13 Theories of grammar and language acquisition 209

with the primary linguistic data. ... The theory that the device has now se-
lected and internally represented specifies its [the child’s] tacit competence,
its knowledge of the language. The child who acquires a language in this
way, of course, knows a great deal more than he has ‘learned’. His know-
ledge of the language ... goes far beyond the presented primary linguistic
data and is in no sense an ‘inductive generalization’ from these data.
(Chomsky 1965: 32–33)

13.1.4.2 Universal grammar

In later models, the idea of a language acquisition device is given up in

favour of the principles and parameters approach. According to this model,
a child is genetically equipped with universal grammar, which Chomsky
(1986: 24) describes as “a characterization of these innate, biologically
determined principles, which constitute one component of the human mind
– the language faculty”. The UG contains principles and parameters, which
Chomsky (1986: 146) describes as follows:
... UG consists of various subsystems of principles; it has the modular struc-
ture that we regularly discover in investigation of cognitive systems. Many
of these principles are associated with parameters that must be fixed by ex-
perience. The parameters must have the property that can be fixed by quite
simple evidence, because this is what is available to the child. ... Once the
values of the parameters are set, the whole system is operative. ... The sys-
tem is associated with a finite set of switches, each of which has a finite
number of positions (perhaps two). Experience is required to set the
switches. When they are set, the system functions.
The transition from the initial state S0 to the steady state Ss is a matter of
setting the switches.
Language acquisition is then a matter of parameter setting. This setting of
parameters takes place on the basis of the input data and eventually leads to
the acquisition of the core grammar of the language.313
In a way, the same types of objections that applied to the earlier versions
of the grammar can be raised against the principles and parameters ap-
proach. Firstly, as was already pointed out above, the concentration on core
grammar, which is an essential element of the theory, may well mean that
important aspects of language are totally ignored. Secondly, as Klann-
Delius (1999: 53) points out, the model still does not offer an outline of the
313
Cf. Klann-Delius (1999: 52–53).
210 Sentences – models of grammar

process of language acquisition as such. Factors such as the interaction of

children with their parents and other people or the special kind of language
parents use when talking to very young children – known as child-directed
speech (> 13.2.3) – are totally ignored. Thirdly, this model lacks empirical
evidence. With the exception of cases such as clitic climbing in Italian,
discussed by Fanselow and Felix (1987: 15–20), there does not seem to be
overwhelming evidence to show that the setting of one parameter affects a
significantly large number of grammatical features of a language – which,
however, would be essential for the model to be attractive. In other words,
there may be good reasons to suspect, as Peter Matthews (1990: 137) does,
that “yet again, we posit that something is innate not because we have
found it to be universal, but because we see no way by which it can be
learned”.314

13.2 Usage-based approaches

13.2.1 Construction grammar

Of the many models of grammar that exist alongside or as modifications of

or alternatives to the Chomskyan version of generative grammar, the ap-
proach of construction grammar is perhaps particularly interesting. On the
one hand, it offers an opportunity of accommodating the findings of corpus
linguists and foreign language linguistics with respect to the item-specific
nature of some linguistic phenomena. On the other hand, construction
grammar and related approaches have had considerable impact on models
of first language acquisition (> 13.2.3).315
Construction grammar must not be perceived as one unified theory. In
fact, Charles Fillmore (1988: 35) speaks of “a set of moving targets with
the same name.” Fischer and Stefanowitsch (2006: 3–4) distinguish three
main approaches – that of the Berkeley constructionists such as Fillmore,
that developed in a cognitive linguistics framework by Lakoff (1987) and
Goldberg (1995, 2006), and that of Radical Construction Grammar, which,
however, share a number of important assumptions:316

314
Compare also Goldberg (2006: 72).
315
For a slightly more sceptical view see Stubbs (2009: 27), Sinclair and Maura-
nen (2006: 31). Cf. also Herbst (2007: 49–53).
316
For an account of these different models see Fischer and Stefanowitsch
(2006) and Croft and Cruse (2004: 257–290).
 13 Theories of grammar and language acquisition 211

The key notion of the model is the term construction, which is defined
as a “form-meaning pair” (Lakoff 1987: 467).317 At first sight, the range of
elements subsumed under the term construction may seem rather wide.
Goldberg (2006: 5), for example, lists elements such as the following as
constructions:318
− morphemes this, paint, pre-, -ing, -s
− idioms (filled) give the Devil his due
− idioms (partially filled) jog <someone’s> memory
− covariational conditional the Xer the Yer
− ditransitive Subj V Obj1 Obj2
− passive SUBj aux VPpp (PPby)
This means that a sentence such as
(3)WG Justifiably, Heron had always been irritated by those literal-minded
critics who insisted in seeking references to the Cornish landscape (…)
in the shapes and contours of his paintings.
is analysed as a combination of different constructions such as the passive
construction, the past perfect construction etc.; furthermore words such as
Heron, always or –ly would be identified as constructions occurring in this
sentence.
The advantage of subsuming all these under the term construction – de-
spite the different degrees of concreteness or abstraction of the whole or
parts of the construction – concerns the fundamental hypothesis of con-
struction grammar, namely the idea of “a uniform representation of all
grammatical knowledge in the speaker’s mind” (Croft and Cruse 2004:
255). In other words, the basic idea of the theory is that language can be
described as a structured inventory of constructions, which can be com-
bined to form utterances, as formulated by Fillmore (1988: 37):

317
Whereas Goldberg (1995: 4) argues in favour of a non-compositional view, a
later formulation by Goldberg (2006: 5) is weaker in this respect: “In addi-
tion, patterns are stored as constructions even if they are fully predictable as
long as they occur with sufficient frequency …”. For the discussion whether
constructions can be compositional or not see Fischer and Stefanowitsch
(2006: 5–6). Compare also Croft and Cruse (2004: 247).
318
Examples taken from Goldberg (2006: 5). Compare Croft and Cruse (2004:
255–256) and Fischer and Stefanowitsch (2006: 6).
212 Sentences – models of grammar

The grammar of a language can be seen as a repertory of constructions, plus

a set of principles which govern the nesting and superimposition of con-
structions into or upon one another.
In this respect, construction grammar differs fundamentally from the gen-
erative approach outlined in 13.1 since it does not posit “underlying syntac-
tic or semantic forms” or transformations (Goldberg 1995: 7).319 Also, of
course, it is interesting to see that construction grammars do not make a
distinction between core and periphery, but rather regard integrating the
idiomatic component of language as central to the model. There is a strik-
ing parallel between Sinclair’s (1991) idiom principle (> 10.5) and the in-
sight formulated by Fillmore, Kay, and O’Connor (1988: 534) that “in the
construction of a grammar, more is needed than a system of general gram-
matical rules and a lexicon of fixed words and phrases”. They demonstrate
the appropriateness of the concept of constructions by providing evidence
for the idiosyncratic and productive properties of constructions such as the
the X-er the Y-er construction
(4a)QE The more carefully you do your work, the easier it will get.
(4b)QE The bigger they come, the harder they fall.
or the let alone construction
(5a)QE Max won’t eat SHRIMP, let alone SQUID.
(5b)QE John hardly speaks RUSSIAN let alone BULGARIAN.
The fact that in this framework of thinking no sharp distinction between
syntax and lexis is made is a further parallel to Sinclair (2004: 164).320

13.2.2 Argument structure constructions

The notion of argument structure constructions as developed by Goldberg

(1995, 2006) provides an interesting framework for discussing certain as-
pects of verb valency and clause structure. Goldberg (1995: 43) makes a
distinction between participant roles – which she describes as frame-
specific roles – and more general argument roles such as agent or patient (>

319
Note that Goldberg (1995: 7) nevertheless sees construction grammar as a
form of generative grammar.
320
See also Bybee (2007: 284–292).
 13 Theories of grammar and language acquisition 213

12.2–3).321 Participant roles and argument roles can be fused under certain
conditions. One of these is the Semantic Coherence Principle according to
which the participant roles of the verb and the argument roles of the con-
struction must be compatible.322
Goldberg (2006: 41) identifies the participant roles loader, loaded-theme
and container for the verb load, which can then be related to the Caused
motion construction and a combination of the causative construction and a
with-construction in the following way:

Caused motion
(6a)QE Pat loaded the hay onto the truck
CAUSE-MOVE cause theme path/location
  
Load loader loaded-theme container

Causative + with constructions

(6b)QE Pat loaded the truck with hay
CAUSE cause patient + INTERMEDIARY instrument
  
Load loader container loaded-theme

Assigning (6a) and (6b) to two different constructional analyses is one way
of doing justice to the subtle differences in meaning between the two sen-

321
Goldberg (2006: 20) makes no claims as to the universal nature of the labels
for argument roles.
322
For the Correspondence Principle see Goldberg (2006: 40): “… profiled
participant roles of the verb must be encoded by profiled argument roles of
the construction, with the exception that if a verb has three profiled roles, one
can be represented by an unprofiled argument role (and realized as an oblique
argument)”. See Herbst (2010) for a discussion from the point of view of
valency theory.
214 Sentences – models of grammar

tences – such as the element in direct object position being presented as

being affected in some special way.323
Furthermore, positing argument structure constructions of this sort may
help to explain aspects of linguistic creativity concerning sentences of the
type
(7)QE She sneezed the foam off the cappuccino.
or, to use a German example,
(8)AM Den letzten habe ich dann wirklich noch daneben zittert.324
The explanation offered by Goldberg (2006: 6 and 94–98) is that the
‘caused-motion’-interpretation of such sentences must be due to the con-
struction since one would not want to argue that the locative complement
off the cappuccino or daneben really represent a valency slot or a separate
meaning of the corresponding verb.325 From that point of view the concept
of argument structure constructions seems quite attractive.
Argument structure constructions are defined in terms of form and
meaning, as illustrated by the following examples (Goldberg 2006: 73):
Form/Example Meaning Construction Label
1. Subj V Oblpath/loc X moves Ypath/loc Intransitive Motion
e.g. The fly buzzed into the room.
2. Subj V Obj Obl- X causes Y to move Zpath/loc Caused Motion
path/loc

e.g. Pat sneezed the foam off the cappuccino.

3. Subj V Obj Obj2 X causes Y to receive Z Ditransitive
e.g. She faxed him a letter.
4. Subj V Obj RP X causes Y to become Zstate Resultative
e.g. She kissed him unconscious.

323
For an account of this in terms of clausal roles see Herbst and Schüller (2008)
and Herbst (2010). See Goldberg (2006: 33–38) for arguments for not treat-
ing such alternations in terms of derivation.
324
Said by Magdalena Neuner on the occasion of winning the gold medal in the
Olympic biathlon race describing her last shot: I really trembled my last shot
wide of the mark.
325
For an account in terms of “nonce applications of the pattern” see Kay
(2005).
 13 Theories of grammar and language acquisition 215

The “formal” categories used such as Obj (object) and RP (resultative

phrase) are determined semantically to a certain degree and thus different
from the complement classes identified in valency theory (> 12.3).326 Se-
mantically, some of these argument structure constructions show a consid-
erable amount of polysemy. Thus for the ditransitive construction, Gold-
berg (1995: 38) identifies a central sense “Agent successfully causes recipi-
ent to receive patient”, which applies to verbs such as give, pass, throw,
bring, take and five further senses such as “Agent enables recipient to re-
ceive patient” (permit, allow) etc.327
Since so far the discussion of argument structure constructions has fo-
cused on a relatively small number of constructions such as the ones listed
above, it is difficult to say whether all verb uses (or all valency patterns)
could be related to argument structure constructions or whether these
should be imagined as prototypical centres in a continuum.

13.2.3 The usage-based view of language acquisition

Construction grammar and other usage-based approaches are closely related

to questions of the representation of language in the mind and offer very
promising alternatives to the Chomskyan approach to language acquisition.
In particular, followers of these theories argue that language can be learnt
and that there is no need to make the special kinds of assumptions about
innate properties of the mind that Chomsky made. Ellis (2003: 64–65)
characterizes the usage-based perspective on language as follows:328
They hold that structural regularities of language emerge from learners’
lifetime analysis of the distributional characteristics of language input and,
thus, that the knowledge of a speaker/hearer cannot be understood as an in-
nate grammar, but rather as a statistical ensemble of language experiences
that changes slightly every time a new utterance is processed.
Thus the idea of storage plays an important role in such accounts, which
also has important consequences for the role of memory. Bybee (2007:

326
For the question of to what extent correlations between verb meaning and use
in particular argument structure constructions are concerned see Faulhaber
(forthcoming). Compare Herbst (2009ab) and Herbst and Faulhaber (2010).
327
Cf. Gries and Stefanowitsch (forthc.) for polysemy of constructions and col-
lexeme classes.
328
Compare Hopper’s (1987) notion of Emergent Grammar.
216 Sentences – models of grammar

291) describes the brain as “a powerful categorization device for the effi-
cient sorting and storing of the pieces of our experience, including the units
of language use”. On the one hand, this accounts for item-specific know-
ledge relating to the phraseological aspects of language, where Bybee
stresses that it is “easier to access, produce, and comprehend a precompiled
chunk”. On the other hand, it is assumed that shared features can be discov-
ered and more abstract schemata will be developed on the basis of stored
examples of a similar nature.329 In Constructing a Language, Tomasello
(2003: 140 and 142) describes this as follows:
In the initial stages ... children’s linguistic competence is most accurately
characterized not as a “grammar”, but rather as an inventory of relatively
isolated, item-based constructional islands. Development after these initial
stages, typically at 2–3 years of age, then proceeds gradually and in piece-
meal fashion, with some constructions becoming abstract more rapidly than
others – mainly depending on the type and the token frequency with which
children hear particular constructions, since this is what provides the raw
material for the schematization process. ... Regardless of details, under no
circumstances does this development look like an instantaneous setting of
parameters in which all verbs and other lexical items immediately partici-
pate in a totally abstract construction.
Abstraction, identification of constituents and formation of schemata are
recognized as important parts of the language acquisition process. At the
same time it is important to note that even in adult language “there is solid
evidence that both item-specific knowledge and generalizations coexist”
(Goldberg 2006: 63). This means that all types of constructions (> 13.2.1)
can be acquired through the same processes of acquisition, which
Tomasello (2003: 6) describes as intention-reading and pattern-finding.
There seems to be a considerable amount of evidence for Tomasello’s
(2003: 6) claim that “children’s early language is largely item-based”.330
For instance, in a study by Lieven, Behrens, Speares and Tomasello (2003)

329
For the nature of storage compare cf. Behrens (2007): “It is as of yet un-
known whether we simply store more and more tokens upon repeated usage,
or whether we store more repeated information on a more general and ab-
stract level when available, or whether we do both.” See also Bybee (2005: 7)
and Goldberg (2006: 62). For the role of the memory and access to memory
see Bybee (2007: 291).
330
Behrens (2009: 394) considers it an “as of yet open issue … where item-
based formulas are the starting point for each aspect of language develop-
ment”.
 13 Theories of grammar and language acquisition 217

the linguistic development of a 2-year old child learning English was moni-
tored relatively closely over a period of 6 weeks and the utterances pro-
duced on the last day were compared with the recordings of the period be-
fore:331
− 79 percent of the things the child said on the target day (and 63 percent
of the multi-word utterances), she had said in exactly the same form be-
fore
− 74 percent of the novel multi-word utterances modified a previously
made utterance only in one place (Where is the butter? on the basis of
Where is the __?)
This is hardly surprising when one considers that the input children get also
seems to be highly repetitive and item-based. A very important factor in
this context is that child-directed speech, sometimes referred to as
motherese, i.e. the language that adults use when talking to very young
children shows certain characteristics which may facilitate the language
acquisition process. Tomasello (2003: 111–113) reports on a study of 12
English-speaking mothers interacting with 2-year old children, which
showed that the majority “of the utterances children hear are grounded in
highly repetitive item-based frames that they experience dozens, in some
cases hundreds, of times every day”.332
What these examples show is that child language can also be imagined
as an inventory of constructions of the same type as construction grammar
posits for adult language.333 Tomasello (2003: 99) points out that children
under the age of 2 have inventories consisting of words (such as bird),
some bound morphemes (plural –s), “frozen phrases” (I-wanna-do-it) and a
variety of item-based mixed constructions (Where’s-the X? I-wanna X) and
comes to the conclusion that “children do not first learn words and then
combine them into sentences via contentless syntactic ‘rules’”.

331
For a more detailed discussion of these results see Tomasello (2003: 307–
308) and Lieven, Behrens, Speares and Tomasello (2003). See also Ellis
(2003: 70), who talks about “a developing set of slot-and-frame patterns” in
this respect.
332
See Cameron-Faulkner, Lieven and Tomasello (2003) for details. For the role
of motherese see also Klann-Delius (1999: 142–146). For the verb-island hy-
pothesis see Tomasello (2003: 117–118).
333
See Tomasello (2003: 105–108) for chunks and analysability in adult lan-
guage.
218 Sentences – models of grammar

The idea of usage-based approaches to language acquisition is that lin-

guistic structure can be accounted for in terms of a few basic psychological
processes such as entrenchment and categorization, as pointed out by
Behrens (2009: 386):334
A central cognitive phenomenon is entrenchment, the fact that repeated en-
counter of a unit leaves memory traces that stabilize the more often this unit
recurs. … However, repetition alone does not lead to the abstraction of
more general information. In order to generalize and form categories, the
mind must recognize similarities as well as dissimilarities. It filters out as-
pects that do not recur, and registers communalities by comparing stored
with new units. New units are categorized along those dimensions where
similarities with stored units are detected. Through abstraction and gener-
alization, schemas are formed.
With respect to generalizations about argument structure constructions or
valency properties of verbs, entrenchment seems to play an important role,
but so does the factor of pre-emption. While it is relatively obvious that if
children hear a verb frequently used in a particular construction this use
will become entrenched, it is not quite so clear how children find out that
sentences such as
(9) *He explained them the situation.335
(10)QE *He disappeared the rabbit.
(11)QE *She giggled me.
(12)QE *Dad said Sue something nice.
do not seem to occur. Tomasello (2003: 178) argues that if one pattern is
established for a verb, children will not use it in another construction in
which they have not experienced it. Furthermore, hearing He made the
rabbit disappear or She made you giggle may pre-empt sentences such as
(10) or (11).336 With respect to example (12), Stefanowitsch (2008: 527)
334
For the concept of entrenchment see Langacker (1987: 59), Ungerer and
Schmid (2006) or Stefanowitsch (2008).
335
See Stefanowitsch (2007: 67–68) for an analysis of examples of explain me
this and a discussion of the notion of grammaticality.
336
For a discussion of pre-emption compare also Goldberg (2006: 96–98), who
points out that She sneezed the foam off the cappuccino would not be pre-
empted by the more frequent intransitive use of sneeze for semantic reasons.
Tomasello (2003: 180) also draws attention to the role of child-parent-
interaction in this context.
 13 Theories of grammar and language acquisition 219

introduces the concept of negative entrenchment, which is based on hy-

potheses about the expected frequency of co-occurrence of linguistic fea-
tures.337 Constraints such as the ones illustrated by (9) – (12) can thus be
explained in a way which is compatible with a view of language acquisition
in terms of learning.
There thus seems to be considerable empirical evidence for a usage-
based view of language acquisition in which the grammar of a language
appears as an emergent phenomenon, although of course it would be wrong
to assume that all scholars would necessarily interpret all the data in the
same way or that all data would point in the same direction.338 Neverthe-
less, there is a large body of evidence to suggest that linguistic competence
can be explained entirely as the result of a learning process and that no
assumptions about a special genetic predisposition of human beings that
would exclusively serve the purpose of language acquisition need to be
made.

Recommended further reading:

− CGEL (1985), CamG (2002)

− Sampson (1980), Sells (1985)
− Herbst and Schüller (2008)
− Palmer (1971)
− Tomasello (2003), Fischer and Stefanowitsch (2006), Behrens (2009)

337
For the hypothesis that “degrees of significant absence correlate with degrees
of unacceptability” see Stefanowitsch (2008: 527).
338
Compare for instance Tomasello’s (2003: 121–122) and Goldberg’s (2006:
77–79) discussions of the role of high-frequency verbs in the acquisition of
argument structure constructions. See also the connectionist model suggested
by MacWhinney (1998). For an account of different theories of language ac-
quisition within the usage-based approach see Ellis (2003) and Behrens
(2009).
 Meaning

14 Semantics: meaning, reference and

denotation

14.1 Meaning

One of the basic ideas about language which is shared by practically all
schools of modern linguistics is de Saussure’s characterization of the lin-
guistic sign as consisting of two components – signifiant (form) and signi-
fié (meaning). Although many of the aspects of language discussed in chap-
ters 4–16 focus on the form-component of the sign (at least in taking form
as the starting point of analysis), reference to meaning was repeatedly
made. For instance, the phoneme was defined as a linguistic unit that has
the function to “distinguish between forms with different meanings” (5.1),
and the morpheme was described as a “unit that serves to carry meaning”
(8.1). It is remarkable – although hopefully it escaped the notice of most
readers – that no precise definition or characterization of the term meaning
was given in these contexts. Similarly, there is nothing unusual about
statements of the sort that talk has a different meaning in English and in
German or that the meaning of the word nice has changed since Jane Aus-
ten’s time. In the same way, speakers of a language (who have had no train-
ing in linguistics) are perfectly able to make statements about meanings of
words, for instance, to the effect that good means the opposite of bad.
Speakers, in very many situations, are also able to describe the meanings of
words to children or even argue about the meaning of a word (in the sense
that there can be disagreement what words such as democracy or freedom
really mean or whether there can be such a thing as a third alternative or
whether the meaning of alternative is ‘one of two possibilities’).339 It thus
seems that, contrary to the word word (> 9.1.1), the way the word meaning

339
Compare the definitions of alternative in NSOED, LDOCE5 and OALD7.
 14 Semantics: meaning, reference and denotation 221

is used in everyday language does not differ greatly from its use in linguis-
tics.
The fact that most linguists would probably not object to any of the uses
of meaning above does not necessarily mean that all linguists would have
the same ideas about what meaning actually is. In science, one is sometimes
confronted with a situation in which it is easier to talk about a particular
phenomenon or to describe some of its properties than to state what the
phenomenon actually is. Thus, physicists, for instance, have been perfectly
able to measure certain qualities of quarks and to distinguish between sev-
eral types of quarks (which they have labelled red, green, etc.), without
necessarily fully understanding as yet the exact nature of quarks. In the
same way, it is easier to make statements about meaning in language than
to describe the character of the phenomenon that we call meaning. As a
consequence, various approaches towards the character of meaning have
been taken, which can perhaps be subsumed under the following three
headings:
− meaning as a purely linguistic phenomenon, a property of the linguistic
sign, which, according to the principles of Saussurean and post-
Saussurean structuralism, can be determined on the basis of the place of
a sign in the system of a language and its relationship to other signs;
− meaning as a psychological phenomenon, i.e. the ideas or concepts
speakers associate with a particular linguistic form; and
− meaning as a phenomenon of use, i.e. a property that can be determined
on the basis of how a word form is actually used by the speakers of a
speech community.

14.2 Meaning and reference

14.2.1 Bloomfield’s misconception of meaning

While it may be relatively difficult to arrive at a precise definition of mean-

ing, it is at least relatively clear what meaning is not. It is accepted today
that the meaning of a word must not be equated with the objects, people,
ideas etc. of the real world to which this word can apply. In fact, it can be
seen as one of the great weaknesses of the approach taken by the most in-
fluential scholar of American structuralism, Leonard Bloomfield, that he
did not make this distinction as rigorously as appropriate. He seemed to
think that the description of the meaning of a word benefits from scientific
222 Meaning

knowledge of the object described (such as its chemical composition).

Since such knowledge is not available for most objects one can describe
using words, the study of meaning was considered particularly difficult by
American structuralists. It is for this reason – not, as is often maintained in
the literature, that they considered it unimportant – that the study of mean-
ing was widely ignored in American structuralism. The point about this
argument is, however, that it is misguided, as is pointed out by Palmer
(21981: 22):
Bloomfield (1933: 139) argued that salt could be clearly defined as sodium
chloride, or NaCl. He was wrong to do so. Salt, for ordinary language, is
the substance that appears on our table. It is no less salt if its chemical com-
position is not precisely that of the chemist’s definition. Salt, for most of us
belongs with pepper and mustard, which do not lend themselves to any
simple scientific specification – and neither should salt in its everyday use.
Also, one could object, as Leech (1981: 2) does, that “Bloomfield’s argu-
ment implies a vision of an eventual period when everything would be ca-
pable of authoritative scientific definition, or in simpler words, when every-
thing there was to be known would be known about everything.”
A further argument against Bloomfield’s conviction that scientific
knowledge about an object is necessary for the description of meaning is
that it ignores the fact that such knowledge may not be available to the
majority of language users and one does not have to know that salt is NaCl
in order to use the word salt correctly.340 In fact, it is more than likely that
children acquire a perfectly appropriate use of the word salt long before
they learn that salt is sodium chloride.341
Whatever the relationship between knowledge about objects and their
properties and knowledge about word meanings may be – and some more
recent psycholinguistic approaches suggest that the two should not be seen
as entirely independent from one another – it is clear that knowledge about
objects (in terms of precise and scientifically accurate chemical, physical
properties etc.) is no prerequisite for mastering the meaning of a word that
can be used for that object.

340
Cf. Ullmann (1972: 59): “… one may wonder whether this is really the mean-
ing of the word for the average speaker who probably has no idea of the
chemical composition of salt”.
341
For an account of Bloomfield’s views see also Sampson (1980: 68–69).
 14 Semantics: meaning, reference and denotation 223

14.2.2 Denotation

In order to describe the relationship between the various entities involved it

makes sense to distinguish between the linguistic sign – with its two com-
ponents form and meaning – on the one hand, and entities of the extra-
linguistic world (which includes the possible worlds of our imagination) to
which linguistic signs can be applied on the other. Thus, on the one hand,
there is a word such as island – consisting of a form, namely /'aɪlənd/ or
<island>, and a meaning, which can be described as ‘a piece of land com-
pletely surrounded by water’.342 On the other hand, there are objects to
which this definition applies in the sense that if speakers of English talk
about such things, they can use the word island.
The set of objects, people etc. of the world to which a word can be ap-
plied is often called its denotatum. Thus the denotatum of the word island
is identical with the set of all islands which exist in the reality of the world
or in the imagination of the speakers. The relationship between a linguistic
sign and its denotatum is its denotation.

342
LDOCE5.
224 Meaning

14.2.3 Reference

14.2.3.1 The general notion of reference

What is important to note is that the denotation is a property of a word

which is independent of its use in any concrete speech situation. However,
when speakers use words they do not necessarily want to talk about the
whole class of objects, persons, etc. which the word denotes but rather they
often wish to identify one particular object, person etc. in order to say
something about them. In the context of Searle’s speech act theory (1969),
the act which speakers perform in such situations has been called refer-
ring. A speaker can refer to an object, person etc. by using a linguistic
expression, usually a phrase, through which it is possible to identify a par-
ticular member of the set comprising the denotatum. The member of the set
referred to is called the referent, the relationship between the respective
phrase and the referent is called reference.343

343
Compare Searle (1969: 26–29), Lyons (1977: 174–215) and Lipka (32002:
74–77). It should also be pointed out at this stage that, especially in research
prior to Searle (1969) and Lyons (1977), the distinction between denotation
and reference is not made in the same way. Thus, Ogden and Richards
(1923/101956: 11) use the model of a semiotic triangle to describe the rela-
tionship of linguistic signs and extralinguistic objects:

Ullmann (1972: 55–56) describes this model as follows: “On this reading,
there is no direct relation between words and the things they ‘stand for’: the
word ‘symbolizes’ a ‘thought or reference’ which in its turn ‘refers’ to the
feature or event we are talking about. There is nothing fundamentally new in
 14 Semantics: meaning, reference and denotation 225

Two things are worth noting here: firstly, a word as such does not have
reference nor does it have a referent. Reference only holds with respect to a
particular speech situation. This can easily be demonstrated by considering
sentences such as the following:344
(1a)AE The cultural landscape of Jersey – its Norman style farmhouses, the
narrow winding lanes and small fields, the French street names – reflect
a fascinating and complex history that has entwined the island in the
fate of two great nations: Britain and France for over one thousand
years.
(1b)AE Situated in the Bay of St. Malo the islands enjoy approximately over
2,000 hours of sunshine per annum.

____________________________
this analysis of meaning; the mediaeval schoolmen already knew that ‘vox
significat mediantibus conceptibus’ (the word signifies through the medium
of concepts).” For a critique of this model see Ullmann (1972: 54–58). For an
outline of the Ogden and Richards model against its behaviourist background
see Lyons (1977: 96–99).
344
Sources of examples in this section: http://www.jersey.com/index.asp,
http://www.gov.gg/, http://www.visitorkney.com/island.html, http://www.tate.
org.uk/stives/default.htm and Munisha Underhill, Perry Green 2000,
http://www.henry-moore.org/pg/henry-moore-research/henry-moores-
life/1898–1925 (last accessed 14 April 2010); and Tom Cross: Painting the
Warmth of the Sun, Tiverton/Cambridge: Westcountry Books/The Lutter-
worth Press (1995).
226 Meaning

(1c)AE The warmth of the welcome and the beauty of the islands will restore
your appetite and your joie de vivre, and enrich your visit to Orkney.
(1d)PW Thus in complete readiness to land he sat looking back at the island.
In all four sentences, the word island is used with the same meaning. Thus
the denotatum of the word, which is independent of its use in any particular
context, is identical with the set of all islands in the world. The referents of
the four uses of the word island are different, however. In (1a) the phrase
the island refers to Jersey, the phrase the islands refers to the islands be-
longing to the Bailwick of Guernsey in (1b) and to the Orkney islands in
(1c). The island in (1d) finally refers to the fictional island where the light-
house in Virginia Woolf’s To the Lighthouse is situated.
Secondly, the linguistic unit that carries reference is not the word but
what Searle (1969: 26–27) has described as a referring expression:
Any expression which serves to identify any thing, process, event, action,
or any other kind of ‘individual’ or ‘particular’ I shall call a referring ex-
pression. Referring expressions point to particular things; they answer the
questions “Who?” “What?” “Which?”
From a grammatical point of view it could be argued that it is the phrase
(not the word) that has the ability to function as a referring expression (>
11.3). It is quite obvious that it is one of the main functions of the deter-
miners (or pre-heads) in a noun phrase, for example, to establish which
elements of the denotatum of the head of the noun phrase are being referred
to in a concrete situation:
(2a)PW The long window in four sections of the large, sunfilled sitting room
overlooking the sea wall and the harbour became the principal subjects
of Patrick Heron’s paintings from 1947 onwards.
(2b)PW Heron made many paintings from St Ives subjects which responded to
the zestful enjoyment of freely applied colour.
(2c)PW In 1950 Patrick Heron was commissioned by the Arts Council to make a
large painting for the Festival exhibition ’60 paintings for ‘51’.

14.2.3.2 Definite and indefinite reference

Nevertheless, the notion of referring expression is not totally clear-cut. First

of all, Searle (1969: 27) recognizes a need to distinguish between definite
and indefinite referring expressions:
 14 Semantics: meaning, reference and denotation 227

Expressions beginning with the indefinite article, such as “a man”, as it oc-

curs in the utterance of the sentence, “A man came”, might be said to refer
to a particular man, but they do not serve to identify or to indicate the
speaker’s intention to identify an object in the manner of some uses of ex-
pressions with the definite article, such as “the man”.
The problem of the degree of definiteness of the kind of reference estab-
lished is also apparent from the sentences under (2). The reference of the
phrase Patrick Heron in (2a) is no doubt definite, as is probably also that of
Patrick Heron’s paintings. This is perhaps less clear in the case of many
paintings in (2b). In any case, the referential character of a large painting
in (2c) might be considered questionable in that it is not apparent from the
text at this point whether Heron actually painted this painting or not.
What is perhaps more important is that Searle (1969: 27), as later Lyons
(1977: 185),345 finds it necessary to speak of
non-referring uses of expressions formed with the indefinite article: e.g., the
occurrence of “a man” in the utterance of “A man came” is to be distin-
guished from its occurrence in the utterance of “John is a man”. The first is
referential, the second predicative.
Similarly, while this great twentieth-century sculptor in
(3)AE The Gallery also manages the Barbara Hepworth Museum and Sculp-
ture Garden, which gives a remarkable insight into the work of this
great twentieth-century sculptor.
is clearly referential in character (which has Barbara Hepworth as its refer-
ent), this is not so clear in the case of a Penzance girl in
(4)PW Margo Maeckelberghe was a Penzance girl.
The same applies to as a painter and draughtsman and as a schoolboy in
(5)PW Patrick showed early talent as a painter and draughtsman. As a school-
boy he was aware of the work of Cézanne and Sickert ...
or a sculptor in346

345
Cf. Lyons (1977: 185): “A definite noun-phrase may occur as the comple-
ment of the verb ‘to be’ and it may then have a predicative, rather than a ref-
erential, function.”
346
Examples (6) and (7a) taken from http://www.henry-moore.org/pg/henry-
moore-research/henry-moores-life/1898–1925 (last accessed 14 April 2010).
228 Meaning

(6)HM During these years Moore began carving in wood and modelling clay,
and he consciously decided to become a sculptor, …
where one could argue that these noun phrases do not contribute to the
identification of an element of the denotatum and thus are not referential.
However, Searle (1969: 27) points out that Russell (1919: 172) treats such
cases as (4) as identity statements. (5) and (6) could be analysed in a similar
way, attributing indefinite reference to painter, draughtsman, schoolboy
and sculptor. Otherwise, the analysis would result in a rather contradictory
treatment of similar noun phrases such as
(7a)HM Henry Moore went to infant and elementary schools in his home town,
and then in 1910, like several of his brothers and sisters he won a
scholarship to Castleford Secondary School, which later became a
Grammar School under subsequent reforms and administrative devel-
opments.
(7b)BNC Joanne works at a grammar school and almost all her teaching experi-
ence has been at this school.
If, as would be in line with the view held by Searle and Lyons, one were to
analyse a grammar school in (7b) as an instance of a singular indefinite
referring expression, and, possibly, in (7a) as a non-referring use, then pre-
sumably only because the school had already been mentioned previously in
the text in (7a) but not in (7b). Since the whole question of reference has
been subject to considerable debate in linguistics and philosophy, it may
suffice at this point to point out that there are differences as to the referen-
tial character of certain phrases in certain contexts. Whether there is any
great descriptive advantage in either solution is perhaps to be doubted. In
any case, it has become obvious that if one makes a distinction between
denotation and reference in the way suggested by Lyons (1977), reference
is a speaker- and utterance-oriented notion, which, strictly speaking, is only
of relatively marginal interest to semantics. The act of referring is appropri-
ately seen by Searle as part of performing a speech act and thus falls under
the scope of pragmatics and speech act theory (> 17.3), whereas referential
relations also constitute an important subject in the analysis of texts and
textual structure (> 18.2). What is more important to semantics is the rela-
tionship between a linguistic sign and its denotatum, where, of course, the
denotatum can be seen as the set of all possible referents.
 14 Semantics: meaning, reference and denotation 229

14.3 The scope of meaning

Up to a point, meaning and denotatum can be seen as complementary no-

tions. Thus, the word with the meaning ‘a boat etc. for transporting people,
vehicles or goods across a stretch of water’ (NSOED) can be applied to any
member of the set of ferries in the real world, which make up its denotatum.
Likewise, any ferry can be referred to by the word ferry. This parallel only
holds to a certain extent, however.
One of the most famous examples to illustrate the fact that meaning may
comprise more than the relationship of denotation was introduced by the
German philosopher Frege, who pointed out that the words Morgenstern
and Abendstern (as well as the word Venus) have the same denotatum but
still would not be considered to be synonymous.347
In a similar way, one could argue that words such as autumn and fall or
child and kid do not differ with the respect to their respective denotata but
rather with respect to who uses them (autumn being a typically British, fall
a typically American use) or in which situation they are being used (kid
being typical of an informal style). They thus contain information which
goes beyond the identification of their denotatum. Such considerations have
led many linguists to talk about different types of meaning. Leech (21981:
23), for instance, introduces a category of social meaning to cover regional,
social and stylistic differences between words.348 Whether these should be
treated as differences of meaning is by no means uncontroversial, however.
Referring to autumn and fall, Palmer (21981: 89) argues:
The works of dialectologists are full of examples like these. They are espe-
cially interested in words to do with farming; depending where you live you
will say cowshed, cowhouse or byre, haystack, hayrick or haymow. … But
these groups of words are of no interest at all for semantics. Their status is
no different from the translation-equivalents of, say, English and French. It
is simply a matter of people speaking different forms of the language hav-
ing different vocabulary items.
Thus, while autumn and fall clearly provide a reader or hearer with infor-
mation about the speaker’s or writer’s background, this kind of information
need not necessarily be seen as one of meaning. Similarly, the stylistic

347
Cf. Frege (1906/2000: 685).
348
Leech (21981: 9–23) distinguishes between the following seven types of
meaning: conceptual meaning, connotative meaning, social meaning, affec-
tive meaning, reflected meaning, collocative meaning and thematic meaning.
230 Meaning

value of a word can be seen as part of its meaning, as in Leech’s (21981)

model, or as separate property of words independent from meaning.
A further element of meaning that is sometimes seen as a separate level
of meaning is that of “emotive or evaluative” meaning, as Palmer (21981:
90) points out:
Some semanticists have made great play with the emotive difference be-
tween politician and statesman, hide and conceal, liberty and freedom, each
implying approval or disapproval. The function of such words in language
is, of course, to influence attitudes. … Nevertheless, it is a mistake to sepa-
rate such emotive or evaluative meaning from the ‘basic’ ‘cognitive’ mean-
ing of words for three reasons: First, it is not easy to establish precisely
what cognitive meaning is, and certainly not reasonable to attempt to define
it in terms of reference to physical properties. On such a definition, most
verbs and adjectives would have little or no cognitive meaning. Secondly,
there are words in English that are used PURELY for evaluative purposes,
most obviously the adjectives good and bad, but it is not normally assumed
that they have no cognitive meaning. Such words are of interest to moral
philosophers, but should not, I believe, have any special place in linguistics.
Thirdly, we make all kinds of judgments and do not merely judge in terms
of ‘good’ and ‘bad’. We judge size and use the appropriate terms – gi-
ant/dwarf, mountain/hill, etc., and we make other kinds of judgments in our
choice of words. The meaning of words is not simply a matter of objective
facts; a great deal of it is subjective and we cannot clearly distinguish be-
tween the two.
As far as word meaning is concerned, there is probably a lot to be said in
favour of Palmer’s rejection of different types of meaning.349 Nevertheless,
what we consider to be part of the meaning is yet another case of a rela-
tively arbitrary decision made by the investigating linguist, the appropriate-
ness of which must be judged against the overall purpose of investigation.

349
Note, however, that some of Leech’s seven types of meaning explicitly refer
to sentence meaning. This applies in particular to his categories of thematic
meaning, and also to affective meaning: “Factors such as intonation and
voice-timbre … are also important here. … Affective meaning is largely a
parasitic category in the sense that to express our emotions we rely upon the
mediation of other categories of meaning – conceptual, connotative, or stylis-
tic. Emotional expression through style comes about, for instance, when we
adopt an impolite tone to express displeasure …, or when we adopt a casual
tone to express friendliness.” (Leech 21981: 16)
 14 Semantics: meaning, reference and denotation 231

This is also true of another type of meaning that is often identified –

connotation.350 In this case, the argument is not so much about whether a
particular feature of a word – such as its stylistic value – is to be seen as
part of its meaning or as a category of its own. Rather, the question is
whether connotation is a linguistic property at all. Connotations of words
can be characterized by three properties:
− Connotations are associations speakers have with a word beyond its
(denotative or conceptual) meaning.
− Connotations arise from the properties of (elements of) the denotatum.
− Connotations are not necessarily shared by the whole speech community
and “they vary considerably … according to culture, historical period
and the experience of the individual” (Leech 21981: 13).
Thus it is quite possible that, with a word such as nuclear power station,
some speakers connote positive attributes such as ‘modern technology’
whereas others may connote negative properties such as ‘danger’, ‘threat to
health or environment’ etc. Similarly, Palmer (21981: 92)351 points out that
“woman has the connotation ‘gentle’ and pig the connotation ‘dirty’” and
argues that “… this is not a matter of the meaning of words or even of
meaning in general. It rather indicates that people (or some people) believe
that women are gentle and pigs dirty”. While in these cases one might in-
deed argue that associations of this kind are related to the denotata, i.e. that
they are clichés attributed to women or pigs by particular groups of speak-
ers, the negative connotations of words such as vergasen or Führer in Ger-
man cannot be explained away quite so easily as a purely extralinguistic
phenomenon. In any case, the study of connotations (whether one considers
them part of semantics or not) touches upon the question of whether one
can (should or must) make a distinction between linguistic and extra-
linguistic knowledge or not. This is also emphasized by Leech (21981: 12–
13):
It will be clear that in talking about connotation, I am in fact talking about
the ‘real world’ experience one associates with an expression when one
uses or hears it. Therefore the boundary between conceptual and connota-

350
The term connotation is often contrasted with denotative or conceptual mean-
ing (Leech 21981: 9) but this must not be confused with the term denotation
as defined above.
351
Compare also Leech’s (21981: 12–13) discussion of the connotations of
woman.
232 Meaning

tive meaning is coincident with that nebulous but crucial distinction … be-
tween ‘language’ and the ‘real world’. This accounts for the feeling that
connotation is somehow incidental to language rather than an essential part
of it, and we may notice, in confirmation, that connotative meaning is not
specific to language, but is shared by other communicative systems, such as
visual art and music. Whatever connotations the word baby has can be con-
jured up … by a drawing of a baby, or an imitation of a baby’s cry.
15 Meaning relations

15.1 Polysemy and homonymy

15.1.1 Polysemy and homonymy in linguistic analysis

Irrespective of how meaning is defined or how the description of meaning

is approached, there is consensus in linguistics that meaning is one of the
components of the linguistic sign that is firmly connected with its form. In
9.1.1 it was pointed out that it may make sense to distinguish between lex-
emes and lexical units. In the terminology employed by Cruse (1986: 80),
“a lexical unit is the union of a single sense with a lexical form”, whereas
“a lexeme is a family of lexical units”. This distinction presupposes that it
is possible for a lexeme to have more than one meaning and that it is possi-
ble to distinguish between these.
That the same word form can occur with different meanings can easily
be demonstrated: One would not expect there to be any disagreement
amongst speakers that bank, ear, press and royalty are not always used in
the same meanings in the following sentences:
(1a)BNC In that same year I was posted to South Shields on the south bank of the
River Tyne.
(1b)BNC I will not accept Eurocheques because when I pay them into the bank I
am losing money because of the transaction charges.
(2a)BNC “You are not at all satisfied”, Susan said softly into the President’s ear.
(2b)BNC If you crush an ear of barley, however, you will not get beer.
(3a)BNC Does freedom of the press mean freedom to choose its own standards?
(3b)BNC I was not to answer the phone; above all I was not to speak to the press.
(3c)BNC The technology of printing now has virtually nothing in common with
the wooden presses and lead type of 15th-century pioneers like Johan-
nes Gutenberg or Nicolas Jensen.
(3d)BNC Two types of presses are used, the horizontal press and the traditional
Champagne press.
234 Meaning

(4a)BNC Apart from demonstrating one of the unwavering laws of British jour-
nalism, that nothing sells newspapers like royalty, and nothing makes a
better editorial column than declamations of simple patriotism, the cu-
rious thing about these assaults is how much they belong to a period.
(4b)BNC And contact with royalty gives the lord lieutenants real influence.
(4c)BNC But then Michael gets an even bigger royalty rate than Madonna.
(4d)BNC This royalty is divided between the writer and the publishing company.
On the basis of such sentences, one could argue that, for instance, (4a-b)
and (4c-d) represent two different meanings of the lexical form royalty and
thus establish the following two lexical units:352

Similarly, two lexical units can be established for ear.

This does not necessarily mean, however, that these lexical units can neces-
sarily be subsumed under one lexeme. In many approaches to semantics, a
distinction is made between polysemy and homonymy:

A lexeme is polysemous when it has more than one meaning, i.e. when
it comprises more than one lexical unit.

Two lexemes (each consisting of one or more lexical units) are ho-
monymous if their forms are identical.353

352
Cf. NSOED.
 15 Meaning relations 235

The problem with the distinction between homonymy and polysemy is that
it can only be made reliably in diachronic linguistics. If one takes the his-
tory of the language into account, then the criterion to distinguish between
polysemy and homonymy is the etymology of the word forms in question:
it makes sense to say that if two lexical units have the same etymological
root, they can be considered as belonging to the same lexeme (which then
must be regarded as polysemous) and if they have different roots they be-
long to two different lexemes (that are homonyms).
On the basis of this criterion, bank and ear present cases of homonymy,
whereas press and royalty are to be classified as polysemous lexemes. Thus
bank1 (‘sloping margin of a river etc.’) is related to Old Icelandic bakki,
whereas bank2 (‘financial institution’) is derived from French banque or its
Italian source banca. Similarly, ear1 (‘organ of hearing’) is derived from
Old English ēare and related to German Ohr, whereas ear2 (‘spike or head
of corn’) is derived from Old English ēar and related to the German word
Ähre. Sound changes operative since Old English have led to a situation in
which two forms which originally were distinct from one another have
become identical. In contrast to bank and ear, press1 and press2 and roy-
alty1 and royalty2 cannot be traced back to different forms in the history of
the language. Rather, one meaning developed out of another meaning of the
lexeme. Thus it is easy to see how the following meanings identified in the
New Shorter Oxford English Dictionary (NSOED) are related: “10a … a
machine for printing”, “10b a place or business of which the printing-press
is the centre …” and “11a … newspapers, journals … generally”. In the
case of royalty, the relationship between the two meanings illustrated above
is perhaps not quite so obvious, but again the meanings identified in the

____________________________
353
Note that lexemes that are only identical in the phonological form are called
homophones, whereas lexemes that are identical in spelling are called homo-
graphs.
236 Meaning

NSOED suggest a plausible semantic development: “A prerogative or right

granted by a monarch, esp. in respect of jurisdiction or over minerals”, “A
payment made by a person working on a mine, obtaining oil, etc., to the
owner of the site or the mineral rights over it; a payment made for the use
of a patent or a technical process; a payment made to an author, editor or
composer for each copy of a work sold or performed …”.
Synchronically, of course, etymological considerations do not provide a
criterion to establish a distinction between homonymy and polysemy. It has
thus been suggested that within a synchronic approach homonymy can be
distinguished from polysemy by arguing that only in the latter case “two
meanings are psychologically related” in the sense that “present-day users
of the language feel intuitively that they are related, and therefore tend to
assume that they are ‘different uses of the same word’” (Leech 21981:
227).354 Leech (21981: 228) points out that etymological relatedness and
psychological relatedness need not coincide and argues that in cases such as
ear1 and ear2, which on the basis of their etymological development are
clearly instances of homonymy, “people often see a metaphorical connec-
tion between them, and adjust their understanding of the words accord-
ingly”.355 In the same way one might argue that the historical relatedness of
the meanings of press1 and press2 or royalty1 and royalty2 is not necessarily
something that the (or all) speakers of the language today would be aware
of when using these words.
Since, as Bloomfield (1933/1935: 436) points out, “the degree of near-
ness of the meanings is not subject to precise measurement”, a synchronic
distinction between homonymy and polysemy is “bound to be subjective
and to some extent arbitrary” (Ullmann 1972: 178) and as such not particu-
larly useful.356

354
In a more precise form, Leech (21981: 228) states: “We may, in fact, say that
two lexical meanings are ‘psychologically related’ if a user of the language is
able to postulate a connection between them by lexical rules, e.g. by the rule
of metaphoric transfer.”
355
Compare also Ullmann’s (1972: 164–165) discussion of such examples.
356
Compare Lyons (1968: 405–407), Lipka (32002: 153–157) or Kastovsky
(1982: 121–124).
 15 Meaning relations 237

15.1.2 Psycholinguistic and lexicographical implications

It has to be said that for the purposes of a synchronic semantic analysis the
distinction between homonymy and polysemy can be regarded as pretty
irrelevant anyway since its subject is the analysis of meanings (or lexical
units).357 It is highly relevant, however, as soon as one tries to investigate
the organization of the mental lexicon and tries to discover how people
access meaning.
This question is closely related to the way that multiple meaning ought
to be treated in dictionaries. Historical dictionaries such as the OED or the
NSOED can obviously draw upon the etymological criterion and provide
two separate entries for bank1 and bank2 or ear1 and ear2 respectively and
deal with the different meanings of press or royalty under a single entry.
But such a principle of organization would hardly be appropriate in a syn-
chronic dictionary, especially in a foreign learners’ dictionary. For this
reason, such dictionaries often make use of the criterion of psychological
relatedness so that sole1 (‘bottom surface of the foot’) and sole2 (‘fish’),
which historically present a case of polysemy, are treated as homonyms in
OALD2 (1963) and LDOCE1 (1978). Because of the inevitable arbitrari-
ness of such decisions outlined above, there is an increasing tendency in
such dictionaries to generally only provide one entry for one lexical form
belonging to a particular word class, thus making polysemy a general prin-
ciple of lexicographical organization.358

15.2 Ambiguity

The use of polysemous lexemes or homonyms in sentences can result in

ambiguity. For instance, a sentence such as
(1c)BNC How sweet the moonlight sleeps upon this bank!
is at least potentially ambiguous in that it is not clear which of the mean-
ings of bank is being referred to.359 Usually, ambiguity can be resolved by
the (linguistic or extralinguistic) context, however, as in

357
Cf. Kastovsky (1982: 122).
358
For a discussion of polysemy and homonymy as organizational principles in
dictionaries see Herbst and Klotz (2003).
359
For a discussion of intended ambiguity compare Ullmann (1972: 188–192).
238 Meaning

(2c)BNC If the subject is instructed to attend to one ear and ignore the sounds
coming into the other ear, all the sounds in the attended ear will pro-
duce an enhanced N100 component in the ERP.
where ear quite clearly illustrates the same meaning of ear as in (2a).

15.3 Problems of identification of meanings and lexical units

Irrespective of the question of homonymy and polysemy, it can be consid-

ered fairly uncontroversial that bank1 and bank2 or press1 and press2 repre-
sent different meanings of bank and press, i.e. different lexical units. How-
ever, this is not always as straightforward. For instance, it is far less clear
how many meanings one would associate with the different occurrences of
the lexical form meet in the following sentences:
(5a)BNC Let’s meet for lunch tomorrow.
(5b)BNC But the pound will face another test on Thursday when the Bundesbank
council meets.
(5c)BNC She first met him when she studied at Trinity College, Dublin, as he
neared the end of his degree course.
(5d)BNC Sampdoria and Barcelona will meet in the European Cup final at Wem-
bley on May 20.
(5e)BNC The stack emission comfortably meets the American standard for
breathing air.
One could argue that the kind of meeting that takes place between two
sports clubs as in (5d) is sufficiently different from that of (5a), for in-
stance, that it might be justified to talk of two different meanings in this
case. Similarly, a case could be made out for treating (5c), which would be
translated by kennenlernen into German, as a separate meaning, while it
could also be said that meeting a person for the first time is getting to know
them. One could establish a difference on the basis of arranged meetings
and meetings by chance, for instance. Similarly, the use in (5e) could be
treated as a different meaning on the grounds that it does not describe a
meeting between two people or groups of people. On the other hand, it
could be argued that in all five cases the verb describes more or less the
same phenomenon, the coming together of two entities, and that whatever
differences between these types of meeting exist they are obvious from the
nature of these meetings and not caused by the verb meaning. This is the
 15 Meaning relations 239

kind of problem that any lexicographical description of words is faced with.

The fact that the number of meanings established for words such as meet
hardly ever coincide between different dictionaries can be taken as an indi-
cation of the fact that no clear decision is possible in such cases.360

15.4 Structural semantics

15.4.1 The idea of contrast

The first approach to the description of meaning to be discussed here is

known as structural semantics. It can be characterized as a purely language-
internal approach in that it attempts to discover semantic properties of lexi-
cal units without taking psychological or extralinguistic factors (such as the
denotatum of a lexical unit) into account.
De Saussure’s conviction (> 2.1.4) that the value of a linguistic sign can
be determined by its relationship to other linguistic signs in the system is
reflected in structural semantics by the idea that word meanings or elements
of word meanings can be elicited by contrasting them with other word
meanings within the same system. One method that has been frequently
employed in this context is that of word fields or lexical fields (> 16.1). It is
an illustration of one of the basic assumptions of a structuralist approach to
semantics, namely that the vocabulary of a language (at least in parts) dis-
plays structure,361 as is expressed by Lyons (1968: 443):
Acceptance of the structural approach in semantics has the advantage that it
enables the linguist to avoid commitment on the controversial question of
the philosophical and psychological status of ‘concepts’ or ‘ideas’ … As far
as the empirical investigation of the structure of language is concerned, the
sense of a lexical item may be defined to be, not only dependent upon, but
identical with, the set of relations which hold between the items in question
and other items in the same lexical system.

360
Palmer (21981: 105–106) points out the limited usefulness of tests to establish
different meanings of a lexeme: “It has been suggested that one test of ambi-
guity is the ‘co-ordination test’. … But these tests do not help, for judgments
about co-ordination depend upon judgments about sameness of meaning, and
the doubtful cases remain. If we judge that Mary cried and so did Ruth is ac-
ceptable in the sense that Mary wept and Ruth shouted, it will be because we
do not regard cry as ambiguous.”
361
Cf. Coseriu (1973: 10–11) and Wotjak (1971: 54–59).
240 Meaning

15.4.2 Semantic relations

15.4.2.1 Hyponymy: unilateral entailment

One of the central semantic relations that can be taken as an indication of

the structuredness of vocabulary is that of hyponymy, which can be de-
scribed in terms of unilateral entailment (Cruse 1986: 88–89), where en-
tailment is defined as follows: “A proposition P is said to entail another
proposition Q when the truth of Q is a logically necessary consequence of
the truth of P (and the falsity of P is a logically necessary consequence of
the falsity of Q)” (Cruse 1986: 14). Thus
(6a) This is a seal.
entails
(6b) This is an animal.
(7a)BNC Set in the dipping wooded hills of Lambton Park, through which the
wide and tidal Wear cuts a deep valley, this grandest of early Georgian
houses looks south down a long wide avenue, smothered solid with daf-
fodils in the spring.
entails
(7b) Set in the dipping wooded hills of Lambton Park, through which the
wide and tidal Wear cuts a deep valley, this grandest of early Georgian
houses looks south down a long wide avenue, smothered solid with
flowers in the spring.
whereas (6b) and (7b) do not entail (6a) or (7a) respectively. Hyponymy
can then be defined as the relationship between a hyperonym and a hypo-
nym:362

A lexical unit X is a hyperonym (or superordinate) of a lexical unit
Y if the meaning of Y entails the meaning of X.

A lexical unit Y is a hyponym (or subordinate) of a lexical unit X if
the meaning of Y entails the meaning of X.

Two lexical units Y and Z are co-hyponyms if they are hyponyms of
the same hyperonym X.
Thus one can identify fly, drive and walk as co-hyponyms of the hypero-
nym go, because

362
See Cruse (1986: 88–89).
 15 Meaning relations 241

(8a) We flew to Kirkwall.

(8b) We drove to Kirkwall.
(8c) We walked to Kirkwall.
all entail
(8d) We went to Kirkwall.
without (8d) entailing (8a), (8b) or (8c). The logical relation of unilateral
entailment can thus be used to establish a particular type of lexical relation.
However, this must be seen as no more than a methodological device that
can be fruitfully applied in structural semantics. In particular, this does not
mean that this kind of test would work in all natural language contexts.
Thus, Cruse (1986: 90) points out that while
(9a)QE Flowers are prohibited.
entails
(9b)QE Dandelions are prohibited.
(10a)QE Flowers make an acceptable present.
does not entail
(10b)QE Dandelions make an acceptable present.
Furthermore, fly could not be replaced by its hyperonym go in a sentence
such as
(8e)BNC We flew to Kirkwall with British Airways, which flies to Orkney and
Shetland regularly, from Glasgow and Edinburgh via Inverness.

15.4.2.2 Synonymy: bilateral entailment

The method of investigating entailment relationships can also be used to

define the notion of synonymy,363 which can then be defined as

363
Compare Cruse (1986: 88), who provides the following definition of cogni-
tive synonymy: “X is a cognitive synonym of Y if (i) X and Y are syntacti-
cally identical, and (ii) any grammatical declarative sentence S containing X
has equivalent truth-conditions to another sentence S1, which is identical to S
except that X is replaced by Y.” For a detailed discussion of synonymy see
Cruse (1986: 265–294).
242 Meaning

Two lexical units are synonymous if they entail each other.

If one considers purchase and buy to be synonyms, then one could argue
that
(11a)BNC Saatchi is known to have purchased five major Warhols in the last six
months, for extremely high prices, which few would argue is the action
of someone dispersing their collection.
entails
(11b) Saatchi is known to have bought five major Warhols in the last six
months, for extremely high prices, which few would argue is the action
of someone dispersing their collection.
and that (11b) also entails (11a). The same holds true in the case of (11c)
and (11d):
(11c)BNC The Association has recently purchased a large, detached house in a
residential area of Melton Mowbray, Leicestershire.
(11d) The Association has recently bought a large, detached house in a resi-
dential area of Melton Mowbray, Leicestershire.
Nevertheless it has to be said that mutual entailment is a definition and not
a test of synonymy. The analysis of sentences such as (11a) and (11b) is
based on the assumption that buy and purchase have the same meaning,
otherwise the relationship of mutual entailment would not hold. In fact,
there are many linguists who consider true synonymy an extremely rare
phenomenon in language. Palmer (21981: 89), for instance, maintains “that
there are no real synonyms, that no two words have exactly the same mean-
ing” and points out that even in pairs such as brotherly/fraternal,
buy/purchase and world/universe the “‘native’ words are often … less
learned”.364 In particular, when one considers meaning from the point of
view of use, it makes sense to argue that while in particular sentences or
contexts there is no difference between the contribution that two words that
can be interchanged in a particular position make to the meaning of a sen-
tence, this does not mean that the lexical units as such are really synony-
mous.

364
For a discussion of synonymy see e.g. Palmer (21981: 88–93).
 15 Meaning relations 243

15.4.2.3 Semantic oppositions

The logical relation of entailment, which was drawn upon in the definition
of hyperonymy and synonymy, can also be used to identify a type of se-
mantic opposition which is exemplified by such pairs as male : female and
good : bad. Thus, for instance,
(12a)BNC Her evidence about being telephoned late at night on 11 January 1987
(…) was that the caller was female and that the call lasted three to five
minutes.
entails
(12b) … that the caller was not male.
just as
(12c)BNC The shape was close enough for Henry to see that it was male.
entails
(12d) The shape was close enough for Henry to see that it was not female.
In the same way one could argue that
(13a)BNC For once the forecast was good.
entails
(13b) For once the forecast was not bad.
and
(13c)BNC Subjectivity in this field is bad.
entails
(13d) Subjectivity in this field is not good.
Nevertheless there is an important difference between the two cases since
(12b) entails (12a) whereas (13b) does not entail (13a). This is due to the
fact that meanings such as those of the lexical units good or bad are grad-
able in a way in which this is not true of pairs such as male and female.365

365
Compare Lyons’s (1977: 272) discussion of these differences in terms of the
traditional logical distinction between contradictories and contraries, which
he however considers insufficient to cover linguistic differences of this kind.
See also Leech (21981), Palmer (21981) or Cruse (1986). Note, however, that
corpus evidence suggests that the distinction between gradable and non-
244 Meaning

(13e)BNC The pollution was pretty bad up there.

(13f)BNC One of the great unanswered questions in my life is why Mrs Spence
was going to antenatal classes that afternoon when, as she had already
had one child, she presumably had a pretty good idea of what to expect.
A further characteristic of pairs such as good and bad or far and near is that
one of the two terms can be considered to be marked and the other un-
marked. The unmarked member of the pair is the one that is used in neutral
questions for example:
(13g)BNC How good is your photographic memory?
(14a)BNC How far are money and the media responsible for the misbehaviour of
football supporters?
Questions with the marked term are somehow presupposing part of the
answer:
(13h)BNC The police will have put out checks on every hospital for miles. How
bad is it?
(14b)BNC How near is your proposed venue to airports (if it involves international
visitors), and the road and rail links?
In the case of pairs such as male and female no such distinction between
marked and unmarked members of the pair can be made. Because of these
differences it may seem appropriate to distinguish between different types
of oppositeness and to refer to pairs such as good : bad, far : near as anto-
nyms and to pairs such as male : female as complementaries.366 Accord-
ing to Cruse (1986: 204) antonyms in this sense share the following charac-
teristics:
____________________________
gradable adjectives is by no means as clear as is sometimes suggested:
Lewis’s debate in the Preface with other Milton critics with Saurat, Eliot, I.
A. Richards or Leavis all seems pretty dead to us now, since not many people
would read any of the aforementioned critics nowadays as critics unless they
were interested in the history of criticism for its own sake.BNC Many provin-
cial towns in the United States have lost almost all their shops to suburban
shopping malls and have become very dead places as a result.BNC He was
very dead, and there was blood all over the floor.BNC It remained very male,
in its content, and in its style.BNC
366
See Lyons (1977: 279), who however also points at the arbitrary nature of the
distinction: “Opinions will differ on the advisability of drawing a termino-
logical distinction in one way rather than another.”
 15 Meaning relations 245

(i) they are fully gradable (…)

(ii) members of a pair denote degrees of some variable property such as
length, speed, weight, accuracy, etc.
(iii) when more strongly intensified, the members of a pair move, as it were, in
opposite directions along the scale representing degrees of the relevant
variable property. Thus, very happy and very light, for instance, are more
widely separated on the scale of weight than fairly heavy and fairly light.
(iv) the terms of a pair do not strictly bisect a domain: there is a range of val-
ues of the variable property, lying between those covered by the opposed
terms, which cannot be properly referred to by either term.
Complementaries, on the other hand, are characterized by Cruse (1986:
198) as follows:
The essence of a pair of complementaries is that between them they exhaus-
tively divide some conceptual domain into two mutually exclusive com-
partments, so that what does not fall into one of the compartments must
necessarily fall into the other. There is no ‘no-man’s land’, no neutral
ground, no possibility of a third term lying between them. Examples of
complementaries are: true : false, dead : alive, open : shut, hit : miss (a tar-
get), pass : fail (an examination).
This is illustrated by sentences such as
(12e)BNC He couldn’t say whether the person was male or female, because the
curtain was drawn halfway across.
It is interesting to see that differences in the entailment relations holding
between antonyms and complementaries can be used to illustrate the differ-
ences between them but that they do not suffice to characterize them fully,
since, as Palmer (21981: 97) points out, in “both cases we must also show
that they belong to the same semantic system or field”. In the case of com-
plementaries “it is not true to say that, if something is not male, it is female,
since it could also be inanimate”. In the same way
(15a) The milk is sour.
entails
(15b) The milk is not fresh.
but (15b) does not entail (15a) since the lexical relationship between sour
and fresh is far too complex to be established on a scale of antonymy.
Apart from antonymy and complementarity, a third type of oppositeness
can be established, for which Lyons (1977: 279) uses the term converse-
246 Meaning

ness but which is more commonly treated as relational opposition.367

This type of oppositeness is characterized by the fact that the members of
the opposition pair express the same relation between two entities, but do
so from a different perspective (which is why perspectival opposition
might be an even more appropriate term to use). This type of opposition
can be exemplified by pairs such as buy : sell, teacher : pupil or husband :
wife.
(16a)BNC They belonged to Hammersmith Council who had bought them from the
BBC for something like three million pounds.
(16b) … the BBC sold them to Hammersmith Council for something like three
million pounds.
(16c)BNC The group did not exercise its rights but sold them to a third party for
£200,000.
(16d) … the third party bought them for £200,000.
(17a)BNC By far the most important of the Venetian composers, after Monteverdi,
was his pupil Francesco Cavalli.
(17b) Monteverdi was Francesco Cavalli’s teacher.
(17c)BNC Miller is a friend of Pringle’s: he was her teacher at University College.
(17d) Pringle was Miller’s pupil at University College.
The distinction between antonymy, complementarity and converseness
shows that oppositeness can be expressed in different ways in the vocabu-
lary of a language.368

367
See Leech (21981: 102–106), Palmer (21981: 98–100), Cruse (1986).
368
For further types see Lyons (1977: 281–287), Leech (1981: 96–108) or Cruse
(1986: 223–264). For oppositeness see also Kastovsky (1982: 131–139).
Compare also the critical discussion by Lipka (32002: 163–165).
16 Ways of describing meaning

16.1 Componential analysis

In the previous chapter it was demonstrated that a number of relations be-

tween lexical units can be established such as hyponymy, different types of
oppositeness and possibly synonymy. This can be taken as a confirmation
of the basic conviction of structural semantics, that the lexical units of a
language form a system and that the meanings of the lexical units can be
determined on the basis of the relations which they enter into with the other
lexical units in that system. While establishing or revealing such relations
between lexical items does not describe the meaning of the lexical units as
such, such relations can be used to contrast the meanings of lexical items
and to discover elements of their meaning on this basis. As pointed out
above, one approach that has been widely used in this context, especially
within the European tradition of structuralist semantics, is that of the analy-
sis of word fields or lexical fields. A lexical field can be defined as a set
of lexical units that belong to the same word class and have at least one
specific semantic component in common.369 This semantic component can
exist as a lexical unit in a language, in which case it is the hyperonym of all
members of the lexical field. Thus, house, cottage, palace, castle, railway
station can be seen as members of a lexical field with building being the
hyperonym of all the members of this field. Kastovsky (1982: 84) points
out, however, that the field of adjectives denoting temperature in English
(of which hot, warm, cool, cold etc. are members) lacks such a hyperonym,

369
See Lipka (32002: 167–168), who uses the further criterion that “field-
membership must be established by objective procedures”. For a detailed
treatment of the history of field theory see Lyons (1977: 250–261). For dif-
ferences in terminology see Lipka (32002: 168), who uses the term lexical
field for fields “consisting of simple or complex lexemes” and word field for
fields “exclusively containing morphologically simple items”. For a critical
account compare Leisi (1973). See also Coseriu (1973: 53). For a discussion
of various approaches and definitions see also Kastovsky (1982: 124–128).
248 Meaning

for example.370 In any case, lexical fields provide a suitable basis for estab-
lishing contrasts of meaning in different lexical units.
Leech (21981: 89), who argues that the “analysis of word-meanings is
often seen as a process of breaking down the sense of a word into its mini-
mal components”, illustrates this approach by contrasting the meanings of
man, woman, boy, girl.

‘male’ ‘female’

‘adult’ ‘man’ ‘woman’

‘young’ ‘boy’ ‘girl’

‘human’

The diagram shows two dimensions of meaning: that of ‘sex’ and that of
‘adulthood’; a third dimension is presupposed by the isolation of the field as
a whole: that between ‘human’ and ‘non-human’ species.
In a slightly more abstract form, these components can be represented as:371
man: + HUMAN + ADULT + MALE
woman: + HUMAN + ADULT – MALE
boy: + HUMAN – ADULT + MALE
girl: + HUMAN – ADULT – MALE
The method of analysing meanings in terms of semantic components which
are gained through contrasting meanings of lexical units (in a lexical field)
is known as componential analysis. 372

370
Coseriu uses the term Archisemem to characterize the semantic features
common to all members of a word field, and Archilexem for cases of lexical
units representing such features. Cf. Coseriu (1973: 54) and also Kastovsky
(1982: 84).
371
See Leech (21981: 90).
372
While it has been frequently employed by European structural semanticists,
often in connection with lexical field analysis, the theory has also received an
important impetus by proposals made within an early generative framework,
especially by Katz and Fodor (1963). For detailed accounts of this approach
 16 Ways of describing meaning 249

One aim of componential analysis in semantics is to draw up an inven-

tory of semantic components similar to that of the distinctive features in
phonology.373 Some proponents of the theory hoped that within this ap-
proach it would be possible to describe any word meaning as a combination
of semantic components (similar to the way that any phoneme in any lan-
guage can be described with a very limited set of distinctive features),
which would make descriptions of this kind particularly suitable for ma-
chine translation, for example. For this to be possible, semantic components
would have to meet four conditions, namely that they should be
− atomic, i.e. it should be impossible to split them up into further semantic
components,
− recurrent, i.e. not just be designed to cover the meaning (or part of the
meaning) of only one lexical unit,
− binary, i.e. describable in terms of ± values, and
− universal, i.e. applicable to all languages.374
It is relatively easy to show that achieving this goal is very problematic. It
is true that semantic components such as the ones identified above are re-
current (and probably also universal) in that they can fruitfully be applied
to other lexical units such as stallion, mare, filly etc. As to atomicity, it has
to be noted that there is considerable disagreement in this respect. While,
for instance, Kastovsky (1982: 88) regards the features “MALE :
FEMALE” as minimal, Lyons (1977: 330) says that “most of the sense-
components that have been postulated by linguists (e.g. MALE, ALIVE, FOR

____________________________
see Kastovsky (1982: 80–150) and Leech (21981: 89–122). See also Palmer
(21981: 108–117) and Lyons (1977: 317–335).
373
Compare also Kastovsky’s (1982: 80) discussion of this parallel. Cf. also
Hjelmslev (1943/1974: 69–71). Compare also Coseriu’s (1973: 58–67) com-
parison of the analysis of lexis and phonology. For this see also Herbst,
Heath, and Dederding (1980: 82–84).
374
Cf. Lyons (1977: 331): “The most extreme form of the thesis of universalism
would combine at least the following three distinguishable sub-theses: (i) that
there is a fixed set of semantic components which are universal in that they
are lexicalized in all languages; (ii) that the formal principles by which these
sense-components are combined to yield as their products the meanings of
lexemes are universal (and presumably innate); and (iii) that the sense of all
lexemes of all languages is decomposable, without residue, into variable
combinations of (homogeneous) sense-components.” For a discussion of the
idea of universalism see Palmer (21981: 114–117).
250 Meaning

SITTING UPON) are not atomic in this sense”.375 However, Kastovsky (1982:
89) also argues that atomicity is not a necessary condition for a semantic
component when he points out that words such as grocer and haberdasher
are to be distinguished with respect to the dimension SOLD OBJECT in
terms of the meanings of the lexeme food or the phrase small articles of
dress. Using food and small articles of dress as semantic components, how-
ever, raises the problem of the recurrent character of such components.
There are two ways of approaching this problem: one is to give up the
claim that all components to be used in a semantic analysis should be ex-
pected to be recurrent and atomic.376 The other way out of the problem is to
give up the claim that componential analysis would provide a suitable basis
for the description of word meanings as a whole. It is indeed difficult to
imagine a component which is atomic and recurrent that could distinguish
daffodil from tulip or dandelion. Lyons (1977: 334) points at a different
problem:377
For example, if the meaning of the lexemes ‘man’, ‘woman’, ‘adult’, ‘girl’,
‘boy’ and ‘child’ are analysed in terms of the sense-components HUMAN,
ADULT and FEMALE, we can readily explain the fact that phrases like
‘adult child’ or ‘male girl’ are semantically anomalous. In doing so, we
must assume (and it is more often assumed than stated explicitly in treat-
ments of componential analysis) that “male” (i.e. the sense of the English
lexeme ‘male’) contains and is exhausted by the sense-component –
FEMALE, that “adult” contains and is exhausted by ADULT, and so on.
On this assumption, however, ‘male child’ should be synonymous with
‘boy’. But it is not.
Taking up this argument, Leech (21981: 121) comes to the conclusion that
“… componential analysis is not the whole story; but that it is part of the

375
Lyons (1977: 330) refers to the ideas of the German philosopher Leibniz in
this context.
376
Note the distinction between so-called markers and distinguishers made in an
early model of componential analysis, developed by Katz and Fodor (1963)
in the context of transformational grammar: in the analysis of one meaning of
bachelor, they identify HUMAN and MALE as markers and as a distin-
guisher ‘who has never married’. For a critical account of this approach see
Kastovsky (1982: 82 and 251).
377
For the problem of semantic notation see, for instance, Leech’s (²1981: 94–
97) concept of “signese”.
 16 Ways of describing meaning 251

story, and a significant part, need not be doubted”.378 Indeed, the insight
that componential analysis alone is probably not an appropriate means of
describing the meanings of lexical units should not be taken as an argument
for rejecting the approach altogether. For instance, componential analysis is
ideally suited to describe semantic relations as the ones identified in 15.4.2:
− The relationship between a hyperonym and its hyponyms can be stated
in componential terms by saying that the hyperonym lacks the semantic
component which distinguishes its hyponyms.
− Synonymy can be described as a total identity of components and their
+ or – values.
− Antonymy, complementarity and converseness can be described as an
identity of all components with the value of one component being dif-
ferent in a pair.
Furthermore, componential analysis serves perfectly well to explain certain
syntagmatic relationships, above all the oddity of a sentence such as
(1a) Wood swims.
– ANIMATE + ANIMATE
on the basis of the incompatibility of the features – ANIMATE (of wood)
and + ANIMATE (required of a subject of the lexical unit swim)379 as op-
posed to
(1b)BNC Wood floats.
– ANIMATE – ANIMATE
It should just be mentioned in passing that the relationship of incompatibil-
ity played an important role in the discussion of the grammaticalness of
sentences in Chomsky’s (1965: 149) Aspects model of generative transfor-
mational syntax, where he described “sentences that break selectional

378
See Leech (21981: 121): “There have emerged three different levels at which
word-meaning may be analysed. Firstly, the word-sense as an entirety may be
seen as a conceptual unit in its own right …; secondly, this unit may be sub-
divided into components or features, by CA; and thirdly, both word-senses
and features, representing prototypic categories, can be broken down into
fuzzy sets of attributes. If this is more complex than the view of word-
meaning with which we started, it is of a similar order of complexity to that
of systems studied in natural science.”
379
Cf. Coseriu’s (1973: 78) concept of Klasseme and the role of such features in
stating selectional restrictions in transformational grammar.
252 Meaning

rules” such as colourless green ideas sleep furiously as “deviant”. Semantic

components were introduced in the form of selectional features to prevent
the generation of such sentences.380

16.2 The structure of vocabulary

While the various approaches that can be subsumed under the heading of
structural semantics have succeeded in revealing and formalizing certain
relations between the lexical units of a language, the overall question as to
what extent the vocabulary of a language can or should be regarded as be-
ing organized in this way is still open.381 There is evidence to suggest that
the relationship between the various lexical units of a language is much
more complex than is suggested by accounts such as the one given above.
For instance, on the basis of the phrase old man one would be inclined to
take young as an antonym of the lexical unit old. However, new would be
the appropriate antonym in cases such as an old manuscript or an old pro-
gram. It is certainly debatable whether this is a sufficient reason to argue
that the old in old man represents a different meaning of old than the old in
old manuscript. Similarly, sell can be seen as the opposite of buy (in terms
of converseness; > 15.4.2.3), but, in other contexts, steal could also be re-
garded as expressing the opposite of buy. This suggests that statements
about oppositeness, like statements about sameness of meaning, are to a
large extent dependent on the context in which a lexical unit occurs.
The same point can be made about linguistic hierarchies. Thus fly and
walk were established as hyponyms of go on the basis of sentences such as
(1)BNC We flew to Kirkwall with British Airways, which flies to Orkney and
Shetland regularly, from Glasgow and Edinburgh via Inverness.
but this leaves open the question of the status of a verb such as travel in the
hierarchy. Lyons (1977: 295, 297–298) also rejects a model of a hierarchi-
cally organized vocabulary of the following form:

380
It may just be noted that the combination green ideas is no longer unusual: …
they no more held a patent on green ideas than feminists did on equal rights
for women.BNC A great many green ideas were produced by Labour that
summer.BNC
381
Compare Lipka’s (32002: 148) description of the lexicon “as the structured
word-store of a language”.
 16 Ways of describing meaning 253

… If we take the most common adjectives in English we will see that there
are no superordinate adjectives at all of which particular subsets are hypo-
nyms. There are no lexemes of which adjectives, denoting differences of
colour, are all hyponyms. We do not say Was it red or coloured in some
other way?; but rather Was it red or (of) some colour? Similarly for sub-
classes of adjectives denoting shape, texture, taste, sound, age, size, state of
mind, etc.
What is perhaps particularly problematic is a view of vocabulary as ex-
pressed by Coseriu (1973: 53) in his characterization of word field:382
Word fields contain … units that divide up amongst themselves a concep-
tual area on the basis of immediate oppositions. An immediate opposition
exists if no third term can be inserted between two terms.
There are two rather interesting claims involved in this statement, namely
(i) that there is a clear delimitation of the meaning of a lexical unit in the
sense that one can identify (or at least assume) clear boundaries between
the meanings of two lexical items and (ii) that all possible meanings or
concepts (within a certain field) are actually expressed by lexical units.
Both claims are more than doubtful. The first has come under attack in
prototype theory (> 16.4) and the second can be shown to be wrong by the
existence of lexical gaps such as the one concerning a neutral colour adjec-
tive outlined above. Furthermore, the contextual view of meaning created
in multi-word expressions as propagated by John Sinclair (> 10.5) contra-
dicts such a structuralist view of vocabulary.

16.3 Vocabulary and conceptualization

The questions addressed in the previous section are closely related to the
way the conceptualization process is imagined. It was pointed out in 2.1.2
that de Saussure’s view of the arbitrariness of the linguistic sign can also be
interpreted as referring to the way that languages divide up or conceptual-
ize the world.383 In other words, the world (or at least parts of it) is imag-

382
Original: “Wortfelder enthalten … Einheiten, die eine Bedeutungszone auf
Grund unmittelbarer Oppositionen unter sich aufteilen. Eine unmittelbare
Opposition besteht dann, wenn zwischen zwei Termen kein dritter mehr ein-
geschoben werden kann.”
383
For a more recent discussion of the notion of arbitrariness see Leech (21981:
26): “This leads on to the question of the partial ‘arbitrariness’ of the catego-
254 Meaning

ined as a (relatively) unstructured continuum, which is broken up by a lan-

guage. This is apparent from the writings of the Danish linguist Hjelmslev
(1943/1961: 52),384 one of the founders of structuralism, who says:
A paradigm in one language and a corresponding paradigm in another lan-
guage can be said to cover one and the same zone of purport, which, ab-
stracted from those languages, is an unanalyzed, amorphous continuum, on
which boundaries are laid by the formative action of the languages.
In this context it was argued by Hjelmslev (1943/1961: 53) – mistakenly, as
will be shown in 16.4 – that the colour spectrum is a good example of such
a continuum. Using the example of the Welsh colour terms gwyrdd, glas
and llywd, he pointed out that Welsh does not draw a boundary in the same
place as Danish does between grøn and blaa. The English translation of
Hjelmslev's text represents this in the following way: 385

The American structuralist Gleason (1961: 4) also points out differences in

the vocabulary denoting colours, referring to English, Shona (spoken in
Zimbabwe) and Bassa (spoken in Liberia):386

____________________________
ries language provides for us. By ‘arbitrariness’, I mean, firstly, that concep-
tual boundaries often vary from language to language in a way that defies
principled explanation. A second kind of arbitrariness, presupposed by the
first, is the arbitrariness of language with respect to experienced reality: lan-
guages have a tendency to ‘impose structure upon the real world’ by treating
some distinctions as crucial, and ignoring others.”
384
Hjelmslev: /jel"msleu̯/
385
The translators of the German version point out that Hjelmslev's basis are
Danish colour terms such as grøn and blaa; see Hjelmslev (1943/1974: 57).
386
See Sampson (1980: 95).
 16 Ways of describing meaning 255

English purple blue green yel- orange red

low
Shona cipswuka citema cicena cipswuka
Bassa hui zĩza

Colour terms were also addressed in the work of the German scholar Trier
(1934), who is known for his work on word fields. His ideas are interpreted
by Lyons (1977: 253) in the following way:
Considered as a continuum, the substance of colour is (…) a conceptual
area (Sinnbezirk); it becomes a conceptual field (Sinnfeld) by virtue of its
structural organization, or articulation, by particular language systems.
While colours may be a particularly good example to demonstrate the struc-
turalist conviction that the vocabulary of a language gives structure to real-
ity, this view is by no means restricted to colour terminology.387
From such considerations it is only a small way towards raising the
question of the relationship between language and thought. If human beings
perceive the world through the conceptual system of their mother tongue
and if the conceptual systems imposed on reality by different languages
differ, then, one might argue, perception and thought are influenced by
language. While such a statement is relatively uncontroversial still today,
the extent to which this is the case has been the subject of much debate.
A particularly extreme form of the claim that human thought is shaped
by language is the so-called Sapir-Whorf-hypothesis, which is based on
ideas of the American anthroplogists and linguists Edward Sapir388 and
Benjamin Lee Whorf, which they developed in the study of the American
Indian languages. One of the most important claims made about one such
language, Hopi, is that it does not have tenses, but uses different verb forms
“according to whether the speaker is reporting a situation or expecting it,
and according to an event’s duration, intensity, or other characteristics”
(Trudgill 31995: 13).389 Such observations on Indian languages led Whorf

387
For a discussion of further differences between languages in the field of lexis
or with respect to tense systems see Hjelmslev (1943/1961: 53–54).
388
Sapir: /sə"pɪə/ /seɪ"pɪə/
389
For a detailed discussion of the Sapir-Whorf-hypothesis see Sampson (1980).
Sampson (1980: 86) describes the situation in Hopi as follows: “… the lan-
guage does not recognize time as a linear dimension which can be measured
256 Meaning

to claim that the world is perceived in a different way by speakers of differ-

ent languages.
Many objections can be raised against these ideas: for instance, that
there is no proof that such differences between languages result in different
perceptions of the world, in fact, there seems to be counterevidence that
this is not the case.390 Also, one has to bear in mind that it is apparently
possible to translate from one language into another and that new ideas
going beyond the concepts provided by a language at any given point in its
history can quite obviously be shaped and put into words, as is obvious
from the study of word formation. Furthermore, quite important differences
can also be observed between European languages (such as the absence of a
distinction between simple and progressive aspect in German). For these
reasons, any kind of strong form of the hypothesis is rejected by semanti-
cists and sociolinguists today.391 Still, the Sapir-Whorf-hypothesis has pro-
vided an important impetus for the study of the relationship of language
and thought.

16.4 Prototype theory

16.4.1 Colour terms

The analysis of colour terms has played a central role in semantics. As was
pointed out in 16.3, structuralist semanticists assume that the spectrum of
colours that the human eye can identify takes the form of a continuum that
is divided up in different ways by different languages. Colour terms thus
present one example of the arbitrariness of the linguistic sign in de Saus-
sure’s sense.
Approaching colour terms from a psychological point of view, however,
there is evidence to suggest that the colour spectrum is not perceived as a

____________________________
and divided into units like spatial dimensions, so that for instance Hopi never
borrows spatial terms to refer to temporal phenomena in the way so common
in European languages …, nor does Hopi permit phrases such as five days
since daytime is not a thing like an apple of which one can have one or sev-
eral. … And since there is no concept of time, there can be no concept of
speed …”.
390
Cf. Steinberg (1982: 109), who points out that “knowing word forms (spoken
or written) may aid memory”. See also 16.4.1.
391
Compare Trudgill (31995: 14–15) or Leech (21981: 26–27).
 16 Ways of describing meaning 257

continuum by human beings. In a famous study, Berlin and Kay (1969)

compared different languages as to their basic colour terms. Basic terms are
words such as red, yellow, orange or green, i.e. terms that cannot be sub-
sumed under any other term (as is the case with crimson), that are morpho-
logically simple (like blue, but not bluish or dunkelblau), frequent and not
restricted to technical language. The remarkable insight provided through
the experiments carried out by Berlin and Kay is that there are shades of a
particular colour which are better examples of that colour than others, in
other words, that some greens are greener than others. Experiments have
shown that speakers differ considerably when asked to indicate the bounda-
ries between different colour terms. This holds both within one language
and across different languages. There is a great deal of agreement, however,
when it comes to picking a particularly good example of a particular colour.
Taylor (32003: 9) summarizes this as follows:
Although the range of colours that are designated by red (or its equivalent
in other languages) might vary from person to person, there is a remarkable
unanimity on what constitutes a 'good red'.
The special status of focal colours is confirmed by a number of experi-
ments. For instance, people seem to be able to name samples of focal col-
ours more easily than samples of non-focal colours.392 The conclusion that
can be drawn from evidence such as this is that cognitively colour catego-
ries “have a centre and a periphery” (Taylor 32003: 14).
This view is different from the structuralist account of colour terms in
two important ways:
− Colour terms do not seem to present a very good example of the arbi-
trariness of linguistic classification. If the colour spectrum contains cer-
tain focal colours which are particularly prominent for human percep-

392
See especially Rosch Heider (1971, 1972) and also Taylor (32003: 10-11).
For the more far-reaching claims made by Berlin and Kay (1969) see also
Clark and Clark (1977: 524–527), Taylor (32003) or Leech (1981: 235–236),
who comes to the following conclusion: “But however much opinions may
vary, psycholinguistic evidence ... points strongly to the conclusion that the
relative uniformity of colour semantics in different languages has much to do
with the uniformity of the human apparatus of visual perception. Whatever
language a person speaks, he will tend to perceive certain focal colour stimuli
as more salient than others; and his language, too, will tend to discriminate
colours on the basis of these perceptionally salient areas.” See also Ungerer
and Schmid (22006: 7-14).
258 Meaning

tion and if it is for these focal colours that languages have basic colour
terms, then the colour spectrum is not divided up by language.
− If focal colours can be named and identified more quickly than non-
focal colours, then category membership cannot be seen in terms of a
yes/no-decision but is rather a matter of gradience.

16.4.2 Prototypes

The question is to what extent the prototypical nature of categories such as

colour terms can be applied to the description of word meanings. If some
greens are greener than others in the sense that some shades of green are
seen as better examples of the meaning of green than others, then perhaps
some types of bird can be regarded as birdier than others or some pieces of
furniture as better examples of the category furniture than others.393
And indeed, so-called goodness-of-example rankings suggest precisely
this. For instance, Rosch (1975: esp. 229) asked American college students
to judge elements of a list of household items as to their membership in the
category furniture on a scale ranging from 1 (very good example) to 7 (very
bad example; not an example at all). The result is a clear gradient: chair
(1.04), sofa (1.04), couch (1.10), table (1.10), ... desk (1.54), bed (1.58), ...
piano (3.64), ... cupboard (4.27), stereo (4.32) ... ashtray (6.35), fan (6.49)
or telephone (6.68).
In a similar way, ratings for different types of birds were established,
which Aitchison (32003: 56) presents in a kind of hierarchy in which the
robin is the most central member and the penguin and the ostrich the most
marginal members.

393
For a more detailed discussion of Rosch’s approach see Taylor (32003),
Aitchison (32003: 53–58).
 16 Ways of describing meaning 259

Birdiness rankings
(taken from Aitchison 32003: 56)

From a psychological point of view, these rankings are certainly relevant.

Rosch (1973) found that usually people (especially children) are quicker to
verify a sentence of the type A pear is a fruit if the test item was a good
example of the category.
To what extent these psychological findings can be applied to the de-
scription of word meaning may be a matter of debate.394 What is certainly
true is that the marginal status of certain members of a category is reflected
by their ability to occur in sentences containing so-called hedges such as
(2a) Strictly speaking, a penguin is a bird.

394
Compare the discussion of the relationship between goodness-of-example
ratings and degree of membership outlined by Croft and Cruse (2004: 79),
where it is pointed out that it could be argued “that an ostrich is a fully paid-
up member of the BIRD category” despite its low goodness-of-example-
rating.
260 Meaning

(3a)QE Strictly speaking, rhubarb is a vegetable.395

which central members of the category cannot
(2b) ?Strictly speaking, a seagull/a guillemot is a bird.
(3b)QE ?Strictly speaking, beans are vegetables.
Following a similar line of thought, Cruse (1986: 16–18) points out that a
feature such as ‘can bark’ is an expected trait of a category such as dog
since a sentence such as
(4a)QE It’s a dog, but it can bark.
is judged as odd, whereas a sentence such as
(4b)QE It’s a dog, but it can’t bark.
is perfectly normal. On the other hand, a sentence such as
(4c)QE It’s a dog, but it can’t sing.
is odd because ‘can sing’ is an unexpected trait of dog. The fact that both
(4d)QE It’s a dog, but it’s brown.
(4e)QE It’s a dog, but it isn’t brown.
are judged as odd shows that ‘is brown’ is a possible trait of dogs, which is
why a sentence such as
(4f)QE It’s a dog and it’s brown.
is normal.
The crucial difference between such observations and the structuralist
approach to meaning is that all components employed in a componential
analysis of a word such as bird would have equal status and be supposed to
apply to all members of the category. Thus ‘can fly’ cannot be used as a
semantic feature of bird in a structuralist approach. On the other hand, one
could argue that if speakers of a language see the ability to fly as a core
characteristic of the category bird, they will also consider it to be an impor-
tant feature for the description of the meaning of the word bird.
That this is indeed the case is shown by the definitions of words such as
bird, seagull or penguin in the Longman Dictionary of Contemporary Eng-
lish (LDOCE4):

395
Example taken from Taylor (32003: 80).
 16 Ways of describing meaning 261

bird ... 1 a creature with wings and feathers that can usually fly. Many birds
sing and build nests, and female birds lay eggs.: wild birds │The dawn
was filled with the sound of birds. │a flock of birds (= a group of birds
flying together) │a wooden bird cage
seagull ... a large common grey or white bird that lives near the sea
penguin ...a large black and white Antarctic sea bird, which cannot fly but uses its
wings for swimming ...
Whereas structural semanticists insist that the linguistic analysis of word
meaning and knowledge of extralinguistic categories should be kept apart,
from a cognitive point of view this does not seem to make sense. Cruse
(1986: 19) would argue “that any attempt to draw a line between the mean-
ing of a word and ‘encyclopaedic’ facts concerning the extra-linguistic
referents of the word would be quite arbitrary”.
As Croft and Cruse (2004: 81–82) point out, prototypes can be defined
in two different ways:396
− in terms of a list of attributes or features: “The centrality of an item in
the category depends on how many of the relevant set of features it pos-
sesses: the more it possesses, the better an example of the category it
will be. A feature is justified if, other things being equal, its presence
leads to a higher GOE [goodness-of-example] rating” (Croft and Cruse
2004: 81).
− on the basis of similarity: “A concept can be thought of as represented
by an ideal exemplar, and membership and centrality of other items can
be defined in terms of their similarity to the prototype” (Croft and Cruse
2004: 82).

16.4.3 Basic level categories

One central concept of prototype theory aims at an explanation for the fact
that words of a particular level of specificity tend to be used rather fre-
quently to refer to certain objects in the world rather than others. Thus it
would be odd to replace chair by a piece of furniture in sentences such as
(5a)BNC/NW Robyn sank down on a chair.

396
See Ungerer and Schmid (22006: 7-45) for attributes, family resemblances
and gestalts with respect to prototypes.
262 Meaning

(5b)BNC/NWLooking at Philip Swallow now, as he seats himself in a low, uphol-

stered chair facing her, Robyn has difficulty in recognising the jet-set
philander of Rupert Sutcliffe’s description.
Similarly, in most situations, it would seem more natural to say
(6a)BNC I’ll get the car out.
than
(6b) I’ll get the vehicle out.
(6c) I’ll get the VW out.
Interestingly, there are 33942 instances of car in the BNC but only 7201 of
vehicle and 150 of VW (and 200 of Volkswagen). Although any car can
also be referred to as a vehicle, this is quite obviously not done very often.
There seems to be a level of specificity which is convenient for many pur-
poses of communication. Thus, for instance,
(7a) I’ll go by car.
contrasts with
(7b) I’ll go by bus.
(7c) I’ll go by train.
or
(7d) I’ll go by bike.
Terms such as car, bus (6666 instances in BNC), train (8164), bike (2212),
lorry (1980) are referred to as basic level terms. This level is described by
Croft and Cruse (2004: 83–84) by the following characteristics:397
(i) It is the most inclusive level at which there are characteristic patterns of
behavioral interaction. …
(ii) The most inclusive level for which a clear visual image can be formed. …
(iii) The most inclusive level at which part-whole information is represented.
…
(iv) The level used for everyday neutral reference. …
(v) Individual items are more rapidly categorized as members of basic level
categories than as members of superordinate or subordinate categories.

397
Cf. Croft and Cruse (2004: 82–92). For a discussion of basic level terms see
also Taylor (32003: 48–55) and Ungerer and Schmid (22006: 64–76).
 16 Ways of describing meaning 263

16.4.4 Problems of prototype theory

It is obvious that prototype theory has an important contribution to make to

the analysis and understanding of meaning. However, there are also a num-
ber of problems.398 One is that many of the prototype effects that have been
observed experimentally depend very strongly on the cultural and situ-
ational environment. Whether or not a seagull is named as a very good
example of the category bird may depend on whether the informants live
close by the sea or not.399 Whether a word is a useful basic level term may
depend on the degree of expertise of a group of speakers. While train is a
useful basic level term in everyday discourse, it probably is not for people
designing railway time tables or responsible for the organization of the
railway network of a country. These, however, are not principal objections
to the approach but only indications of the fact that we are dealing with a
great number of subsystems.
What is more difficult to account for is the fact that prototype effects
can be observed even with clearly defined categories such as odd and even
numbers. Apparently, informants were happy to name a good example of a
category such as odd number, 3 getting a high degree of membership, 91 a
low one (Taylor 32003: 73; Armstrong, Gleitman and Gleitman 1999: 241).
This contradicts the common understanding of odd and even numbers as
categories that on the mere basis of their defining criteria do not allow for
any gradience of category membership.400
The most important deficit of prototype theory, as Croft and Cruse
(2004) point out, is that it does not offer a satisfactory explanation for cate-
gory boundaries. Although the notion of fuzzy category boundaries is justi-
fied on the grounds that different subjects locate category boundaries, such
as those between different colours, in different places, “it should be pointed
out that even a fuzzy boundary has a location” (Croft and Cruse 2004: 91).
Furthermore, one could argue that while it is tempting to analyse the
uses of a colour term such as white in combinations such as white coffee,

398
For some objections to prototype theory see Leech (1981: 84–86).
399
Compare Croft and Cruse (2004: 79) on priming effects: “The priming effect
correlates with the GOE score of the category member, that is, for Britons,
FRUIT will speed up the response to APPLE to a greater degree than the re-
sponse to, for instance, DATE.” See also Ungerer and Schmid (22006: 45–58,
esp. 53), who draw attention to the fact “that cultural models are not static but
changing” [GOE = goodness of example].
400
See Croft and Cruse (2004: 88) for a discussion.
264 Meaning

white skin or white wine as uses which are not central to the category, that
in itself does not explain why the combination white wine rather than yel-
low wine is used in the language.

Recommended further reading:

− Leech (21981), Palmer (21981), Cruse (1986)

− Ungerer and Schmid (22006)
− Croft and Cruse (2004)
 Utterances

17 Pragmatics

17.1 Word, sentence and utterance meaning

17.1.1 Sentence meaning

As was shown in Chapter 16, it is extremely difficult to answer a question

such as What does such and such a word mean? In fact, from a linguistic
point of view one might argue that this is not a sensible question to ask.
Nevertheless, questions about the meanings of words do get asked, espe-
cially perhaps in foreign language contexts. Questions of the type What is
the German word for “chair”? are by no means uncommon. In the case of
chair and Stuhl most people who know both languages would probably not
hesitate to answer the question. A question such as What does the word
“chair” mean? is probably slightly more problematic because one might
have to think about such attributes as to whether a chair can have arms or
not, whether it has to have four legs or not etc. Even if we accept that word
meanings are often perceived and explained in terms of prototypes, it is
remarkable that speakers feel able to answer such questions at all.
In the case of chair, the kind of meaning that speakers would probably
think of is that used in sentences such as
(1a)VDE He relaxed back into his chair, letting his pipe burn itself out in his
hand.
(1b)VDE A padded chair covered with green velvet was the only article of furni-
ture in the room.
and not
(1c)VDE You could apply for a personal chair.
(1d)VDE Hawking, inheritor of Sir Isaac Newton’s chair at Cambridge, origi-
nally played with the idea that time might run backwards if the universe
stopped expanding and collapsed in on itself.
266 Utterances

It is obvious that in semantics one would tend to treat this as a case of

polysemy as far as the meaning of the word chair is concerned. On the
other hand, it is unlikely that anyone would consider sentences such as (1a),
(1b), (1d) or even (1c) ambiguous. The reason for this must be that speakers
are able to select the appropriate meaning of a polysemous word on the
basis of information provided by the (linguistic or extralinguistic) context.
Although psycholinguists seem to disagree as to whether all meanings of a
word are activated when a particular sound chain or string of letters is per-
ceived or not, there can be no doubt that some kind of disambiguation
process must take place.
The guiding principle behind the process of working out the meaning of
a sentence seems to be to look for sense. This may sound trivial, but at the
same time it is probably a fact essential to the understanding of human
communication that hearers or readers assume that a particular utterance is
meaningful. It is for this reason that one would immediately rule out the
reading of ‘university chair’ as the appropriate meaning of chair in a sen-
tence such as
(1e)VDE Each chair will vary, so you’ll have to measure up and make a paper
pattern before cutting out the fabric.

17.1.2 The meaning of utterances

The principle of assuming utterances to be meaningful – which has been

termed the “no-nonsense-principle” or the “reality principle”401 – has wider
consequences than merely selecting the appropriate sense of a word in a
given context. It can equally be applied to the interpretation of an utterance
in a particular context of situation. For instance, a sentence such as
(2)VDE A visit to the site is truly memorable.
can be interpreted as a characterization of a particular site which stresses its
significance or impressiveness. In the appropriate context, however, it can
also be interpreted as a piece of advice to the effect that it is worthwhile
visiting this particular site. The descriptive or advisory character of the
utterance can also be seen as part of its meaning, but one which is not cap-
tured in a semantic description. In the same way, an utterance such as
(1c)VDE You could apply for a personal chair

401
See Clark and Clark (1977: 72–73).
 17 Pragmatics 267

can be meant as a piece of advice or as consolation or as information about

a possible course of action to take. This kind of meaning can only be ar-
rived at in the concrete situation of an utterance or by imagining situations
in which a sentence such as (2) or (1c) could be uttered.
In linguistics, a distinction is made between semantic meaning or, in the
terminology employed by Leech (1983: 17), sense and pragmatic meaning
or force.402 While semantics deals with sense, pragmatics, which Leech
(1983: 6) defines as “the study of meaning in relation to speech situations”,
deals with force. If, like Leech (1983: 14), one reserves “terms like sen-
tence or question for grammatical entities derived from the language system
… and the term utterance for instances of such entities, identified by their
use in a particular situation”, then one can say that semantics deals with the
meanings of words and sentences, whereas pragmatics deals with the mean-
ing of utterances.
What makes the study of pragmatics so interesting is that there can be
rather interesting discrepancies between semantic and pragmatic meaning.
Thus all four of the following sentences
(3a)VDE Can you clear the feathers off the mattress?
(3b)VDE Can you give me an example?
(3c)VDE Can you delay him in some way?
(3d)VDE Can you dismiss the management company if they fall down on the job?
can be classified as questions, more precisely even as ‘ability questions’.
From a pragmatic point of view, however, it is arguable whether they all
are interrogative in character. Whereas this is (relatively) clearly the case in
(3d) and (3c), utterances such as (3a) or Can you open the window? tend to
be requests rather than ability questions in that by uttering them a speaker
as a rule wants the hearer to do something rather than merely find out about
the hearer’s ability to do something.
There is of course considerable overlap between the various pragmatic
interpretations. (3b), for instance, might be said by an examiner and thus be
a request also testing ability. In fact, it seems that the interpretation of ut-
terances in terms of their force is also dependent on the assumption that an
utterance is meaningful. Since it would hardly make sense to ask an adult
person whether they are able to open a window (provided it is not some sort
of special window or a window that has caused other people trouble), one

402
Compare also Bublitz (22009: 19–24).
268 Utterances

would tend to interpret the utterance Can you open the window? as a re-
quest or a command. This is different, of course, when one is talking to a
small child who might not be able to open the window, and where, as a
consequence, asking an ability question does make sense.403
In the same way, of course, a sentence such as
(1c)VDE You could apply for a personal chair
could receive the ‘furniture’ interpretation in a context where the existence
of personal chairs in this sense might appear plausible. Thus it is probably
arguable whether the distinction between semantics and pragmatics can be
drawn as clearly as is sometimes suggested.

17.2 Principles

17.2.1 The co-operative principle and conversational implicature

In order to describe how human communication works, Paul Grice (1975:

45) established the so-called co-operative principle, which comprises
four maxims:404
QUANTITY: Make your contribution as informative as is required (for the
current purposes of the exchange). Do not make your contribution more in-
formative than is required.
QUALITY: Do not say what you believe to be false. Do not say that for
which you lack adequate evidence.
RELATION: Be relevant.
MANNER: Be perspicuous.
Avoid obscurity of expression.
Avoid ambiguity.
Be brief (avoid unnecessary prolixity).
Be orderly.

403
The question is, of course, whether we should interpret these facts to say that
the interpretation of an utterance such as (3a) relies on certain inferences to
be made by the hearer or whether one could see such questions as pre-
fabricated phraseological units. For a discussion of idiom theory and infer-
ence theory see Levinson (1983: 268–270). Compare also Leech (21981:
338–339). See also 17.3.4.2.
404
Cf. Levinson (1983: 97–166) and Bublitz (22009: 197–214).
 17 Pragmatics 269

It is rather unfortunate that Grice phrased the co-operative principle in

terms of imperatives because thus it can easily be misinterpreted as a list of
instructions or a style guide. Hence it is important to point out that the idea
behind the co-operative principle is purely descriptive in that it claims that
when people communicate they follow these principles. In this context it
must be understood that, as Leech (1983: 8) points out, Grice’s principles
and maxims differ from rules in a number of important ways:
(a) Principles/maxims apply variably to different contexts of language use.
(b) Principles/maxims apply in variable degrees, rather than in an all-or-
nothing way.
(c) Principles/maxims can conflict with one another.
(d) Principles/maxims can be contravened without abnegation of the kind of
activity which they control.
The co-operative principle helps to explain how hearers interpret utterances
beyond their explicitly stated content. For instance, a dialogue such as
(4) A: I have to be home by 5.
B: There is a bus at 4.20.
can be interpreted to mean that if A gets that bus A will be home by 5 –
otherwise B’s utterance would not be relevant. Similarly, in the case of
(5)QE A: When is Aunt Rose’s birthday?
B: It’s sometime in April.
B’s response at first sight appears like a violation of the quantity maxim.
Since there is no reason to assume that B is uncooperative in making that
response, the only explanation for the vagueness of the response is that B
does not wish to violate the quality maxim. Thus A can infer that B does
not know the exact date of Aunt Rose’s birthday. Leech (1983: 31) de-
scribes this process as follows:
The three stages of this inference are (i) rejection of the face-value interpre-
tation as inconsistent with the CP; (ii) search for a new interpretation con-
sistent with the CP; (iii) finding a new interpretation, and checking that it is
consistent with the CP. The new interpretation includes an implicature …
This implicature (in this case that B does not know the precise date of Aunt
Rose’s birthday) is necessary in order to make B’s utterance consistent with
the co-operative principle. Of course, one could also have come to the con-
clusion that B does not want to co-operate and deliberately does not tell A
270 Utterances

when Aunt Rose’s birthday is. Such conversational implicatures differ

from semantic inferences in that they are, as Levinson (1983: 104) says,
“inferences based on both the content of what has been said and some spe-
cific assumptions about the co-operative nature of ordinary verbal interac-
tion”.
The co-operative principle thus enables hearers to infer meaning ele-
ments in utterances that are not necessarily compatible with an analysis in
terms of semantics or logic. For instance, an utterance such as
(6a)QE Many of the delegates opposed the motion
will normally be interpreted to mean
(6b)QE Not all delegates opposed the motion
although it is not inconsistent with
(6c)QE All the delegates opposed the motion.
The reason for this is, as Leech (1983: 9) explains, “that if the speaker
knew that all of the delegates opposed the motion, the first Maxim of Quan-
tity … would have obliged him to be informative enough to say so.” (6b)
thus is a conversational implicature of (6a).405

17.2.2 Further principles

While Grice’s account of how people use and interpret utterances is re-
stricted to the co-operative principle, Leech argues that further principles
should be introduced such as the Irony principle, the Banter principle or the
Pollyanna principle to account for particular instances of language use. The
most important of these is the Politeness principle: ‘Minimize (other
things being equal) the expression of impolite beliefs’ (Leech 1983: 81).
Thus, for instance, in
(7)QE A: We’ll all miss Bill and Agatha, won’t we.
B: Well, we’ll all miss Bill.
or
(8)QE P: Someone’s eaten the icing off the cake.
C: It wasn’t ME.

405
See Levinson (1983: 102–103), Brown and Yule (1983: 31–33) and espe-
cially Bublitz (22009: 218–236).
 17 Pragmatics 271

impolite beliefs such as “We won’t miss Agatha” or “You have eaten the
icing off the cake” are not expressed explicitly. One of the maxims within
the politeness principle is the tact maxim – ‘Minimize the cost to h’ and
‘Maximize the benefit to h’,406 which enables Leech (1983: 107–109) to
establish different scales of politeness:

17.3 Speech acts

17.3.1 Performatives and constatives

Speech act theory, which is mainly associated with the British philosopher
J.L. Austin and the American philosopher J.R. Searle, focusses on describ-
ing the use of language as the performing of actions. The starting point of
speech act theory is the observation – made by Austin in a famous book

406
h = hearer.
272 Utterances

entitled How to do things with Words407 – of the special character of utter-

ances such as
(9)QE ‘I name this ship the Queen Elizabeth’ – as uttered when smashing the
bottle against the stem.
(10)QE ‘I do (sc. take this woman to be my lawful wedded wife)’ – as uttered in
the course of a marriage ceremony.
Austin (1962: 6) introduces the term performative for utterances of this
kind and distinguishes them from other utterances, which he calls consta-
tives, by saying that performatives do “not ‘describe’ or ‘report’ or constate
anything at all” (1962: 5). In particular, Austin points out that constative
utterances such as
(11)BNC The functionalist approach to the study of mind characterizes much of
the work currently being done in cognitive psychology, artificial intelli-
gence and linguistics.
(12)BNC I come to the beach every week.
can be assigned a truth value, i.e. they can be said to be true or false in a
particular situation, whereas performative utterances such as (9) or (10)
cannot be true or false in the same way. The reason for this is that by say-
ing something such as I name this ship the Queen Elizabeth, the person
does not describe an event or a fact, but carries out an action or performs an
integral part of an action, or, in Austin’s (1962: 5) words,
the uttering of the sentence is, or is part of, the doing of an action, which
again would not normally be described as saying something.
In fact, it is difficult to imagine the naming of a ship taking place without
such words being uttered – just as certain other ceremonial actions will
have to be performed as well. Similarly, the making of a promise is carried
out by uttering the words I promise and a wedding ceremony involves cer-
tain linguistic utterances (I will – I hereby pronounce you man and wife) as
well as non-linguistic actions (exchanging rings – kissing the bride etc.).
Performative utterances typically exhibit certain grammatical or lexical
features such as the use of present tense verbs, the use of the pronouns I
and we or the presence of hereby: thus

407
How to do things with Words is a series of lectures given in Harvard by Aus-
tin in 1955, which were published posthumously in 1962. See Austin (1962:
5).
 17 Pragmatics 273

(13)SP PICKERING: ... Never mind crying a little, Miss Doolittle: you are
doing very well; and the lessons won’t hurt. I promise you I wont let him
drag you round the room by your hair.
can certainly be interpreted as the making of a promise, whereas
(14)SP MRS PEARCE: ... I had to promise her not to burn it; but I had better
put it in the oven for a while.
(15)SP MRS HIGGINS: ... Henry! ... What are you doing here today? It is my
at-home day: you promised not to come.
(16)BNC So I promised I wouldn’t go out with him again.
are constative utterances that make a statement about a promise being
given. However, performative utterances do not have to meet these criteria,
as is shown by the following examples:408
(17)SP MRS HIGGINS: If you promise to behave yourself, Henry, I’ll ask her to
come down. If not, go home; for you have taken up quite enough of my
time.
HIGGINS: Oh, all right. Very well. Pick: you behave yourself. Let us
put on our best Sunday manners for the creature that we picked out of
the mud.
...
MRS HIGGINS: Remember your promise, Henry.
(18)BNC The Artist hereby authorizes the Publisher to pay all fees due to her
agents ...
(19)BNC Readers are reminded that the term “United Kingdom” (UK) includes
Northern Ireland, whereas “Great Britain” (GB) does not.
(20)BNC You are fired.
Furthermore, it is apparent also from Austin’s discussion of the distinction
between performatives and constatives that the view of utterances as ac-
tions need not be restricted to performatives: when one makes a constative
utterance, one also performs an action, namely the action of making a
statement. This is why Bublitz (22009: 85–86), who also points out that it is

408
See Austin (1962: 53–66). For an outline and a critical discussion of the fea-
tures of performatives see Bublitz (22009: 71–86). See also Levinson (1983:
229–236). Compare also Austin’s (1962: 32–33) distinction between explicit
(such as I order you to go) and implicit performatives (such as go).
274 Utterances

difficult to draw a clear dividing line between constatives and performa-

tives, still considers the distinction to be a milestone in the development of
pragmatics:409
We owe the following insights to Austin’s discussion of these two types of
utterance:
– that we use language to act, in particular to act in a goal-oriented way,
– that there are various types of linguistic actions,
– that there are verbs whose use counts as the performance of the action
they denote,
– that in some cases – apart from the use of a performative verb – further
(non-linguistic) conditions have to apply for the action to count as per-
formed
– that performative verbs are used predominantly for particular types of
institutionalized action (BAPTIZING, CONVICTING or DECLARING
OPEN)
– that speech acts are usually implicit, but (almost always) can be made
explicit,
– that speech acts can go wrong in different ways, but can only be true or
false under certain conditions.

409
Original text: “Wir verdanken Austins Überlegungen zu diesen beiden Äuße-
rungstypen die Erkenntnisse,
− dass wir mit Sprache handeln, und zwar zielgerichtet handeln,
− dass es mehrere Arten von sprachlichen Handlungen gibt,
− dass es Verben gibt, deren Gebrauch als Vollzug der von ihnen de-
notierten Handlung gilt,
− dass in manchen Fällen zusätzlich zu den performativen Verben andere
Bedingungen erfüllt sein müssen, damit die genannte Handlung auch voll-
zogen werden kann (sprechhandlungsunterstützende nicht-sprachliche
Bedingungen)
− dass performative Verben vor allem für den Vollzug bestimmter institu-
tionalisierter Handlungen explizit verwendet werden (TAUFEN,
VERURTEILEN oder ERÖFFNEN)
− dass Sprechhandlungen gewöhnlich implizit sind, aber (meist) explizit
gemacht werden können
− dass Sprechhandlungen in verschiedener Hinsicht misslingen, aber nur
unter bestimmten Bedingungen wahr oder falsch sein können.”
 17 Pragmatics 275

17.3.2 Locutionary, illocutionary and perlocutionary acts

Producing an utterance can be seen not only as one single act but as a num-
ber of different acts: in Austin’s (1962) theory, three types of act are distin-
guished which a speaker performs when making an utterance:410

a locutionary act by formulating and pronouncing a grammatical and
meaningful utterance,411

an illocutionary act by expressing the communicative intention of the
speaker, i.e. “asking or answering a question, giving some information
or an assurance or a warning, announcing a verdict or an intention, pro-
nouncing sentence, making an appointment or an appeal or a criticism,
making an identification or giving a description, and the numerous like”
(Austin 1962: 98-99),

a perlocutionary act by producing certain effects in the hearer.412
Leech (1983: 199) illustrates this with the following example:
LOCUTION:s says to h that X.
(X being certain words spoken with a certain sense and reference)
413
ILLOCUTION:In saying X, s ASSERTS that P.
PERLOCUTION:By saying X, s CONVINCES that P.

Searle (1969), in his model, splits up Austin’s locutionary act and empha-
sizes the act-character of referring (> 14.2.3) and predicating as being dis-
tinct from the mere uttering of words. Consequently, the following four
kinds of acts are identified by Searle (1969: 24–25):

utterance acts: “uttering words (morphemes, sentences)”

propositional acts: “referring and predicating”414

illocutionary acts: “stating, questioning, commanding, promising, etc.”
to which he adds

410
Cf. Bublitz (22009: 87–100).
411
Austin (1962: 95) describes the locutionary act in terms of “the phonetic act,
the phatic act, and the rhetic act”.
412
Austin (1962: 117) makes a distinction between perlocutionary object and
perlocutionary sequel, which Coulthard (21985: 19) describes as “intended re-
sult of the illocutionary act” and “an unintended or secondary result”.
413
P stands for proposition.
414
Cf. Bublitz (22009: 101–103).
276 Utterances

perlocutionary acts: “... the consequences or effects such [illocution-

ary] acts have on the actions, thoughts, or beliefs etc. of hearers.”
This can be demonstrated by looking at the following passage taken from
P.G. Wodehouse’s novel Blandings Castle:415
(21)AE ‘Beach,’ said Lord Emsworth.
‘M’lord?’
‘I’ve been swindled. This dashed thing doesn’t work.’
‘Your lordship cannot see clearly?’
‘I can’t see at all, dash it. It’s all black.’
The butler was an observant man.
‘Perhaps if I were to remove the cap at the extremity of the instrument,
m’lord, more satisfactory results might be obtained.’
‘Eh? Cap? Is there a cap? So there is. Take it off, Beach.’
‘Very good, m’lord.’
‘Ah!’ There was satisfaction in Lord Emsworth’s voice. He twiddled and
adjusted, and the satisfaction deepened. ‘Yes, that’s better. That’s capi-
tal.’...
If we take the sentence This dashed thing doesn’t work as an utterance of
(the fictional character of) Lord Emsworth, then we can say that Lord
Emsworth performs an utterance act by saying these words. The proposi-
tional act involves establishing reference to the binoculars he is trying to
use by the noun phrase this dashed thing and by making a predication about
it. The illocutionary act can probably be described as a statement, while the
perlocutionary act can be seen as prompting a question as a response.
It is important to realize that these acts should not be seen as unrelated
actions taking place simultaneously. Leech (1983: 201) points out that this
would be like describing an event in a football match as “The centre-
forward has kicked the ball; moreover, he has scored a goal; and further-
more, he has won the match.” Searle (1969: 24) also makes this perfectly
clear when he outlines the relation between utterance acts, propositional
acts and illocutionary acts:
I am not saying, of course, that these are separate things that speakers do, as
it happens, simultaneously, as one might smoke, read and scratch one’s
head simultaneously, but rather that in performing an illocutionary act one
characteristically performs propositional acts and utterance acts. Nor should
it be thought from this that utterance acts and propositional acts stand to il-

415
P.G. Wodehouse: Blandings Castle, Harmondsworth: Penguin (1954: 9).
 17 Pragmatics 277

locutionary acts in the way buying a ticket and getting on a train stand to
taking a railroad trip. They are not means to ends; rather, utterance acts
stand to propositional and illocutionary acts in the way in which, e.g., mak-
ing an “X” on a ballot paper stands to voting.

17.3.3 Felicity conditions

Although only certain utterances can be true or false, performatives of the

type
(9)QE I name this ship the Queen Elizabeth.
(22)BNC I declare you man and wife.
can go wrong or be non-felicitous416 – for instance, if the person making
the utterance is not authorized to do so, if there is no ship or no couple pre-
sent, etc., then the conventionalized speech acts of naming or of marriage
will not have been carried out. Searle (1969) suggests a distinction between
different types of felicity conditions for the performance of illocutionary
acts from which he derives four different types of rule, which can be exem-
plified rather well describing the speech act of promising as in (13) I prom-
ise you I wont let him drag you round the room by your hair:417
(1) propositional content rules:
− A promise can only be uttered “in the context of a sentence (or larger
stretch of discourse) T, the utterance of which predicates some future act
A of the speaker S” (Searle 1969: 63), in the case of (13) not allowing
Higgins to drag Eliza round the room.
(2) preparatory rules:
− A promise can only be uttered if the hearer would prefer the speaker’s
carrying out the act A to the speaker’s not carrying it out and if the
speaker believes that this is so.

416
For infelicities see Austin (1962: 14–24).
417
For the character of these “rules for the use of the illocutionary force indicat-
ing device” see Searle (1969: 54–56). For a more detailed discussion of the
felicity conditions of promising compare Searle (1969: 57–64) and Searle
(1971: 46–53). For a description of different types of illocutionary act in
these terms see Searle (1969: 66–67). Compare also Levinson (1983: 238–
240) and Bublitz (22009: 106–110).
278 Utterances

− Furthermore, a promise can only be uttered if it is not obvious to

speaker and hearer that the speaker will carry out the promised action
anyway.
(3) sincerity rule:418
− A promise can only be uttered if the speaker intends to carry out the
action.
(4) essential rule:
− The utterance of a promise “counts as the undertaking of an obligation
to do A” (Searle 1969: 63).
In contrast,
(23)SP PICKERING. Higgins: I’m interested. What about the ambassador’s
garden party? I’ll say youre the greatest teacher alive if you make that
good. I’ll bet you all the expenses of the experiment you cant do it. And
I’ll pay for the lessons.
is a case of betting: again, there is a sentence which predicates some future
act (teach Eliza proper English); the preparatory rules would be that the two
people betting take different views as to the outcome of this act; the sincer-
ity rule in this case requires that Pickering will pay for the experiment
should Higgins succeed, and the essential rule can be seen as the undertak-
ing of an obligation in very much the same way as with a promise.

17.3.4 Types of speech act

17.3.4.1 Searle’s taxonomy

Both Austin and Searle propose classifications of speech acts which focus
on the illocutionary aspect of the speech act. Austin’s (1962: 150) classifi-
cation has been criticized for not being a classification of speech acts but
basically one of verbs describing illocutionary acts.419 On the basis of a

418
For insincere promises see Searle (1969: 62).
419
Austin (1962: 150) distinguishes between (1) verdictives, (2) exercitives, (3)
commissives, (4) behabitives, and (5) expositives. For a critical view see
Bublitz (22009: 110–112). See also Leech (1983: 176) and Levinson (1983:
239). For Searle’s classification see Bublitz (22009: 116–120). Compare also
the approach taken by Leech (1983: 198–226).
 17 Pragmatics 279

number of different criteria, Searle (1979a) arrives at a distinction between

five “basic categories of illocutionary acts”:420

Searle (1979a: 12–17) paradigm cases

(Levinson 1983: 240)
assertives “commit the speaker (in varying asserting, concluding,
(representatives) degrees) to something’s being etc.
the case, to the truth of the
expressed proposition”
directives “attempts (of varying degrees requesting, question-
...) by the speaker to get the ing
hearer to do something”
commissives “commit the speaker (again in promising, threaten-
varying degrees) to some future ing, offering
course of action”
expressives “express the psychological state thanking, apologiz-
specified in the sincerity condi- ing, welcoming, con-
tion about a state of affairs gratulating
specified in the propositional
content”
declarations “the successful performance ... excommunicating,
brings about the correspon- declaring war, chris-
dence between the propositional tening, firing from
content and reality, successful employment
performance guarantees that the
propositional content corre-
sponds to the world”

In these terms, the following passage from Pygmalion could be analysed as

follows:
(15)SP MRS HIGGINS: [dismayed] Henry! <expressive> [Scolding him] What
are you doing here today? <directive> It is my at-home day <represen-
tative>: you promised not to come. <representative> [...]

420
Searle (1979a: 12) lists as criteria “illocutionary point, and its corollaries,
direction of fit and expressed sincerity conditions”. See also Bublitz (22009:
120–122) for a discussion of the criteria employed by Searle.
280 Utterances

HIGGINS: Oh bother! <expressive> [...]

MRS HIGGINS: Go home at once. <directive>
HIGGINS: [kissing her] I know, mother. <representative> I came on
purpose. <representative>
MRS HIGGINS: But you mustn’t. <representative> I’m serious, Henry.
<representative> You offend all my friends: <representative> they stop
coming whenever they meet you. <representative>

17.3.4.2 Direct and indirect speech acts

One of the problems of this kind of analysis is that although What are you
doing here? in (15) can be interpreted as the speech act of asking a question
(which falls into the category of directives) in the context of situation its
main communicative function can be described as an expression of surprise
and disapproval and thus be classified as belonging to the category of ex-
pressives. Searle (1979b: 30) accounts for this dual function of illocution-
ary acts by introducing the notion of an indirect speech act:421 in Searle’s
view, an utterance such as
(24)QE Can you reach the salt?
has two illocutionary forces – that of question and that of a request. The
fact that such utterances can serve as requests can be explained by two fac-
tors: firstly, the addressee’s ability to reach the salt can be seen as a pre-
paratory condition for the request to pass it. Secondly, in particular contexts
hearers will interpret such questions as only being relevant to the conversa-
tion if the speaker wishes them to carry out the action described according
to the co-operative principle established by Grice (1975) (> 17.2).
Searle’s view (1979b: 31) of indirect speech acts as “cases in which one
illocutionary act is performed indirectly by way of performing another” has
been criticized for a number of reasons. Thus while in the case of (24) one
could argue that a “literal” (question) interpretation and an indirect inferred
(request) interpretation are possible, Coulthard (21985: 27) points out that
this is by no means always the case since, for instance, teacher directives
such as the following “don’t seem to admit verbal responses easily”:
T: How many times have I told you to ... P: ? Seven, sir.
T: Who’s talking now? P: ? Me, sir

421
Compare Levinson (1983: 263–276). See also 17.1.
 17 Pragmatics 281

T: Can I hear someone whistling? P: ? Yes, sir

Such examples could be seen as an argument in favour of analysing such
cases simply as idiomatic constructions. However, Levinson (1983: 270–
271) argues that any such explanation “will need to be complemented by a
powerful pragmatic theory that will account for which interpretation will be
taken in which context” in cases in which more than one interpretation is
possible.
Nevertheless, the notion of indirect speech act is by no means unprob-
lematic. In particular, one has to bear in mind, as Levinson (1983: 274) and
Bublitz (22009: 141) show, that the claim of indirectness is based on the
assumption that there is a direct correlation between syntactic structures
such as declarative, interrogative and imperative and speech act functions
such as ‘question’ for interrogative. The “indirectness” then arises from the
fact that the hearer infers from the context that the speaker may not have
been meaning to ask a question at all so that the utterance has to be reinter-
preted as, for example, a request. If one does not maintain such a strong
link between syntactic form and illocutionary force, it is possible to attrib-
ute utterances illocutionary force independent of their form.422 As a conse-
quence, a distinction between direct and indirect speech acts as such is no
longer possible or necessary. Rather, indirectness can be seen as “a matter
of degree” (Leech 1983: 38) and be integrated into a general model of
pragmatics.423 Separating the classification of speech acts from syntactic
structure also makes it considerably easier to assign certain utterances to
particular types of speech act.

17.3.4.3 Problems of classification

Irrespective of the notion of indirect speech acts, one has to say that
Searle’s classification of speech acts is not entirely unproblematic. In par-
ticular, as Levinson (1983: 240) criticizes, there “is no reason ... to think
that it is definitive or exhaustive”.424 Indeed, one may wonder, for instance,
whether the fact that the class of directives comprises questions as well as

422
Levinson (1983: 274) describes this position as follows: “Illocutionary force
is then entirely pragmatic and moreover has no direct and simple correlation
with sentence-form or -meaning.”
423
Cf. Bublitz (22009: 145–147) and Coulthard (1977: 26–28) for a discussion
of indirectness.
424
For further criticism of Searle’s typology see Levinson (1983: 240).
282 Utterances

commands is particularly well-founded – other linguists such as Leech

(1983: 211) make a distinction between directive and rogative, for example.
Similarly, one might argue, as Coulthard (21985: 25) does, that “proposi-
tionally empty items” such as hello cannot be accommodated easily in
Searle’s typology. On the other hand – and here there is a parallel to Fill-
more’s (1968) model of case grammar – any classification of this kind will
have to provide for more discrete categories to be identified as subcatego-
ries of the major classes established. In fact, the level of discretion required
depends very much on the purposes of the analysis.
What is perhaps the most surprising aspect of speech act theory as de-
veloped by Austin and Searle is that although it is concerned with what
speakers “do” with language they do not investigate what “real speakers”
have done in the sense that they would base their theories on the analysis of
actual spoken or written text. Rather, speech act theory is developed on the
basis of invented examples in envisaged linguistic and extra-linguistic
situations of use. In the formulation of the theory, this is reflected in two
ways: firstly, the utterances discussed within the framework of speech act
theory mostly consist of sentences. One might argue that it might be appro-
priate to consider a sequence of several sentences as a single speech act
with one illocutionary force. Thus, in (15), the whole passage uttered by
Mrs Higgins could be seen as expressing ‘reproach’.
A second consequence of the fact that speech act theory has not been
driven by the analysis of real language use is that certain aspects of dis-
course structure such as the relation of different elements of a conversation
to each other etc. have not been taken into account.425 These issues are at
the centre of work on discourse analysis.426

425
Cf. Mey (1993: 181–185), Levinson (1983: 279).
426
For an account of research on discourse analysis see Levinson (1983: 284–
370) and Brown and Yule (1983) or Coulthard (21985).
18 Texts

18.1 The notion of text

18.1.1 Cohesion and coherence

Texts can be seen as a level of linguistic description above that of the sen-
tence.427 Such a level is needed for a number of reasons: one is that most
utterances used in discourse comprise more than one sentence; another that
it would be wrong to imagine texts as a mere series of unrelated sentences.
This can be shown by looking at a sentence such as
(1)TI From about 1950 until her death she also lived there.
Sentence (1) is highly unlikely to occur as the first sentence of a text. In
fact, it does not make much sense if read in isolation. This is because it
contains four linguistic elements which can only be interpreted with refer-
ence to the sentence preceding it in the text from which it is taken – namely
her, she, also and there.
(2)TI Barbara Hepworth worked in Trewyn Studios from 1949, ten
years after she first arrived in West Cornwall.
Following (2), it is clear that she and her refer to Barbara Hepworth and
there refers to Trewyn Studios since these are explicitly mentioned in the
first sentence of the text. Similarly, also is used to express the contrast be-
tween working in Trewyn Studios and living in Trewyn Studios. These are
examples of items that depend for their interpretation on elements of the
surrounding text. Such items can thus be seen as establishing links between
elements of a text or as creating cohesion within the text.
The term cohesion can be applied to sentence-internal and inter-
sentential relations: there does not seem to be any great difference in the
function of but in (32) and (33) apart from the fact that in the first case it

427
This does not necessarily contradict the view expressed by the famous Ameri-
can linguist Leonard Bloomfield (1933/1935: 170), who said: “It is evident
that the sentences in any utterance are marked off by the mere fact that each
sentence is an independent linguistic form, not included by virtue of any
grammatical construction in any larger linguistic form.”
284 Utterances

links two sentences and in the second it links two clauses belonging to the
same sentence.428
1
(3)TI When Barbara Hepworth first arrived at Trewyn studios she was
still largely preoccupied with the stone and wood carving which
was central to her work. 2But during the 1950s she increasingly
made sculpture in bronze as well. ... 3Some of the bronzes now in
the garden may be familiar to visitors from casts situated else-
where, but here they are seen in the environment in which they
were created.
It should be noted, however, that some linguists, in particular Halliday and
Hasan in their influential book Cohesion in English (1976), tend to focus
on relations across sentence boundaries because these have been studied
less widely than sentence internal structures. 429
Cohesion can arise from structural relations or parallels between differ-
ent sentences in a text but also from the relations between the lexical items
occurring in the text. The latter can be illustrated by text (4), in which the
repeated use of the same words (Cornwall in sentences 41 and 43, art in 41,
42, 43 and 45 etc.) or of morphologically related words (such as art and art-
ist) contributes to cohesion:
1
(4)TI Any account of modern art in Cornwall must begin by acknowl-
edging a three thousand year legacy of human activity in the re-
gion now known as West Penwith. 2The extraordinary presence of
Bronze Age standing stones and Celtic carvings and sculpture, as
well as the heritage of indigenous craft traditions, have regularly
surfaced in art made in the region in the twentieth century.
3
Nevertheless, most histories of modern art in Cornwall begin by

428
The same is true of the function of she in (1) and (2), of course. The fact that,
especially in spoken language, sentence boundaries are by no means always
clear, provides a further argument for not restricting the use of the term cohe-
sion.
429
For a more detailed survey of this model see Halliday and Hasan (1976) and
Schubert (2008: 31–58). Although Halliday and Hasan restrict the use of the
term cohesion to relations across sentence boundaries, it is important to real-
ize that they only do this in order to focus on the textual aspect of cohesion.
Halliday and Hasan (1976: 6) explicitly point out that the “parts of a sentence
or a clause ‘cohere’ with each other” and thus also “display texture”. Note
that other linguists such as de Beaugrande and Dressler (1981: 50) apply the
term cohesion to both sentence-internal and inter-sentential relations.
 18 Texts 285

noting Turner’s visit to the county in 1811. 4Turner’s sketches at-

test to the natural beauty of the region and the romance of visit-
ing the furthest flung limb of Southern England. 5They inaugurate
the dialogue between artist and landscape that dominates the
subsequent history of art in the region.
What is even more important for (4) to be perceived as a text, however, is
that the different elements of the text can be related by the reader so that it
appears as a meaningful whole. This property of texts is generally known
as coherence. The notion of coherence thus refers not only to linguistic
relations between certain words or constructions in a text but also to
knowledge about real-world items and affairs: thus it requires extra-
linguistic knowledge to know that Cornwall is a county of England (and
thus to relate the county in (43) to Cornwall) just as it takes some geo-
graphical knowledge of the British Isles to relate the furthest flung limb of
Southern England to West Penwith (or Cornwall). To what extent a text is
perceived as making sense depends, to some extent at least, on the world
knowledge and the inferences made by the reader: readers not aware of the
fact that William Turner is a famous British painter will perhaps perceive
the text as less coherent than those who are, although they may infer this
from the phrase Turner’s sketches in (44). It is thus important to note that,
as de Beaugrande and Dressler (1981: 6) say, “[c]oherence is clearly not a
mere feature of texts, but rather the outcome of cognitive processes among
text users”.
In a way, cohesion and coherence constitute different perspectives from
which a text can be analysed. The analysis of cohesion focuses on the text-
internal links between linguistic units such as words and structures,
whereas the investigation of coherence concerns such aspects as to whether
a sequence of words and linguistic structures presenting certain facts,
events or relations between them make sense. In some cases, it is difficult
to decide whether what one can call the texture of a text arises from cohe-
sion or coherence: would one want to say that the repeated occurrence of
words such as art and words related to art is a purely linguistic phenome-
non that makes the text cohesive or is that the result of the way in which
certain ideas are developed in the text? This question also concerns issues
such as whether expressions such as three thousand year, Bronze Age,
twentieth century, 1811, history can be analysed as being related in terms of
the categories and relations identified in semantics and whether the occur-
rence of histories (43) and history (45) contributes to the cohesion of the
286 Utterances

text despite the fact that the word history occurs in two different senses
here.
As far as the relation between cohesion and coherence is concerned, one
should bear in mind that a sequence of utterances can be perceived as co-
herent without any obvious signs of cohesion:430
(5)QE A: There’s the doorbell.
B: I’m in the bath.
This example makes clear that whether an utterance appears as meaningful
(and coherent) or not may depend on the world-knowledge of the speakers,
the context of situation or the linguistic context. A sentence such as (6) can
be perceived as a coherent utterance in different situations:
(6)TI This is the first work by Barbara Hepworth to which she gave a
title referring directly to landscape.
If, for instance, sentence (6) is uttered by a guide showing visitors round a
museum, they will interpret this as referring to the object which they are
looking at and consider the utterance coherent in the context of the situa-
tion. In the book from which it is taken, it could be interpreted with respect
to the headline
1
(7)TI Dame Barbara Hepworth 1903–1975
2
Landscape Sculpture 1944, cast 1961 ...
3
This is the first work by Barbara Hepworth to which she gave a title
referring directly to landscape.
or/and by looking at a photograph of the sculpture printed next to the text.
In all of these cases the use of this would appear coherent but only when it
is part of the text presented as (7) there is also cohesion.431

430
Example quoted from Brown and Yule (1983: 196).
431
Note in this context the following view by Sinclair (1994: 16): “Text is often
described as a long string of sentences, and this encourages the practice of
drawing links from one bit of the text to another. I would like to suggest, as
an alternative, that the most important thing is what is happening in the cur-
rent sentence. ... The state of the discourse is identified with the sentence
which is currently being processed. ... The previous text is part of the imme-
diately previous experience of the reader or listener and is no different from
any other, non-linguistic experience.”
 18 Texts 287

18.1.2 Texts as utterances

Both cohesion and coherence are important aspects of the analysis of texts.
Since there are texts which are coherent without being cohesive and there
are also sequences of words that could be analysed as being cohesive but
not as being coherent432 (Brown and Yule 1983: 197) it makes sense to
recur only on coherence in a definition of the notion of text. Further criteria
have been suggested – for instance, by de Beaugrande and Dressler (1981)
– of which the fact that a text should have a communicative function ap-
pears the most central.433
It is for this reason that Brinker (31992: 17) and Schubert (2008: 26)
identify coherence and communicative function as central to the definition
of text.
A text can then be defined as a linguistic utterance – spoken or written –
consisting of one or more sentences (minimally consisting of one word)
that can be attributed the property of coherence and that has a communica-
tive function.434

18.2 Cohesive relations

18.2.1 Explicit linking expressions

Perhaps the most obvious way in which the cohesion of texts manifests
itself is the use of words or phrases that link two sentences or structure the

432
Cf. the example given by Enkvist (1978: 110–111): “I bought a Ford. A car
in which President Wilson rode down the Champs Elysées was black. Black
English has been widely discussed. The discussion between the presidents
ended last week. A week has seven days. …” Cf. Brown and Yule (1983:
197).
433
See also Esser (2009: 19–20).
434
Compare also Quirk, Greenbaum, Leech and Svartvik (CGEL 1985: 1423):
“... a text – unlike a sentence – is not a grammatical unit but rather a semantic
and even a pragmatic one. A text is a stretch of language which seems appro-
priately coherent in actual use. That is, the text ‘coheres’ in its real-world
context, semantically and pragmatically, and it is also internally or linguisti-
cally coherent. For this latter facet, the term ‘cohesive’ has been applied, re-
ferring to the actual forms of linguistic linkage.” See also the seven standards
of textuality posited by de Beaugrande and Dressler (1981) and Schubert’s
(2008: 22–23) critique of that model.
288 Utterances

text explicitly – a phenomenon for which Halliday and Hasan (1976) use
the term conjunction.
(8)PH I wish you could see the place today in its Mediterranean brilliance of
light and colour! Yesterday, though, we were wreathed in mist all day ...
(9)TI Barbara Hepworth worked in Trewyn Studios from 1949, ten years after
she first arrived in West Cornwall. From about 1950 until her death she
also lived there.
(10)TI The involvement of the Tate in this project reflects the fact that, whilst it
holds only a small group of major Newlyn School paintings, its holdings
of painting and sculpture from St Ives are rich and broad, capable of
providing changing displays of variety and quality at a new gallery. But
the idea of a gallery for St Ives also chimed well with the Tate’s de-
clared policy of presenting the national collections of British and Mod-
ern Art to diverse audiences away from London. That is why the Trus-
tees agreed to take responsibility for the gallery that Cornwall County
Council had committed to build.
While in the case of conjunction linguistic items are used that have an ex-
plicit linking function, in other cases of cohesion the link is less explicit but
the presence of other items in the text is still presupposed for the interpreta-
tion of particular expressions.435

18.2.2 Grammatical aspects of relating referents and meanings

Many cohesive effects have to do with the fact that the same referent (of a
real or imagined world) is being referred to more than once or that the same
sense (i.e. meaning of a word) is being expressed more than once. This can
be done in a number of ways, which have been described in great detail by
Halliday and Hasan (1976).
(1) A phrase can depend on another element in the text for the identification
of its referent. Halliday and Hasan (1976) use the term reference for this
type of cohesion and make a distinction between

personal reference – established by personal pronouns (including their
genitive forms often called possessive pronouns)

435
For asyndetic coordination see CGEL (1985: 13.1).
 18 Texts 289

demonstrative reference – established by the definite article the and

the determiner-pronouns this – these, that – those or words such as
there.
(11)PH Midsummer Reds: July 1982 is a painting unique in Heron’s work. Its
robust textures and rugged forms, with the exception of the great lem-
ony disc and a couple of smaller shapes, are created entirely by palette-
knife.
(12)PH The maximal sensation created by a vast field of a subtly differentiated
pure hue, the optical vibrations generated by colour juxtaposition at
linear boundaries, and the dynamic equilibrium of disparate formal
elements by which motive and shape oscillate between negative and
positive: these were the means to artistic purposes that only oil painting
could fulfil.
In the case of personal reference and demonstrative reference there is co-
reference between the pro-form and the item to which the pro-form refers.
Thus in (11) both the noun phrase Midsummer Reds: July 1982 and its refer
to the same painting by Patrick Heron. A slightly different situation is pre-
sented by (13), where the item this takes up the previous sentence (so that
the idea expressed by (131) can be regarded as the referent of this in
(132).436
1
(13)PH ‘The feeling of a sort of marriage of indoor and outdoor space, through
the aperture of the window frame, itself roughly rectilinear and parallel
to the picture surface, was really the main theme of all my paintings –
or nearly all – between 1945 and 1955.’ 2This, as we have seen, is not
entirely true of Heron’s production in that period ...
(2) A word can take up another word, part of a phrase or a larger unit of the
text in terms of the meaning expressed by it. Halliday and Hasan (1976) use
the term substitution for this phenomenon and distinguish between

436
For the categories of extended reference and textual reference see Halliday
and Hasan (1976: 52–53). Note that Halliday and Hasan (1976: 89) describe
reference as “a relation between meanings”. In the case of (13), there is no
co-reference since (131) as a sentence does not have a referent. Halliday and
Hasan (1976: 32–33) make a distinction between endophoric reference,
which can be anaphoric (i.e. to preceding text) or cataphoric (i.e. to following
text), and exophoric reference (i.e. to the context of situation). Only endo-
phoric reference is cohesive since exophoric reference is text-external. Com-
pare in this respect Sinclair (1994) (see note 431).
290 Utterances

nominal substitution where the pro-forms one, ones or same replace the
head of a noun phrase

verbal substitution in which do acts as a pro-form for a verb or a whole
predicate (and possibly also adjuncts)

clausal substitution by so or not.
(14)TI This work exists in two versions. The original elmwood carving of 1944,
is on permanent loan to the Barbara Hepworth Museum. The one illus-
trated is a bronze cast made in 1961.
(15)ZD ‘… They say you met Miss Coyne – the Treveals’ Clare.’...
‘I did’.
(16)ZD Flushed, giggling, they start to put themselves to rights.
But not Clare.
(3) Cohesion can also arise if the element taken up from surrounding text is
not expressed formally at all. This is the case, for example, if the partici-
pants of a verb (which are part of its semantic valency) are not expressed
by complements in the same sentence but when their referents can be re-
trieved from the context.
(17)ZD ‘Will you be going over to Newlyn?’ she asks, collecting teacups. Her
face is averted, her voice casual.
‘Why, no – you know I only went there last Thursday. I have no business
there today. Why do you ask?’
‘I only just wondered – I wasn’t sure.’ [unexpressed participant:
whether you were going over to Newlyn]
(18)ZD Go on, then – open it. I know you want to. [unexpressed participant:
open it]
(19)ZD ‘I don’t want to go home.’ she says.
‘No. Why should you?’ he responds. [unexpressed participant: want to
go home]
While the above cases can be accounted for in terms of contextually-
optional valency slots of the corresponding verbs which have no formal
expression (> 12.3.5), in the case of (20) and (21) certain elements of the
clause could be seen as adjuncts and complements of the verb in the pre-
ceding sentence:
(20)ZD ‘You been walking over to St Ives by the coast,’ he remarks to Law-
rence. It is not a question.
 18 Texts 291

‘Yesterday, you mean? No, not all the way. But I walked along the cliff-
path.’
‘That’s what I heard. ...’
(21)ZD ‘These are pencil drawings. And I do water-colours.’
‘What do you draw?’
‘Flowers, mostly.’
‘Do you? May I see?’
In Halliday and Hasan’s (1976) model such cases are referred to as ellip-
sis.437 However, they treat cases which are very similar in character such as
(22) – (24) as comparative reference, although they can also be explained
in terms of contextually-optional valency slots:
(22)TI ‘... I also don’t like personal publicity which leads to being known or
recognised by the general public or even art students as this interferes
with my private life & it is this which produces the work.’
Barbara Hepworth’s situation was very different at the time. [unex-
pressed participant: from what it is now]
(23)PH ... The exception was Rothko. ... [unexpressed participant: from the
group of artists discussed]
(24)TI I have always been intrigued by observing the way in which first the
colour on one side and then the colour on the other side of a common
but irregularly drawn frontier dividing them seems to come in front. As
your eye moves along such a frontier the spatial positions of the colour-
areas alternate ... [unexpressed argument: as the one mentioned]

18.2.3 Lexical aspects of cohesion and coherence

Halliday and Hasan (1976) subsume all the above types of cohesion under
grammatical cohesion, which they distinguish from lexical cohesion.438 It
has to be said that the same or at least very similar mechanisms are at work
in lexical cohesion and in grammatical cohesion and also that different

437
It may seem preferable to speak of unexpressed participants here since the
term ellipsis suggests a kind of underlying element which got deleted. The
point is, however, that the referent of the participant can be identified, not
that exact words should be retrieved.
438
For an account of lexical cohesion see Halliday and Hasan (1976: 274–292)
and Schubert (2008: 45–54).
292 Utterances

ways of making a text coherent interact or overlap. Thus the use of general
nouns such as works or event is very similar to the relation described as
reference with respect to items such as it or this in (11) – (13) and often
goes hand in hand with the use of the article the:439
(25)TI This painting is a turning point in Lanyon’s work. It was one of the
large works commissioned by the Arts Council for the Festival of Brit-
ain in 1951.
(26)PH In July 1965, at the ICA in London, Heron made a short speech in
which for the first time in public he criticized American aesthetic chau-
vinism in general and the role played by Clement Greenberg in the
critical promotion of American painting in particular. The event caused
something of a stir: ...
The use of general nouns can be seen as a special type of reiteration, a
term which also covers explicit repetition of a lexical item as in
1
(27)TI It may be hard news for new visitors to Tate Gallery St Ives to imagine
that it is built on the site of the town’s old gasworks. 2The site had, by
the 1980s, become derelict and dangerous. However, it presented obvi-
ous attractions as a site for a future gallery of modern art: it overlooks
Porthmeor Beach, but is also close to the old town around the harbour
and the modern development to the north-west, including the streets
where artists have lived for four generations.
Repetition of lexical items certainly contributes to the cohesiveness of a
text, but it could be argued that this also applies to groups of words and to
parallels in syntactic construction.
In any case, there seem to be different levels at which cohesion between
different items can be claimed. This can be illustrated by text (28):
1
(28)TI Since the mid-nineteenth century two ‘schools’ of art have grown up in
West Cornwall, at Newlyn and at St Ives. 2The existing public galleries
in Penzance and Newlyn provide for the display of Newlyn School paint-
ing but no similar arrangement has hitherto existed for St Ives. 3The
idea of a permanent home in the region for the distinctive modern art of
St Ives has long been cherished by many who live in or visit the area.
4
The Tate Gallery St Ives is the realisation of that idea.
5
The involvement of the Tate in this project reflects the fact that,
whilst it holds only a small group of major Newlyn School paintings, its

439
For the textual function of nouns see Götz-Votteler (2008: 135–141).
 18 Texts 293

holdings of painting and sculpture from St Ives are rich and broad, ca-
pable of providing changing displays of variety and quality at a new
gallery. 6But the idea of a gallery for St Ives also chimed well with the
Tate’s declared policy of presenting the national collections of British
and Modern Art to diverse audiences away from London. 7That is why
the Trustees agreed to take responsibility for the gallery that Cornwall
County Council had committed to build.
In this text we can identify (a) identical lexical forms representing the same
lexeme (Cornwall, Newlyn, art etc.), (b) different morphological forms of
the same lexeme (2galleries – 7gallery but also 1have, 2has, 7had and 4is –
5
are), (c) items related by word formation (5holds – 5holdings) and (d) lexi-
cal forms representing different senses of the same lexeme (2painting –
5
paintings – 5painting).
Halliday and Hasan (1976: 284) argue that cohesion can also be
“achieved through the associations of lexical items that regularly co-occur”
(a type of cohesion for which they rather confusingly use the term colloca-
tion).440 This type of cohesion can be based on semantic relations such as
synonymy, different types of oppositeness, and hyponymy or on the fact
that words belong to the same lexical field or to the same lexical set.441 The
fact that Halliday and Hasan (1976: 284) are rather vague in their descrip-
tion of this category442 and call it the “most problematical part of lexical
cohesion” may have to do with the fact that these relations between words
can be more appropriately described in terms of coherence than in terms of
purely linguistic relations between the words in question.
This applies in particular to cases such as the following, where the use
of the definite article is of interest.

440
Cf. Schubert (2008: 51–54) and also Herbst (1996) for this use of the term
collocation.
441
Schubert (2008: 52–53) makes use of Lipka’s (32002: 166–174) distinction
between lexical fields and lexical sets, where a lexical field is defined by its
members belonging to the same word class and sharing at least one semantic
feature, whereas lexical sets are not defined entirely on the basis of purely
linguistic criteria (> 16.1).
442
Cf. Halliday and Hasan (1976: 285): “The members of any such set stand in
some kind of semantic relation to one another, but for textual purposes it does
not much matter what this relation is.” Compare, however, Hasan’s (1984)
more convincing account in terms of identity and similarity chains.
294 Utterances

1
(29)TI Until he moved permanently to Cornwall, Heron and his family regu-
larly stayed in St Ives in a house in St Andrew Street, overlooking the
harbour. 2This painting is one of a series based on the main room of the
house and the view from its windows.
The use of the in the noun phrase the house in (292) can be explained by the
fact that it refers to a house in (291). There is no prior mention of main
room, however. It would be difficult to describe the relation of main room
and house as a semantic relation since not only houses but also other build-
ings such as inns, huts, etc. can have main rooms.443 Such relations can be
accounted for much more convincingly in terms of cognitive models of
knowledge representation which have made use of concepts such as frame,
scenario or script.444 A frame can be defined as “a schematic representation
of speakers’ knowledge of the situations or states of affair that underlie the
meanings of lexical items” (Fillmore 2007: 130).445
Within such models, one can then account for the use of the definite ar-
ticle in (292) by saying that the word house opens up a house-frame, which
accounts for the fact that it is part of the speaker’s knowledge about houses
that they can have a main room, (a kitchen, a bathroom etc.), which is why
main room does not have to be introduced as a “new” entity. The same
would hold for the use of the with view since it could be argued that speak-
ers know that houses have windows and that windows can have views.

443
Cf., however, the discussion of the semantic relation of meronymy by Croft
and Cruse (2004: 150–163).
444
For a discussion of frame, scenario, script and related concepts see Brown
and Yule (1983: 236–256), Bublitz (2009: 179–194), Ungerer and Schmid
(22006: 207–217) and Schubert (2008: 71–75).
445
Compare also Fillmore and Atkins (1992: 76–77): “Semantic theories
founded on the notion of cognitive frames or knowledge schemata ... ap-
proach the description of lexical meaning in a quite different way. In such
theories, a word’s meaning can be understood only with reference to a struc-
tured background of experience, beliefs, or practices, constituting a kind of
conceptual prerequisite for understanding the meaning. Speakers can be said
to know the meaning of the word only by first understanding the background
frames that motivate the concept that the word encodes. Within such an ap-
proach, words or word senses are not related to each other directly, word to
word, but only by way of their links to common background frames and indi-
cations of the manner in which their meanings highlight particular elements
of such frames.” See also Fillmore (1976).
 18 Texts 295

In the same way one could explain the use of the definite article in the
sea in (301) and (302) or of the sand in (304) on the grounds that a beach-
frame contains knowledge of the fact that a beach is on the sea (or a lake
etc.) and very often consists of a stretch of sand, whereas trees are not an
integral part of such a frame.
1
(30)PH Take, for example, a beach extending as far as the eye can reach, bor-
dered, on the one hand, by trees, and, on the other, by the sea. 2There is
enough green in the sea to relate it to the palms. 3There is enough of the
sky reflected in the water to create a resemblance, in some sense, be-
tween them. 4The sand is yellow between the green and the blue.
These examples show that certain aspects of the coherence of texts can be
accounted for more satisfactorily by referring to the world-knowledge of
speakers than to the language-internal relations that form the basis of cohe-
sion.

18.3 Thematic structure and information structure

18.3.1 Theme and rheme – given and new information

Since the way that a message is presented in a sentence can depend on the
surrounding text, the information structure of sentences is an important
aspect of the analysis of texts. The framework most widely used to describe
such phenomena was originally developed by scholars such as Mathesius
(1975) and Firbas, important representatives of the Prague School, and
applied to the description of English, for instance, by Halliday (1970b).
The first distinction to be made is that between theme and rheme. The
theme can be defined as the left-most constituent of a clause, the rheme
then being the remaining part of the sentence.
(2) Barbara Hepworth worked in Trewyn Studios from 1949, ten years
after she first arrived in West Cornwall.
theme rheme
Halliday (1994: 37) describes the theme as the “element which serves as the
point of departure of the message; it is that with which the clause is con-
cerned”. This distinction between theme and rheme bears a certain resem-
blance to that between subject (in the sense of an element ‘about which
something is said’) and predicate (‘what is said about the subject’) in tradi-
296 Utterances

tional grammar. However, the theme need not always be the subject of the
clause:446
(31)TI From the 1930s Barbara Hepworth made sculptures which con-
onwards sisted of a single upright form.
theme rheme
(32)SP Remember that you are a human being with a soul and the
divine gift of articulate speech: that your native
language is the language of Shakespear and
Milton and The Bible ...
theme rheme
While many linguists – for example, Esser (2009: 32) – describe the theme
as that part of the clause which is known to the hearer and the rheme as the
part that contains new information, Halliday (1970: 160–164) makes a dis-
tinction between theme and rheme on the one hand and ‘given’ and ‘new’
on the other. Given information can be described as information that the
speaker assumes to be familiar to the hearer, whereas new information is
information that the speaker assumes not to be known to the hearer. It is
obvious that theme and rheme and ‘given’ and ‘new’ represent distinctions
that do not necessarily coincide, as can be illustrated by the first two sen-
tences of the Foreword of a brochure sold at the Tate Gallery St Ives:
1
(33)TI The Tate Gallery is unique.
St Ives project
theme rheme
new new
2
It allows the visitor to see the works of art in the
area in which they were conceived and close to
the landscape and the sea which influenced
them.
theme rheme
given new
While the categories of given and new serve to illustrate some aspects of
the information structure quite well, one has to recognize that they are not
always easy to apply to actual textual analysis and that more refined cate-

446
For a detailed description of possible themes see Halliday (1994: 37–67). For
a definition of theme and rheme in this sense see Hoffmann (2000).
 18 Texts 297

gories recognizing different degrees of givenness or newness may be re-

quired.447 Firbas’s (1992: 8–11) notion of communicative dynamism tries to
capture this by saying that the information value of the various units of a
clause tends to increase with linear progression: this means that as a rule
the first unit has a very low information value and the last unit has the
highest information value in the sentence, while the other units of the
clause have an information value that lies somewhere in between.

18.3.2 End-focus and marked focus

Quirk, Greenbaum, Leech and Svartvik (CGEL 1985: 18.3/9) use the terms
theme and focus to describe the information structure of sentences (where
focus is defined in terms of stress). They establish the principle of end-
focus (which operates alongside the principle of end-weight) to say that
new and important information is “most neutrally and normally” placed at
the end of a clause.448
However, the information structure of sentences can deviate from this in
a number of ways. For instance, the intonation nucleus of the sentence can
be on elements other than the last one, thus creating marked focus. In a
neutral context,
(25)TI This painting is a turning point in Lanyon’s work.
will have the nucleus on Lanyon’s work but marked focus can be achieved
by putting the nucleus on this, painting or turning point. A similar effect of
giving special prominence to an item can be achieved by a number of syn-
tactic constructions such as a cleft sentence as in (34) or a pseudo-cleft
sentences as in (35):449

447
For a discussion of different degrees of ‘givenness’ see Brown and Yule
(1983: 179–188). See also Prince (1981: esp. 225–232). For an attempt to de-
scribe different types of thematic progression in texts see Daneš (1970).
Compare Götz-Votteler (2008: 68–80). For a critical account of the Prague
School concepts see Newmeyer (2001).
448
For a precise definition of the terms theme and focus see CGEL (1985: 18.1–
18.11). The principle of end-weight refers to the fact that syntactically heavy
(or long) constituents tend to occur at the ends of clauses: … the stripe paint-
ings did seem to reduce to the bare minimum the formula for the making of
pictures.PH
449
See CGEL (1985: 18.20–43) for grammatical devices and information struc-
ture.
298 Utterances

(34)PH It was in Arts in March 1956 that Heron wrote the celebrated article
that recorded the first collective showing in London of the painters ...
(35)PH What has changed is the underlying formal structure of the paintings.

18.4 Spoken and written texts

Texts can be classified from different perspectives and according to differ-

ent criteria. Werlich (1975), for instance, distinguishes between different
text types such as description, narration, exposition, argumentation and
instruction with regard to the way in which texts are structured.450 We will
concentrate here on the difference between spoken and written texts, i.e. a
difference caused by the medium of communication.
Spoken and written language differs in a number of respects, which can
be attributed to a number of factors.451 First of all, one has to consider the
fact that the spoken and the written medium offer a different range of pos-
sibilities for expressing meaning. In the spoken language, there is variety of
phonetic or prosodic means of expression which the written language lacks.
For instance, as pointed out above, stress can be used to highlight particu-
larly important elements of the message; similarly, intonation provides a
way of expressing meanings such as ‘statement’ or ‘question’, or of signal-
ling ‘surprise’, ‘disbelief’, ‘encouragement’ or ‘disapproval’ for instance.452
In the written medium, punctuation or different typefaces can be used to
express some of these meanings.453

450
Compare Werlich’s (1976: 39) definition of text type as “an idealized norm
of distinctive text structuring”. For the distinction between text type and
genre see Schubert (2008: 89). For different text types see Werlich (1975:
27–43) and Schubert (2008: 91–94).
451
For a detailed description of the features of the spoken and the written lan-
guage see e.g. Barnickel (1980: II.84–151) or Brown and Yule (1983: 14–
17). Compare also Schubert’s (2008: 133–161) discussion of conversation
analysis.
452
For a survey of such functions of different tones in English see Gut (2009:
121–126).
453
Note that the spoken language can also contain indications of meaning of a
different kind when one considers factors such as speed of delivery, voice
quality, which often, however, are not used intentionally. See Brown (1977:
125–155). For the relevance of intended and non-intended meaning elements
to film translation cf. Herbst (1994: 226–237).
 18 Texts 299

Further differences are related to the circumstances of the utterance and

are thus to a much greater extent a matter of degree. In spoken conversa-
tion, for example, the interlocutors are usually in the same room and can
see each other, which enables them to make use of visual symbols such as
gestures or to establish reference by deictic words such as this, over there
or here. This does not apply in the same way to telephone conversations, of
course.454
Similarly, in real (as opposed to fictional) conversation, speech tends to
be produced spontaneously, which means that speakers hesitate, produce
false starts or change constructions while they are speaking – and some-
times carry out repairs of such performance phenomena. Most situations in
which written texts are produced allow more time for planning and correc-
tion so that as a result written texts may appear more structured.455
Corpus analyses, such as the ones that form the basis of the Longman
Grammar of Spoken and Written English (1999) by Biber et al. (LGSWE),
provide evidence for the linguistic manifestation of differences between the
written and the spoken language.456 Thus it was found, for instance, that

tags are typical of conversation or written representations of speech
(LGSWE 3.4.4),

noun phrases with a modifier are rare in conversation, whereas almost
60% of all NPs in academic prose contain a modifier (LGSWE 8.1.1) or
that

passives tend to be more frequent in written texts, especially academic
prose, than in conversation (LGSWE 11.3).

454
For a discussion of deixis see Bublitz (2009: 237–258).
455
Cf. Crystal and Davy (1975: 86): “In the world of written English, discourse
has a regular, predictable pattern of connectivity. Sentences are regularly
identifiable ... The general impression is one of premeditation and conscious
organization. Errors of expression and changes of mind, if they occur, can be
carefully erased, and eliminated from a final draft. If a word or phrase does
not come to mind, the writer may pause until he finds it, or choose some al-
ternative. The page you are reading now is errorless: it does not show the
various stages of revision from manuscript to printer’s proof which gave it
the form it now has.” See also Brown and Yule (1983: 15), who point out that
spoken language contains “many incomplete sentences, often simply se-
quences of phrases” and little subordination.
456
LGSWE compares subcorpora of the four registers conversation, fiction,
newspaper language and academic prose.
300 Utterances

It is important to note that such features are not necessarily due to the spo-
ken or the written medium but the circumstances in which one tends to use
spoken or written language. Thus there is a wide range of texts in between
the typical spoken text of spontaneous conversation and the typical written
text as it is to be found in academic writing, for example. Novels may con-
tain dialogues which show typical features of spontaneous conversation
such as question tags or adjacency pairs of the type hello – hello or ‘ques-
tion’ and ‘answer’. The dialogues of plays or films differ from natural con-
versation in a number of ways, for example, with respect to hesitation phe-
nomena, repairs, overlap or interruption. In fact, these kinds of dialogue
belong to the category of texts that are written to be spoken. News broad-
casts on the radio or TV also fall into this intermediate category. On the one
hand, these texts clearly represent written language: since there is time to
plan and construct the utterance, the text will usually be well-structured. On
the other hand, radio news broadcasts have to be designed in such a way
that they can be understood by listening because the hearer does not have
the chance to go back and re-read a complex sentence as in a newspaper
article, for instance. TV news often combines spoken text with written text
and as such combines the two media of linguistic expression. In particular,
it should be noticed that the use of the internet has resulted in new forms of
texts which, although written, also display many characteristics of the spo-
ken language.457
Thus – with the exception of differences directly caused by the medium
of expression – it is very difficult to find general criteria for distinguishing
between spoken and written texts. As a consequence, a classification of
texts must be based on a number of criteria such as458

whether a text is spoken to be heard (as in conversation), spoken to be
written (dictation), written to be spoken (news broadcast, play) or writ-
ten to be read (e-mail, novel),

whether there is direct interaction between the people communicating
(as in an oral conversation or an internet chatgroup) or not (as in the
case of books or films),

457
For a detailed discussion of the language of e-mails, virtual worlds, chat-
groups etc. see Crystal (2001: esp. 24–48). See also Schubert (2008: 117–
128).
458
Compare Söll and Hausmann (1985: 46) and Herbst (1994: 150–153). See
also Crystal (2001: 42–43).
 18 Texts 301

whether there is face-to-face contact between the participants (as in

many conversations or also university lectures) or not (as in a telephone
conversation, a chatgroup or, of course, a novel),

to what extent the language produced is spontaneous (as would typically
be the case in conversation or – to a lesser degree perhaps – chatgroups)
or pre-planned (as in news broadcasts or written texts).

Recommended further reading:

− Bublitz (22009)
− Schubert (2008)
Variation

19 Variation in language

19.1 Registers and dialects

As was pointed out in Chapter 1, a language must not be imagined as a

monolithic system but the use of language is characterized by a consider-
able amount of variation in the sense that there are different ways of ex-
pressing the same content or meaning. Some of this variation entails a
choice on the part of the speakers; other types of variation are subject to
external factors. Thus the form a particular utterance takes may depend on

whether the medium of expression is spoken or written (as shown in the
preceding chapter),

what is being discussed and whether specialized vocabulary or termi-
nology is being used in the discussion,

whether the situation of utterance is (or is perceived by the speaker as)
relatively formal or relatively informal,

the regional and social background of the speaker (and the extent to
which this is apparent from their use of language),

whether the speaker is a woman or a man.459
Varieties that are determined by factors such as medium, field of discourse
or attitude are often subsumed under the labels of register and style (the
latter only referring to attitudinal variation). Biber et al., in the Longman
Grammar of Spoken and Written English (1999: 15), characterize registers
as “varieties relating to different circumstances and purposes” and distin-
guish them from dialects as “varieties associated with different groups of

459
For differences between the language used by men and women see e.g.
Coates (1986). For a corpus-based study of parameters such as the use of
empty adjectives, colour terms, non-standard language, question tags by male
and female speakers of British and American English see Grimm (2008).
 19 Variation in language 303

speakers”. Halliday, McIntosh, and Strevens (1964: 87) speak of varieties

of a language “distinguished according to use” and “according to user”
respectively.460
Such a distinction is too rigorous, however, because there are no clear-
cut boundaries between the different varieties and because, up to a point,
the factors determining the choice of a variety may be seen as interrelated.
For instance, speakers may well be competent in more than one dialect, i.e.
be able to speak, say, the standard language and some form of a regional
dialect. For such speakers, the extent to which they make use of “regional
elements” in their language may well be influenced by situational factors
such as who they are talking to and how formal they perceive the situation
to be.461 It is for such reason that Hudson (1980: 51) rejects what he refers
to as a variety-based model and argues in favour of an “item-based model”,
in which a linguistic item (lexical item, form of pronunciation, grammatical
construction) “is associated with a social description which says who uses
it, and when”.462
Indeed, sociolinguistic experiments have shown considerable variation
with respect to particular features. In a famous experiment the American
sociolinguist Labov investigated whether people working in three different
New York department stores pronounced an [r] in words such as fourth
floor or not. Labov (1966: 63–87) found that the use of [r] correlates to
social status because more [r]-forms occurred in the more prestigious shops
and to careful speech because people used more [r]-forms when they re-
peated the forms because they thought the researcher had not understood
them. So it seems that the use of this particular variable is typical of a

460
For a survey see Barnickel (1980: esp. 18–21) or Esser (1993: 38–43).
461
Compare Hudson’s (1980: 56) remark about varieties: “… there are yet more
reasons for not taking the notion seriously as a part of sociolinguistic theory,
since so-called varieties may be hopelessly mixed up together even in the sa-
me stretch of speech”. For the overlap between dialects and registers see also
Hudson (1980: 51), for the concept of code-switching see Hudson (1980: 56–
58).
462
Note that the approach advocated by Hudson (1980: 232) shows strong paral-
lels to the model of the mental lexicon as envisaged by Bybee (1995) in the
context of morphology and other usage-based approaches (> 9.7 and 13.2)
since he emphasizes the role of storage: “… the evidence that we have sur-
veyed suggests that we do in fact remember a vast amount of information
concerned with particular lexical and other types of item, and that linguistic
items may be related individually to social context”.
304 Variation

group of speakers (in the sense of a social dialect) as well as of a particular

style of speech.463 Similarly, studies carried out by Trudgill (31995: 32, 35–
36) revealed that the use of a word-final [n! ] (as opposed to RP [N] in words
such as walking, “h-dropping” in words such as hat, the use of glottal stops
as realizations of /t/ in words such as better or the use of third person singu-
lar verbs without an {S}-suffix (She like him very much) correlate with the
social class of the speakers:464

percentage of percentage of percentage percentage

[n! ]-realiza- no realization of glottal of verbs
tions in walk- of <h> in stops in without 3rd
ing, running, hammer, hat, butter, bet person
class etc. etc. etc. singular
{S}
middle middle 31 6 41 0
lower middle 42 14 62 2
upper working 87 40 89 70
middle working 95 59 92 87
lower working 100 61 94 97

One interesting factor determining this variation is accommodation, i.e. the

phenomenon that speakers tend to adapt their own speech to that of their
interlocutors. For instance, Trudgill (1986: 7–8) found that as an inter-
viewer he tended to use more glottal stops in words such as better or bet
when he was talking to informants with a high score of glottal stop realiza-
tions.465 In any case, the figures presented above are a clear indication of
the fact “that it is not possible to talk legitimately of discrete social-class
accents – again there is a continuum, with most speakers using sometimes
one pronunciation, sometimes another” (Trudgill 31995: 36).

463
For a description of Labov’s experiments and further studies on the correla-
tion of social class and particular linguistic forms see Hudson (1980: 148–
157) and Holmes (1992: 144–161). Compare also Trudgill (31995: 31–38).
464
For a detailed account of social and stylistic variation see Trudgill (1974: esp.
55–63, 90–96 and 130–132).
465
This does not apply to the groups with the highest scores for glottal stops. For
details see Trudgill (1978: 7–8).
 19 Variation in language 305

19.2 Accent, dialect, standard and prestige

19.2.1 Standard English and its pronunciations

When we talk about regional and social varieties of language it is important

to consider that such varieties should not be taken to exist in a discrete and
clearly distinguishable form.466 First of all, it is not really possible to draw a
clear dividing line between one dialect and another because dialect bounda-
ries tend to be gradual in character. Thus, the notion of dialect is already a
generalization of some sort. Similarly, the distinction between social and
regional dialects is artificial up to a point because there is strong interde-
pendency between the two dimensions. Thus one can safely assume that the
language typically used by speakers at the lower end of the social scale
usually shows more regionally marked forms than that of so-called “edu-
cated speakers”. Trudgill (1975: 21) uses the following representation to
show the interrelatedness of the social and the regional dimension in Brit-
ain:

R.P. accent

Standard English dialect

Social variation

Low status accents and

dialects

Regional variation

466
Compare Chapter 1 for the global spread of English and regional varieties.
306 Variation

A special position is taken by the form of English usually referred to as

Standard English. According to Barnickel (1980: 28–30), a standard can be
defined as a variety467

that is used for communication between speakers of different regional
dialects and is intelligible to all speakers in a country

that is generally used in printing

that has received a certain form of codification in that it is the variety
most usually described in grammar books and dictionaries

that is usually taught to foreigners.
Although Quirk, Greenbaum, Leech, and Svartvik (CGEL 1985: 18) point
out that the “degree of acceptance of a single standard of English through-
out the world … is a truly remarkable phenomenon”, different national
standards of English such as British and American English can be distin-
guished.468 Trudgill and Hannah (52008) identify two ENL (English as a
native language) varieties that are most commonly used as models in EFL
teaching – British English and North American English. The latter they
describe as “English as it is written and spoken by educated speakers in the
United States of America and Canada” (52008: 5), whereas British English
is characterized in very broad terms by Trudgill and Hannah (52008: 4):
As far as grammar and vocabulary are concerned, this generally means
Standard English as it is normally written and spoken by educated speakers
in England and, with certain differences, in Wales, Scotland, Northern Ire-
land, The Republic of Ireland, Australia, New Zealand and South Africa.
While Standard English allows for a certain amount of regional variation at
the levels of vocabulary, grammar or pronunciation, within England and
Wales there is one pronunciation of it, called Received Pronunciation
(RP), which shows no regional variation. RP is a highly prestigious accent
which is defined entirely in social terms. Historically, it is associated with
public school education and for a very long time RP was extremely impor-
tant as a social marker in British society, as documented, for example, in

467
Compare the criteria given by Haugen (1966/1972: 110): selection, codifica-
tion, elaboration of function, acceptance by the community. For linguistic cri-
teria of Standard English see Trudgill and Hannah (52008: 2–3): See also
Trudgill (1975: 18).
468
CGEL (1985: 1.25), however, describes a variety such as Irish English as a
national standard. For national varieties of English see also Barnickel (1980:
44–139) and Hansen, Carls, and Lucko (1996).
 19 Variation in language 307

George Bernard Shaw’s play Pygmalion. Since today this type of accent is
more appropriately characterized in terms of its use on radio and television
by BBC national newsreaders, for example, the term RP is sometimes re-
placed with BBC English.469 While Trudgill and Hannah (52008: 15) esti-
mate that only 3–5 per cent of the population of England are RP speakers,
Wells (1982: 10) points out that in the United States the majority of speak-
ers use an accent that “reveals little or nothing of their geographical ori-
gins”. This accent is referred to as General American or Network Eng-
lish.470

19.2.2 Quality judgements

It is important to realize that the standard language is, as Kortmann (2005:

256) or Trudgill (31995: 8) emphasize, not in any way superior to other
regional or social dialects of the language with respect to any structural
aspects or intrinsic linguistic properties. The standard language rather has a
special status because of its functions in a speech community. It is a variety
used for certain purposes of communication and also a variety used by so-
called “educated speakers”.471
The use of the standard raises a number of highly relevant educational
issues. For instance, one could argue that it is important to teach the stan-
dard at school because it is the variety of the language which is most ap-
propriately used in certain communicative situations in which it would be a
disadvantage not to be able to use the standard. Even Peter Trudgill (1975:

469
See, for instance, Gimson’s introduction to the fourteenth edition of the Eng-
lish Pronouncing Dictionary (EPD14, 1977: x), where he points out that the
term “Public School Pronunciation” used by Daniel Jones in the first editions
of the EPD and his description of that type of pronunciation as “that most
usually heard in everyday speech in the families of Southern English persons
whose menfolk have been educated at the great public boarding-schools” is
“hardly tenable today”. EPD15 (1997: v) uses the term BBC English. For
variability within RP today see Wells (1982: 279–301). See also Upton
(2008: 237–240).
470
For a more detailed account of General American see Wells (1982: esp. 467–
472).
471
Cf. CGEL (1985: 1.22). Kortmann (2005: 257) estimates that the standard
variety is used by a minority of 12% to 15% of the population.
308 Variation

80), whose views on these matters can be regarded as relatively radical,

says:
We accept … that the teaching of written standard English can be carried
out in schools, not because this dialect is ‘correct’ or even ‘appropriate’, but
because it may be socially and economically advantageous to children when
they leave – and because it is possible.
However, it is important that when the standard is taught at school to chil-
dren who do not normally use it, they learn to understand that this is a vari-
ety of the language that is useful for particular purposes and that the lan-
guage they normally use is not in any way inferior and is equally useful for
other purposes of communication.472
It is interesting to see that the use of a particular dialect or, especially
perhaps in Britain, the use of a particular accent can influence the way a
particular utterance or its speaker is perceived by interlocutors. One ex-
periment that can be quoted in this context was carried out by Elyan, Smith,
Giles, and Bourhis (1978: 128) in which it was tested how listeners react to
utterances spoken by two women once using RP and once using a Northern
English accent.
It was found that listeners considered the RP speakers to be higher in self-
esteem, clearer, more fluent, intelligent, self-confident, adventurous, inde-
pendent, feminine and less weak than the regional accented speakers. In ad-
dition, the former were more likely to have a job which was well-paid and
prestigious and an egalitarian relationship with their spouse in the home,
but less likely to have children than Northern accented speakers. At the
same time, regional speakers were perceived to be more sincere and like-
able, and less aggressive and egotistic than their RP counterparts.
Irrespective of whether the results of such experiments are totally reliable
in every detail, there is strong evidence to suggest that people connect cer-
tain qualities with particular varieties. It has been shown, however, that
such judgements do not result from particular intrinsic features of those
varieties. Thus, for example, not pronouncing word-initial h is considered
ugly in an English context, where it occurs in low-prestige varieties such as
the London accent of Cockney. At the same time it is a prominent feature
of the French language, which is generally regarded as elegant. Rather,
aesthetic judgements seem to reflect qualities associated with the speakers

472
Cf. Trudgill (1975: 65–83). For a discussion of the deficit hypothesis, re-
stricted and elaborated code cf. Hudson (1980: 214–230).
 19 Variation in language 309

of the varieties in question or areas where they are spoken.473 Within Eng-
land, there is a tendency for Received Pronunciation, the standard pronun-
ciation of England and Wales, to obtain high prestige scores, followed by
accents of pleasant rural areas such as Somerset, with the Birmingham ac-
cent at the absolute bottom.474 Edwards (1982: 25) concludes:
Overall, these British (and Irish) studies of accent evaluation show that
speech samples may evoke stereotyped reactions reflecting differential
views of social groups. Standard accents usually connote high status and
competence; regional accents may be seen to reflect greater integrity and at-
tractiveness.
The fact that different varieties apparently evoke certain stereotypes indi-
cates that different varieties are associated not only with particular user
groups but also with particular situations of use.

19.3 Levels of differences between regional and social varieties

Differences between regional and social varieties can be observed at all

levels of linguistic description:475
As far as pronunciation is concerned, it is interesting to see that some
accents of English distinguish different words by different sounds which
are pronounced the same in others: thus, RP and many other accents of
English have two different vowels in words such as put, book, push and
words such as putt (in golf), cut or much:

RP Northern English
put, book, push U
U
putt, cut, much V

473
See the discussion of the social connotations hypothesis by Trudgill and Giles
(1978).
474
In a study by Giles (1970: 218) ten accents of English were ranked as to their
aesthetic content in the following way: RP, Irish English, South Welsh,
Northern England, Somerset, North American, Cockney, Affected RP, Bir-
mingham. See also Coupland and Bishop (2007). For further such studies and
possible objections see also Herbst (1994: 91). For German see Hundt (1992).
475
For detailed descriptions of varieties of English world-wide see Trudgill and
Hannah (52008).
310 Variation

In Northern accents of English, however, both groups of words are pro-

nounced with the same vowel because they do not have a sound contrast
corresponding to that between RP /ä/ and /å/.
There are also cases where two varieties express the same sound con-
trast, i.e. they can be said to have the same phonemes (> 5.1.1) but use
them in different sets of words. Both RP and Northern English and General
American use different vowels in the words in lists (i) and (ii):476

RP Northern English
General American
(i) park, father, calm, hard vowel 1
(ii) ant, hat, latter vowel 2

In words belonging to set (iii), however, RP has the vowel of group (i),
whereas Northern English and General American have the vowel of group
(ii):

RP Northern English
General American
(iii) castle, aunt, banana,
vowel 1 vowel 2
example

These differences explain why “[i]n a northern accent, then, put and putt
are typically homophones, [pät], while gas and glass rhyme perfectly, [gas,
glas]” (Wells 1982: 349).
If we say that two accents make the same distinctions this does not mean
that the sounds expressing these distinctions have to be phonetically identi-
cal. Thus, although from a certain point of view it makes sense to say that
words such as castle or aunt have the same vowel phoneme in General
American and Northern English, the pronunciation of these words in the
two accents differs considerably, which can be represented like this:477

476
For a description of different accents of English with reference to standard
lexical sets see Wells (1982). See also Schneider (2008b).
477
For the vowel systems and the use of the symbols compare Wells (1982: 118,
120, 363–365), who gives /—–/ for Birmingham and /°–/ for Leeds. For /&/ in
RP and American English see 5.2.3 and 7.5.
 19 Variation in language 311

Northern General
RP
English American
(i) park, father, calm, hard —– —– —
(ii) ant, hat, latter, trap ë ° ë
(iii) castle, aunt, banana,
—– °– ë
example, bath

A further very important distinction to be made in the case of accents of

English is that between rhotic and non-rhotic accents. Rhotic accents have
an overt /r/-realization before consonants (park, farm) or in word-final po-
sition (far, core, here), whereas non-rhotic accents do not. Rhotic pronun-
ciation is typical of many American accents including General American
and Canadian English, as well as the accents of Ireland, Scotland or the
South West of England, whereas Received Pronunciation and most other
accents of England and Wales, Australian English, New Zealand English or
South African English are non-rhotic.478
One important feature distinguishing RP and American English con-
cerns the pronunciation of /ju;/ and /u;/ in words such as knew, student or
enthusiasm.479 Furthermore, there can be differences in pronunciation re-
lated to stress (BrE /@d"v3;tIsm@nt/ vs. AmE /&dv@"taIzm@nt/ or simply
concerning individual words (BrE /¥Ÿ"≠—–¥Ÿä/ vs. AmE /¥Ÿ"≠•Í¥}Øä/).
Lexical differences are also very important when it comes to identify-
ing the differences between different varieties of a language. This becomes
clear when we ask how a particular object or thing of the real world is re-
ferred to in a particular variety. Thus, for instance, if speakers of American
English are talking about a ‘railway system operating below the streets of a
town or city’, they will probably be using the word subway, whereas
speakers of British English will use the word underground, when talking
about what in German is called a Fußgängerunterführung speakers of Brit-
ish English will use subway, whereas speakers of American English will
use the word underpass.

478
See Wells (1982: 75–76).
479
For Yod-dropping in British and American accents see Wells (1982: 147–
148). Generally, /u;/ pronunciations can be regarded as more typical of
American English, but note that the pronunciations given in EPD17 and
LPD3 do not always coincide.
312 Variation

From the point of view of the lexicon, one can thus say that

certain words only occur in particular varieties: thus words such as lorry
are used in British English but not in American English,

certain words have meanings that are exclusive to or typical of a particu-
lar variety: thus subway has the meaning of ‘underground railway’ in
American English and the meaning of “a path that goes under a road …”
(OALD6) in British English, British and American English share the
general meaning of waste disposal, but using it to refer to a waste dis-
posal unit is British English; the word tea is used for a meal only in
British English; etc.

some words occur in different varieties but with different meanings:
coffee shop in American English is “a restaurant that serves cheap
meals” and in British English “a place in a large shop or a hotel that
serves meals and non-alcoholic drinks” (LDOCE5).

Some words marked BrE or AmE in LDOCE5

BrE AmE
communication cord – cooling off attorney – blindside – boom box –
period – dustman – golden jubilee – freeway – garbage collector – heavy
motorway – post-code – school-leaver hitter – raise (noun) – restroom –
– tea brake – tea-cake soda fountain – zip code

Some word meanings marked BrE or AmE in LDOCE5

BrE AmE
circle ‘the upper floor of a theatre’ – blind ‘a small shelter from which you
copper ‘police officer’ – hide ‘place can watch birds or animals’ – figure
from which you can watch animals or ‘calculate an amount’ – outlet ‘place
birds’ – scoop ‘win a prize or award’– on a wall where you can connect
way out ‘a door or passage through electrical equipment to the supply of
which you leave a building’ electricity’

Such differences in the lexicon do not only apply to individual words but
also to multi-word units: the idioms not give a monkey’s or sail close to the
wind are typical of British English, for instance. Mittmann (2004) has dem-
onstrated that certain clusters of words tend to be used in one variety rather
than the other: for example, kind of or kinda was found significantly more
 19 Variation in language 313

often in a corpus of spoken American English than in a comparable corpus

of spoken British English, whereas sort of/sorta is more typical of British
spoken English. Similarly, combinations such as I suppose, I reckon, I
should think are overwhelmingly used by British speakers, whereas I guess
and I figure represent typical American usage. Similarly, more than 90 per
cent of the occurrences of a little bit were found in the British corpus but
more than 60 per cent of a little and a bit in the American corpus.480
It is important to realize that such differences concerning combinations
of words are often a matter of frequency rather than absolute features of
one variety. This is also true of some grammatical features: the Grammar
of Spoken and Written English by Biber et al. (1999: 462) states that in
conversation American English shows more uses of the progressive aspect
than British English does, whereas British English news broadcasts show a
higher proportion of forms of the perfect aspect.
Differences in the grammar of different varieties can, however, also be
more dramatic. For instance, multiple negation can be found in many varie-
ties of English world-wide (Kortmann 2008: 492, Schneider 2008a: 769).
One further such feature is the occurrence of double modals (I tell you what
we might should do) to be found in Britain in Northern English, Scottish
and Shetland/Orkney English (Kortmann 2008: 491) but also in Southern
American dialects and Jamaican Creole (Schneider 2008a: 766).481
As far as British and American English are concerned, there are a num-
ber of differences concerning inflexion (such as BrE got – AmE gotten, BrE
learnt – AmE learned) or the use of the articles in certain phrases (BrE go
to university – AmE go to a university).482 Similarly, differences in valency
can be found: Trudgill and Hannah (52008: 70) give examples such as
− that uninflected come and go occur with an [INF]-complement in US
English but not in English English (Can I come have a cup of coffee
with you?)
− that verbs such as seem or look occur in English English with an [NP]-
complement (It seems a good idea), whereas a [like NP]-complement

480
Cf. Mittmann (2004: 212, 217). See also Herbst and Mittmann (2003).
481
Compare Kortmann (2008) for a survey of morphological and syntactic varia-
tion in the British Isles and Schneider (2008a) for the Americas and the Car-
ribean.
482
For these and further differences between English English and American
English see Hannah and Trudgill (52008: 59–82).
314 Variation

can occur both in English English and US English (It seems like a good
idea).
A further level where differences between British and American English
can be observed is spelling:

BrE AmE
individual words cosy, monologue cozy, monolog/monologue
-our vs. -or colour, rigour, splendour color, rigor, splendor
-ise vs. -ize analyse, realize/realise analyze, realize
-re vs. -er centre, kilometre, theatre center, kilometer, theater
doubling of conso- travelled, travelling traveled, traveling
nants
20 Linguistic change

20.1 Types of linguistic change

This chapter is concerned with linguistic change and illustrates some of the
important changes that have taken place in the history of the English lan-
guage in the areas of pronunciation, lexis and grammar.483
The causes of linguistic change are generally described as falling into
two different categories. The first of these – external causes – are related
to language contact situations in which a language, or a particular variety of
a language, comes into contact with another language or another variety.
Thus, for instance, the Scandinavian settlements in the Northern parts of
England during the Old English period led to a considerable amount of
contact between speakers of English and speakers of Norse and a situation
of bilingualism in the North of England484 just as the Norman Conquest
resulted in contact between people who spoke Norman French and people
who spoke English. Depending on the direction of the influence, one can
distinguish between

substrate influence, i.e. a situation in which a language (or variety)
exerts influence on that of speakers in socially or politically superior po-
sition as in the case of Celtic influence on English,485

483
For a discussion of the nature of linguistic change see Traugott (2002: esp.
21): “From a ‘functionalist’ perspective … change is the result of strategic in-
teraction, specifically of choice-making on the part of speakers/writers in in-
teractional negotiation with addressees/readers. This includes, but is not limi-
ted to, conveying of information. On this view, language change is the result
of innovation in the individual and spread of the innovation to the commu-
nity, as suggested by Weinreich, Labov and Herzog (1968).”
484
See Kastovsky (1992: 329–332).
485
See Lutz (2009: 229): “… a substratum language may exert far-reaching
structural influence on the phonology, morphology, and syntax of the su-
perstratum if many speakers of the substratum abandon their own language
for the superstratum, i.e. in case of large-scale language shift. Thus, scholars
looking for Celtic loanwords in English as a result of the Anglo-Saxon con-
316 Variation

superstrate influence, i.e. a situation in which the language of a group

of speakers in a socially or politically superior position influences that
of the speakers in an inferior position as in the case of French influence
on English after 1066, and

adstrate influence, i.e. a situation in which a language influences a
language or variety of a group of speakers of equal status.486
As opposed to such external causes of linguistic change there are also in-
ternal causes. Brinton and Arnovick (2006: 56–57) mention factors such
as ease of articulation (which may show in assimilation or elision of
sounds), perceptual clarity (which may conflict with ease of articulation) or
tendencies such as the devoicing of final consonants or the loss of final [n].

20.2 Sound change

20.2.1 The phoneme systems of Old English and RP

That the sound system of a language is subject to constant modification is

obvious. The extent of the changes becomes clear when one compares the
phoneme system of Old English with that of RP today,487 which for the
purpose of easy comparison will be represented in the following form:
Old English Present-Day British English (RP)
Monophthongs /iː/ /i/ /yː/ /y/ /uː/ /u/ /iː/ /ɪ/ /uː/ /ʊ/
/eː/ /e/ /oː/ /o/ /e/
/зː/ /ə/ /ɔː/
/æː/ /æ/ /æ/
/ʌ/
/ɑː/ /ɑ/ /ɑː/ /ɒ/

____________________________
quest are looking for the wrong type of traces of this kind of contact …”. See
also Vennemann (1998: esp. 245–248).
486
For a more detailed discussion of the causes of change see Brinton and Arno-
vick (2006: 60–61). For a discussion of the notion of adstrate see also Schnei-
der (2007).
487
RP-system adapted from Gimson (41989: 80 and 82), Old English system
based on Hogg (1992: 85–86) and Lass (2006: 53–54). Lass (2006: 54) also
includes double consonants; cf. also Hogg (1992: 89). For unstressed vowels
see Hogg (1992: 88). For a more detailed account see Hogg (1992: 84–95).
 20 Linguistic change 317

Diphthongs /eːo/ /eo/ /æɑː/ /æɑ/ /eɪ, əʊ, aʊ, aɪ, ɪə, ʊə, ɔɪ, eə/
Consonants /p/ /b/ /t/ /d/ /k/ /g/ /p/ /b/ /t/ /d/ /k/ /g/
/f/ /θ/ /s/ /f/ /v/ /θ/ /ð/ /s/ /z/
/ʃ/ /h/ /ʃ/ /ʒ/ /h/
/tʃ/ /dʒ/ /tʃ/ /dʒ/
/m/ /n/ /m/ /n/ /ŋ/
/l/ /r/ /l/ /r/
/j/ /w/ /j/ /w/

It is immediately obvious that there are fewer differences in the consonant

system than in the vowel system. In the case of the vowels, Old English has
rounded front vowel phonemes /y;/ and /y/ (in words such as mys [my;s]
(‘mice’) and cyning [kynɪŋg] (‘king’)), which modern English lacks. What
is also very noticeable is the fact that there are considerable discrepancies
between the lists of diphthongs and in particular that RP (but not American
English and other rhotic varieties) has the series /ɪə/ /eə/ /ʊə/. The most
striking fact about the consonants is that the voiceless and voiced counter-
parts of the fricatives /f/ /θ/ /s/ are to be treated as allophones of a single
phoneme since they are in complementary distribution:488

/f θ s/ [v ð z] between voiced sounds [su;ðan] ‘from the South’

[f θ s] in all other positions [sæ;] ‘sea’

[su;θ] ‘South’

It should be clear that a comparison of the phoneme systems at two differ-

ent stages of the language can lead to valuable insights about the nature of
systems. However, it does not allow any conclusions as to the stability or
development of the system or of individual sounds. Thus while both in Old
English and in present-day RP an /i;/-phoneme can be identified, these do
not occur in the same words: in fact, the /i;/ of sea must be traced back to
Old English /æ;/ in sæ,489 whereas /i;/ can be found in a word such as drifan
/dri;fan/ in Old English, which has developed into /aɪ/.

488
Similarly, [ŋ] occurs as an allophone of /n/ before velar consonants. The
phoneme /h/ is analysed by Hogg (1992: 92) as having three allophones in
Old English: [h] [x] [C].
489
Note that the symbol æ is meant to represent “æ”.
318 Variation

20.2.2 Types of sound change

A distinction can thus be made between sound changes that affect the na-
ture of the phonological system and those that do not. A phoneme
merger occurs when a phonological distinction is given up. An example of
this kind of change is the contrast between Old English /y;/ and /i;/, which
was given up at the end of the Old English period when /y;/ developed into
/i;/ (/my;s/ > /mi;s/, which later became /maɪs/) and the opposition was lost.
The opposite development is that of a phoneme split, which happens
when a new phonological opposition is created. This happened, for in-
stance, in the case of the fricatives /f θ s/, where, caused by a number of
developments such as the loss of morphological endings and French influ-
ence in early Middle English,490 the voiced sounds [v ð z] came to occur in
the same positions as [f θ s]. Thus the sounds that had originally been allo-
phones of each other were no longer in complementary distribution and
created a phonological opposition. As a result, separate phonemes /f/ and
/v/, /θ/ and /ð/ as well as /s/ and /z/ have to be identified. When a change
just affects the quality of sounds but not the number of phonological oppo-
sitions in a language, one speaks of a phoneme shift. An example of a
phoneme shift in English is the Great Vowel Shift, which will be discussed
in 20.2.3.2.
While the distinction between phoneme mergers, phoneme splits and
phoneme shifts refers to the phonological system of a language, the distinc-
tion between conditioned and unconditioned changes refers to the condi-
tions under which a change can take place. An unconditioned sound
change is one which affects the sound in question irrespective of the envi-
ronment in which it occurs, whereas a conditioned sound change only
happens under certain circumstances. The development of Old English /y;/
into /i;/ is an example of such an unconditioned change since /y;/ was sub-
ject to that change irrespective of its position in the word or any condition
determining a particular phonetic context, in other words, all Old English
/y;/s were affected by this change. On the other hand, the emergence of /y;/
in words such as /my;s/ in Old English can be traced back to a conditioned
change that is known as i-mutation (> 20.2.3.1).
A third distinction concerns the phonetic character of sound changes,
where qualitative sound changes, which (like i-mutation) affect the

490
For details see Lass (2006: 62).
 20 Linguistic change 319

quality of the sounds, must be distinguished from quantitative sound

changes such as lengthening or shortening.

20.2.3 Important sound changes in the history of English

20.2.3.1 I-mutation

I-mutation (also called i-umlaut) took place between Germanic and Old
English and is a good example of a conditioned sound change since it only
affected vowels that were followed by a syllable containing [j] or [i]. Hogg
(1992: 113) describes i-mutation as follows:
… Old English vowels harmonised to an /i/ or /j/ following them in the
same word. This caused all back vowels to front and all short front vowels
(except, naturally, /i/) and diphthongs to raise when an /i/ or /j/ followed in
the next syllable. We can tabulate this as follows:
i; y(;) u(;)

e ø(;) o(;)

æ(;) ɑ(;)
Since some Germanic noun plural endings contained [i] and [j], i-mutation
can be seen as the cause of those plural forms in modern English that show
a different vowel from the singular such as goose – geese, tooth – teeth
(Old English /o;/ and /e;/ < /o;/ in the forms that gave rise to the present-
day singular and plural forms) or mouse – mice (Old English /u;/ and /y;/ <
/u;/ respectively).491

20.2.3.2 The Great Vowel Shift

One of the most important sound changes in the history of English is the
so-called Great Vowel Shift. It affected the Middle English long vowels,
all of which were raised or diphthongized between the Middle English pe-
riod and Present-Day English, as can be seen from the following examples:

491
It is important to bear in mind that /ø(;)/ was unrounded to /e(;)/ in West
Saxon. For a detailed description of i-mutation see Hogg (1992).
320 Variation

Middle GVS Further changes up to Present-Day English

English

/i;/ > /ei/ > /ai/ life

/e;/ > /i;/ deed

/ε;/ > /e;/ > /i;/ deal

/a;/ > /ε;/ > /eɪ/ make

/ɔ;/ > /o;/ /ou/ > /əʊ/ home

/o;/ > /u;/ food

/u;/ > /ou/ /aʊ/ house

Part of this development is due to the Great Vowel Shift. There has been
considerable discussion about what started the Great Vowel Shift, including
explanations in terms of a “drag chain” or a “push chain”: the idea of a drag
chain, proposed by Otto Jespersen (1909: 232), rests on the assumption that
the development started by a diphthongisation of the close vowels /i;/ and
/u;/ as a result of which /e;/ and /o;/ moved upwards. Luick, however, saw
the raising of /e;/ and /o;/ as the starting point of the development so that
/i;/ and /u;/ were pushed out of their original positions.492 More recently,
Lass (1999: 80) proposes distinguishing between two phases, “the early
push chain initiated by the raising of ME /e; o;/” and “the later raising of
the lower vowels”.493 After the Great Vowel Shift, a number of qualitative
changes can be observed as well as one major development, namely the
phonological merger of Middle English /e;/ (see) and /ε;/ (sea).
The fact that English spelling had widely been fixed by the time the
Great Vowel Shift set in explains some of the major discrepancies between

492
For a more detailed account of these controversial views see Lass (1999: 74–
75). Compare Luick’s (1940: 554) remarks about the “first impulse” (“erster
Impuls”) and Jespersen’s (1909: 233) view “that the whole shift began at the
upper end”. See also Stockwell’s (2002) discussions and the account given by
Brinton and Arnovick (2006: 308), who also point out that “[s]cholars argue
over whether this was really one shift or a series of separate ones”.
493
For the identification of a “First Push” in the form of an upward movement of
the two late Old English vowels /æ; / and /ɑ; / see Lutz (2004: 220–221.)
 20 Linguistic change 321

pronunciation and spelling in English, as is reflected in the orthographic

distinction between sea and see, for instance.494

20.2.3.3 Quantitative changes

As far as lengthening of vowels is concerned, two important conditions for

such lengthening ought to be mentioned:495
− before consonant clusters consisting of a nasal or /l/ or /r/ followed by a
plosive with the same place of articulation (a so-called homorganic plo-
sive). It is due to this type of lengthening that words such as climb or
child have /aɪ/ in present-day English. Since this lengthening did not
take place when a third consonant followed, children has /ɪ/, for in-
stance.
− in stressed open syllables of disyllabic words. This type of lengthening
resulted in a long vowel which became diphthongized in words such as
name.
Shortening occurs under the following conditions:
− before double consonants except homorganic clusters (wisdom < /i;/)
− in the first syllable of a trisyllabic word (southern < suðerne /u;/)

20.2.3.4 Present-day reflections

One interesting aspect of the operation of such sound laws is that although
they can be regarded as regular processes taking place in the language, the
result is one of increasing irregularity.496 Thus, the successive operation of
various of the sound laws described above can describe a considerable
number of irregularities of present-day English:

494
For a definition of the terms grapheme and allograph and an outline of cor-
respondences between English pronunciation and spelling see Arnold and
Hansen (81992: 57–84); for the different ways in which English phonemes
can be spelt compare also Gimson (41989).
495
See Brunner (1960/21984: 251) and Bammesberger (1989: 58–60). For quan-
titative changes relevant to the Great Vowel Shift see also Lutz (2004).
496
Compare Strang’s (1970: 293) discussion of examples such as clean and
cleanse or wise and wisdom in this context.
322 Variation

− The vowel change in the singular and plural forms of nouns such as
goose – geese or mouse – mice can be explained by the fact that the
Germanic plural suffix caused i-mutation, creating a difference of vowel
between singular and plural. Both the vowels of the singular and the
plural were later subject to the Great Vowel Shift.
− The difference in the vowels in the pairs /faɪv/ – /fɪfθ/ and /saʊθ/ –
/sʌðən/ in present-day English can be explained by the fact that in fifth
and southern the long vowels /i;/ and /u;/ were shortened, /ʊ/ later de-
veloping into /ʌ/, whereas the vowels in five and south remained unaf-
fected by such shortening and were diphthongized in the Great Vowel
Shift.
− The irregularity in verbs such as keep – kept – kept or meet – met – met
has similar causes. In Old English, cepan and metan both had a long
/e;/-vowel. Both verbs were weak verbs, forming their past tense with a
suffix –ed, which resulted in forms such as mette, with long /e;/ and a
double consonant, which caused shortening of the vowel to /e/. Since
only long vowels were subject to the Great Vowel Shift, present-day
English shows vowel alternation between /i;/ in the present tense forms
and /e/ in the past tense forms and the past participle.

20.3 Lexis

20.3.1 New words

The most obvious type of lexical change is the use of new words. New
words can come into existence on the basis of existing language material by
productive word formation processes (> 9.1.2) or by a process of borrowing
in which words are “taken over” from another language. The amount of
loan words in present-day English is enormous – Kastovsky (2006: 202)
estimates that about 70 per cent of the vocabulary of present-day English
are loans, mostly from French or Latin and Greek.497 The following list
contains a few examples:498

497
The influence of one language on another need not necessarily take the form
of a complete word being taken over: in some cases, a word that exists in the
language already may take over a new sense under the influence of another
language (a process that can be referred to as semantic borrowing) or a new
word or phrase may be coined on the model of another language but using
 20 Linguistic change 323

donor language
Latin abbot, mile, port, street (Old English: continental period
up to 1066) – cause, contradiction, history, include, in-
credible, interrupt, polite, summary (Middle English) –
focus, formula, nucleus, premium, status (Early Modern
English)
French appetite, authority, court, crown, dinner, government,
justice, orange, parliament, plead, pork (Middle English)
– grotesque, police, scene, soup (Early Modern English)
Scandinavian law, knife (Old English) – against, anger, both, fog, give,
husband, seem, sky, smile, take, they/their/them, Thurs-
day, want, window (Middle English)
Greek alphabet, drama (via Latin; Early Modern English)
Italian bankrupt, volcano, sonata, umbrella (Early Modern Eng-
lish)
Spanish creole, banana, sherry (Early Modern English)

It is important to note that words taken from other languages differ with
respect to the degree of integration. Thus for a word such as corpus the
Latin plural corpora can be used in present-day English alongside a form
corpuses, other words such as stimulus or alumnus only have i-plurals,
whereas words such as apparatus, bonus, campus only take regularized
plural forms in -s.499

____________________________
existing language material – thus Kastovsky (2006: 258) gives make/pay a vi-
sit, in detail, in particular as examples of a category loan translation. It has to
be pointed out, however, that such distinctions are by no means unproblema-
tic. Cf. the classification suggested by Betz (21965). See also Carstensen
(1975). For the inadequacy of this classification for the description of angli-
cisms in German and an alternative approach see Carstensen (1993: 53–62).
For a description of different types of anglicisms arising in film translation
see Herbst (1994: 129–150).
498
Examples taken from Kastovsky (2006) and Kastovsky (1992).
499
Cf. CGEL (1985: 5.92–101). The degree of integration is often taken as the
basis for a distinction between Lehnwort and Fremdwort. Cf. e.g. Gneuss
(1955); for a discussion of the distinction see e.g. Herbst (1994: 146–150).
324 Variation

The reverse process should also be mentioned, namely that words dis-
continue being used. Thus Old English words such as niman (correspond-
ing to German nehmen) or eag-Þyrel (‘window’) died out.

20.3.2 Changes of meaning

Another important type of change in the field of vocabulary concerns

changes of meaning: words can become polysemous, i.e. new lexical units
can come into existence, or meanings may die out. That such changes of
meaning can be quite dramatic is shown by the fact that the German word
Zaun ‘fence’, the English word town and the Dutch word tuin ‘garden’ or
English silly and German selig go back to the same Germanic words.
Changes of meaning can take different forms; thus holiday (originally
‘holy day’), sanctuary (originally ‘holy place’) or bonfire (originally ‘fire
of bones’) can be seen as examples of the sense of a word becoming more
general, whereas in the case of sermon (originally ‘speech’) or lust (origi-
nally ‘desire’) the present-day meaning is more restricted or specialized.500
Kastovsky (2006: 216) points out how such changes can affect words in the
same semantic field – the meaning of bird having become more general (in
Old English it meant ‘young bird, chicken’) and that of fowl (Old English
fugol meaning ‘bird’) having been extended. New senses can develop
through words being used metaphorically as in the case of mouse (‘com-
puter tool’). Interestingly, such “metaphorical” senses can come to be used
more frequently than the original sense as seems to be the case with a word
such as treadmill, for example.501 These few examples can serve as an indi-
cation of the complex nature of semantic change and the difficulty in de-
scribing it in terms of such general categories as generalization, specializa-
tion etc.502 In particular, one should remember that in many cases it may be
a matter of debate whether one should analyse a word as having developed

500
For a discussion of these examples see Brinton and Arnovick (2006: 77).
501
For examples such as current and mouse see Kastovsky (2006: 216). For
treadmill see Moon (1987: 92).
502
For categories such as weakening and strengthening, pejoration and ameliora-
tion, figurative shifts etc. see e.g. Brinton and Arnovick (2006: 76–85). Com-
pare also Leisi (1973: 114–132), who points out the disparate nature of such
criteria. For a detailed survey of types of change see also Hock (1991: 280–
308).
 20 Linguistic change 325

a “new” sense (and thus polysemous) or regard the sense of a word as hav-
ing become more general.

20.3.3 Homonymy

While semantic change can lead to polysemy (as in the case of mouse, for
example), homonymy is caused by phonetic and orthographic changes in
the language: thus both Old English ear (‘ear’) and Old English eare
(‘corn’) developed into Modern English ear, which can then be analysed as
a case of homonymy (> 15.1).

20.4 Grammar

20.4.1 Differences between Old English and Modern English

As far as grammatical differences between Old English and Modern Eng-

lish are concerned, it can be stated that Old English had a much more so-
phisticated system of inflexional endings than Modern English does.503

Gender: Old English nouns can be classified according to grammatical
gender (masculine, feminine and neuter), which is also reflected in ad-
jective inflexion. Grammatical gender was lost during the Middle Eng-
lish period.

Case: In Old English nouns have four distinct case forms (nominative,
accusative, genitive and dative), and adjectives and some pronouns dis-
tinguish a fifth case (instrumental). In contrast, present-day English
nouns show two morphologically marked case forms (a common case
and an ’s-genitive) and some pronouns three (nominative: he – genitive:
his – accusative: him).

503
For details of Old English morphology see Quirk and Wrenn (21957: 19–41)
or Lass (1994: 123–174). For a justification of the use of terms such as nomi-
native and accusative see CamG (2002: 456), for the use of the terms com-
mon case, subjective case and objective case see CGEL (1985: 5.112 and
6.2).
504
Traugott (1992: 273) points out that Old English word order was “not free;
rather, different word order patterns co-existed”. For a detailed description of
word order in Old English see Traugott (1992: 273–285).
326 Variation

Number: with respect to number, Old English personal pronouns do not

only show a contrast between singular and plural but also have special
dual forms (thus expressing a contrast between one, two, and more than
two).

Person: Old English verbs have three morphologically distinct forms for
first, second and third person singular indicative present tense forms and
a separate plural form for all persons, while Modern English only has
{S} to mark third person singular.
These examples serve to illustrate how the English language developed
from a predominantly synthetic towards an analytic language. It is impor-
tant to realize the role of phonological and morphological changes between
Old English and Middle English in this context: due to the reduction of
vowels in unstressed syllables many inflexional endings were levelled or
lost, which means that the role of inflexion as a means of expressing mean-
ing (and assigning syntactic functions such as subject) was reduced. At the
same time, the relatively free word order typical of Old English main
clauses became more fixed (i.e. the number of word order patterns available
in the language decreased)504 and periphrastic constructions were increas-
ingly being used.505 Similarly, the semantic role of a BENEFICIARY or
RECIPIENT typically associated with the dative case in Old English or Ger-
man can be expressed by word order (traditionally described as the indirect
object preceding the direct object) or by a to_NP-construction.
There does not seem to be agreement about the causes of these changes:
on the one hand, the loss of inflexions is often seen as the reason of the
emergence of periphrastic constructions and a more fixed type of word
order. On the other hand, one could imagine that the fact that the respective
meanings were expressed by new constructions meant that inflexions

505
In Old English, as in present-day German, case still played an important role
in distinguishing subjects and objects, for example. If there is no contrast bet-
ween nominative and accusative in noun inflexions, case does not provide a
means of indicating the subject of a clause, for instance – a function which
can also be expressed by a word order in which subjects precede verbs in
declarative clauses. Note, however, that nominative and accusative were by
no means formally distinguished with all noun classes in Old English. Thus
a-stems had identical forms for nominative and accusative (sg. dæg – pl. da-
gas), for example. See Lass (1994: 129–139) for case forms of nouns. For a
more detailed account of these developments in Middle English see Brinton
and Arnovick (2006: 266–297).
 20 Linguistic change 327

ceased to have an important function and were lost.506 In any case, it is

important to realize that the English language underwent dramatic changes
between late Old English and Early Modern English in this area; in fact
Brinton and Arnovick (2006: 295) describe the structural changes of the
Middle English period as “the most significant and far-reaching grammati-
cal changes in the history of the language”.

20.4.2 Analogy

One important type of change in the field of grammar is presented by anal-

ogy.507 This is particularly noticeable in the field of morphology, for in-
stance, when speakers start using “regular” {S}-plurals with words that
(used to) have irregular plural forms such as corpuses instead of corpora.
In a similar way, the plural form of one of the Old English noun stems was
extended to words of other classes so that plural forms such as books,
names or horses can be seen as analogical.508
Similarly, the fact that a considerable number of Old English strong
verbs (> 8.4) began developing regular {D}-past tense forms in the Middle
English period can be seen as being caused by analogy to the large number
of verbs following the regular pattern. This means that the ablaut-forms of a
verb such help (helpan – healp – hulpon – geholpen in Old English) were
regularized.509 Bybee (1995: 426) argues that irregularity is linked to token
frequency because irregular forms will only be stored in the mental lexicon
if they occur frequently enough. Thus it is not surprising that the verb be
should distinguish the greatest number of morphological forms of all verbs
in English (e.g. was – were) because each of them is sufficiently frequent510
(> 9.6).

506
See Brinton and Arnovick (2006: 295–297) for a discussion of different
views; compare also the discussion and the references given in Fischer and
van der Wurff (2006: esp. 185–187). See also Görlach (1974: 96–98).
507
For a detailed discussion of analogy see Hock (1991: 167–236).
508
Cf. Lass (1992: 96).
509
For a detailed outline of the development of the Old English strong verbs see
Brunner (1962: 209–252).
510
Note, however, that the frequency criterion does not apply in the same way to
irregular plural forms of nouns.
328 Variation

20.4.3 Grammaticalization

A further very important type of change can be described as grammaticali-

zation, which Hopper and Traugott (22003: xv) define “as the process
whereby lexical items and constructions come in certain linguistic contexts
to serve grammatical functions, and, once grammaticalized, continue to
develop new grammatical functions”.511
A good example of the concept of grammaticalization is the be going to-
construction of present-day English, which is quite obviously related to the
verb go. Following Hopper and Traugott (22003: 2–3) one can imagine a
sentence such as I am going to marry Bill to be interpreted originally as
containing a main verb go (in the sense of leave or travel) and a purpose
clause (in order to marry Bill.). A (new) meaning (of ‘futurity’) can be
inferred from the meaning of ‘purpose’ and become prominent if no direc-
tion or motion is being expressed. A very important factor in grammaticali-
zation is reanalysis: so one could argue that I am going to marry Bill can be
reinterpreted with marry being the main verb and am going to having a
kind of auxiliary function (as in the classification of be going to as a semi-
auxiliary in CGEL (1985: 3.47). Hopper and Traugott (22003: 3) point out
that the reanalysis is manifest when going to is used with verbs or in con-
texts in which a ‘purpose’ reading is not possible or at least very unlikely
as in I am going to like Bill or I am going to go to London. This reanalysis
can also result in phonological reduction, which only occurs in cases of
going to being used as a semi-auxiliary:
(1)QE Bill’s gonna go to college after all.
(2)QE *Bill’s gonna college after all.
As far as the meaning of go is concerned, Hopper and Traugott argue that
the components of motion and directionality of go (which it still has in
main verb uses) have been lost in the process of grammaticalization but
some new, specifically temporal meanings have been added.
There are numerous examples of grammaticalization: Brinton and Ar-
novick (2006: 74) mention the use of of as a marker of possession or the

511
For a discussion of the history of the term and different approaches to gram-
maticalization see Krug (2000: 11–18), Hopper and Traugott (22003: 18–31),
Traugott (2002) or Noël (2007). See also Lenker (2010: esp. 184–185) espe-
cially with respect to the aspect of uni-directionality in grammaticalization
processes.
 20 Linguistic change 329

emergence of a category of modal verbs out of Old English full verbs, for
example.512 Krug (2000: 250–251), in a corpus-based study on the gram-
maticalization of combinations such as have to, stresses the importance of
frequency in the development of new categories.513 Fischer and van der
Wurff (2006: 133) point out that the fact that the modal will is used with
inanimate subjects can be seen as an example of grammaticalization.
Whether in the light of the counterarguments provided in the major refer-
ence grammars of English this is a good enough reason for arguing in fa-
vour of a category future tense in present-day English depends on one’s
perspective and on one’s definition of tense (> 2.3).514
The discussion of the grammaticalization phenomena in present-day his-
torical linguistics can easily be related to the view expressed by corpus
linguists such as Sinclair or construction grammarians that there is no clear
dividing line between lexis and grammar (> 10.5 and 13.2).

512
For the development of modal verbs out of preterite-present verbs see Krug
(2000: 44–45). See also Lightfoot (1988: 311–313) and Fischer and van der
Wurff (2006: 146–152).
513
For the postulate of a new category of emerging modals see Krug (2000:
214).
514
The position of Fischer and van der Wurff (2006: 131) or of Brinton and
Arnovick (2006: 74) differs from that taken in CGEL (1985: 4.3) or CamG
(2002: 208–210).
Postscript

I very much hope that this book has succeeded in giving readers an idea of
the wide range of topics that fall under the scope of linguistics. The idea of
an introduction of this kind can only be to provide a relatively selective
survey of different fields of study and different approaches to their study
and in particular to raise curiosity as to what else there is. I have tried to
give a few hints as to where one could continue reading in order to get a
more profound account of individual aspects or an alternative view of
things from the one presented here.
Two points should perhaps be mentioned as a sort of personal conclu-
sion: firstly, given the diversity of perspectives in the analysis of language,
it is interesting to see that over the last few decades researchers in the fields
of corpus linguistics, foreign language linguistics, historical linguistics and
cognitive linguistics (in particular construction grammar and usage-based
theories of language acquisition) have come to conclusions about the nature
of language and the ways that the human mind processes language that
point in a relatively similar direction, which opens up very encouraging
perspectives for future research.
Secondly, it is important to realize that the insights gained by linguistic
analysis can be fruitfully applied to many fields which have not been
treated in this introduction. One particularly significant area that could be
mentioned here is speech therapy. Other important fields are related to
situations which involve more than one language. The insights that Sinclair
subsumed under the term idiom principle are highly relevant to, for in-
stance, translation studies and lexicography. In fact, the amount and type of
coverage that multi-word units of different kinds have received in some of
the more recent editions of English foreign learners’ dictionaries is a good
example of the direct relation between linguistic research and its applica-
tions. It should always be borne in mind that practical descriptive work, as
is done in dictionaries, or the analysis of difficulties arising in translation
can always be exploited for gaining theoretical insights into the nature of
language because whatever creates a problem in the use of language ought
to be accounted for in any theory of language. This applies equally to for-
 Postscript 331

eign language teaching, a field for which more recent developments in lin-
guistic theory offer very interesting perspectives indeed.
A third point that I would like to add as a sort of personal conviction is
that it is an extremely valuable exercise to try to put one’s own ideas about
language to the test by applying them to authentic language material. This
is one of the reasons why most of the examples used in this book were ei-
ther taken from corpora such as the British National Corpus or from books I
happened to be reading while working on this text. To those who have read
the examples carefully, it may not come as a great surprise to hear that parts
of this book were written in Cornwall. Nevertheless, it was not the prime
end of this book to create an interest in Cornish art (although this would be
a not altogether unwelcome side-effect) but to demonstrate the immense
fascination of the many facets and perspectives of the study of language.
Bibliography

Aarts, Bas
2000 Corpus linguistics, Chomsky and fuzzy tree fragments. In Corpus
Linguistics and Linguistic Theory: Papers from the Twentieth Inter-
national Conference on English Language Research on Computer-
ized Corpora (ICAME), Christian Mair, and Marianne Hundt (eds.),
5–13. Amsterdam/Atlanta: Rodopi.
Aarts, Flor, and Jan Aarts
2
1988 English Syntactic Structures. New York/Leyden: Prentice
Hall/Martinus Nijhoff. First edition 1982.
Abercrombie, David
1967 Elements of General Phonetics. Edinburgh: Edinburgh University
Press.
Ágel, Vilmos
2000 Valenztheorie. Tübingen: Narr.
Aijmer, Karin, and Bengt Altenberg (eds.)
1991 English Corpus Linguistics. London/New York: Longman.
Aitchison, Jean
3
2003 Words in the Mind: An Introduction to the Mental Lexicon. Oxford:
Blackwell.
Alexander, Richard J.
1992 Fixed expressions, phraseology and language teaching. Zeitschrift
für Anglistik und Amerikanistik 40 (3): 238–249.
Allen, J. P. B., and P. van Buren
1971 Chomsky: Selected Readings. London: Oxford University Press.
Allerton, David J.
1982 Valency and the English Verb. London: Academic Press.
Allerton, David J.
1990 Language as form and patterns: Grammar and its categories. In An
Encyclopaedia of Language, N. E. Collinge (ed.), 68–111. Lon-
don/New York: Routledge.
Altenberg, Bengt
1998 On the phraseology of spoken English: The evidence of recurrent
word-combinations. In Phraseology: Theory, Analysis, and Applica-
tions, A. P. Cowie (ed.), 101–121. Oxford: Clarendon Press.
 Bibliography 333

Anderson, Stephen R.
1988 Morphological theory. In Linguistics: The Cambridge Survey, Fre-
derick J. Newmeyer (ed.), 146–191. Cambridge: Cambridge Univer-
sity Press.
Arnold, Roland, and Klaus Hansen
8
1992 Englische Phonetik. Leipzig: Langenscheidt/Enzyklopädie. First
edition 1975.
Armstrong, Sharon Lee, Lila R. Gleitman, and Henry Gleitman
1999 What some concepts might not be. In Core Readings, Eric Margolis,
and Stephen Laurence (eds.), 225–259. Cambridge, Mass./London:
MIT Press.
Aronoff, Mark, and Frank Anshen
1998 Morphology and the lexicon: Lexicalization and Productivity. In The
Handbook of Morphology, Andrew Spencer, and Arnold M. Zwicky
(eds.), 237–247. Oxford/Malden: Blackwell.
Austin, John L.
1962 How to Do Things with Words. London: Oxford University Press.
Ayto, John
1996 Lexical life expectancy: A prognostic guide. In Words: Proceedings
of an International Symposium Lund, 25–26 August 1995, Konferen-
ser 36, Jan Svartvik (ed.), 181–188. Stockholm: Kungl. Vitterhets
Historie och Antikvitets Akademien.
Bahns, Jens
1996 Kollokationen als lexikographisches Problem: Eine Analyse allge-
meiner und spezieller Lernerwörterbücher des Englischen. Tübin-
gen: Niemeyer.
Bammesberger, Alfred
1989 English Linguistics. Heidelberg: Winter.
Barnickel, Klaus Dieter
1980 Sprachliche Varianten des Englischen. 2 Vols. München: Hueber.
Bauer, Laurie
1983 English Word-Formation. Cambridge: Cambridge University Press.
Bauer, Laurie
1994 Watching English Change: An Introduction to the Study of Linguis-
tic Change in Standard Englishes in the Twentieth Century. Lon-
don/NewYork: Longman.
Baumgärtner, Alfred Clemens
1970 Konsistenz und Dependenz. In Vorschläge für eine strukturale
Grammatik des Deutschen, H. Steger (ed.), 57–77. Darmstadt: Wis-
senschaftliche Buchgesellschaft.
de Beaugrande, Robert-Alain, and Wolfgang Ulrich Dressler
1981 Introduction to Text Linguistics. London/New York: Longman.
334 Bibliography

Behrens, Heike
1999 Was macht Verben zu einer besonderen Kategorie im Spracherwerb?
In Das Lexikon im Spracherwerb, Jörg Meibauer, and Monika Roth-
weiler (eds.), 32-50. Tübingen: Francke.
Behrens, Heike
2005 Wortartenerwerb durch Induktion. In Wortarten und Grammatikali-
sierung: Perspektiven in System und Erwerb, Clemens Knobloch,
and Burkhard Schaeder (eds.), 177–198. Berlin/New York: de
Gruyter.
Behrens, Heike
2007 The acquisition of argument structure. In Valency: Theoretical, De-
scriptive and Cognitive Issues, Thomas Herbst, and Katrin Götz-
Votteler (eds.), 193–214. Berlin/New York: Mouton de Gruyter.
Behrens, Heike
2008 Corpora in language acquisition research: History, methods, perspec-
tives. In Corpora in Language Acquisition Research: History, Meth-
ods, Heike Behrens (ed.), xi–xxx. Amsterdam: Benjamins.
Behrens, Heike
2009 Usage-based and emergentist approaches to language acquisition.
Linguistics 47 (2): 383–411.
Berlin, Brent, and Paul Kay
1969 Basic Color Terms: Their Universality and Evolution. Berkeley/ Los
Angeles: University of California Press.
Betz, Werner
2
1965 Deutsch und Lateinisch: Die Lehnbildungen der althochdeutschen
Benediktinerregel. Bonn: Bouvier. First edition 1949.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad, and Edward Fine-
gan
1999 Longman Grammar of Spoken and Written English. Harlow: Long-
man. [LGSWE]
Bloomfield, Leonard
1933 Language. New York: Holt. (21935), London: Allen & Unwin Ltd.
Bolinger, Dwight
2
1975 Aspects of Language. New York: Harcourt Brace Jovanovich.
Brekle, Herbert Ernst
1985 Einführung in die Geschichte der Sprachwissenschaft. Darmstadt:
Wiss. Buchgesellschaft.
Brinker, Klaus
1977 Modelle und Methoden der strukturalistischen Syntax. Stuttgart:
Kohlhammer.
 Bibliography 335

Brinker, Klaus
3
1992 Linguistische Textanalyse: Eine Einführung in Grundbegriffe und
Methoden. Berlin: Schmidt.
Brinton, Laurel J., and Leslie K. Arnovick
2006 The English Language: A Linguistic History. Oxford: Oxford Uni-
versity Press
Brown, Gillian
1977 Listening to Spoken English. London: Longman.
Brown, Gillian, and George Yule
1983 Discourse Analysis. Cambridge: Cambridge University Press.
Brunner, Karl
1962 Die englische Sprache. Vol. 2. Tübingen: Niemeyer.
Brunner, Karl
2
1984 Die Englische Sprache. Vol. 1: Allgemeines. Lautgeschichte. Tübin-
gen: Niemeyer. First edition 1960.
Bublitz, Wolfram
2
2009 Englische Pragmatik: Eine Einführung. Berlin: Schmidt.
Burgschmidt, Ernst
1973–5 System, Norm und Produktivität in der Wortbildung: Aufsätze. Er-
langen: E. Burgschmidt. 3. Vols.
Burgschmidt, Ernst
1977 Strukturierung, Norm und Produktivität in der Wortbildung. In Per-
spektiven der Wortbildungsforschung, H. E. Brekle, und D.
Kastovsky (eds.), 39–47. Bonn: Bouvier.
Burgschmidt, Ernst
1978 Wortbildung im Englischen. Dortmund: Lambert Lensing.
Burgschmidt, Ernst, and Dieter Götz
1974 Kontrastive Linguistik Deutsch/Englisch. München: Hueber.
Bußmann, Hadumod
4
2008 Lexikon der Sprachwissenschaft. Stuttgart: Kröner.
Bybee, Joan
1995 Regular morphology and the lexicon. Language and Cognitive Proc-
esses 10 (5): 425–455.
Bybee, Joan
2005 From usage to grammar: The mind’s response to repetition (manu-
script: http://www.unm.edu/~jbybee/Bybee%20plenary.pdf
[1.4.2007]).
Bybee, Joan
2007 The Emergent Lexicon. In Frequency of Use and the Organization
of Language, Joan Bybee (ed.), 279–293. Oxford: Oxford University
Press.
336 Bibliography

Bybee, Joan, and Paul Hopper

2001 Introduction to frequency and emergence of linguistic structure. In
Frequency and the Emergence of Linguistic Structure, Joan Bybee
and Paul Hopper (eds.), 1–24. Amsterdam/Philadelphia: Benjamins.
Cameron-Faulkner, T., Elena Lieven, and Michael Tomasello
2003 A Construction Based Analysis of Child Directed Speech. Cognitive
Science 27: 843–873.
Carstensen, Broder
1975 Amerikanische Einflüsse auf die deutsche Sprache. In Jahrbuch für
Amerikastudien, Ernst Fraenkel, Hans Galinsky, Dietrich Gerhard,
and H. J. Lang (eds.), 34–55. Heidelberg: Winter. First edition 1963.
Catford, J.C.
1988 A Practical Introduction to Phonetics. Oxford: Clarendon Press.
Chomsky, Noam
1957 Syntactic Structures. The Hague: Mouton.
Chomsky, Noam
1959 Review of Skinner Verbal Behavior. Language 35: 26–58.
Chomsky, Noam
1965 Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press.
Chomsky, Noam
1966 Topics in the Theory of Generative Grammar. The Hague/Paris:
Mouton.
Chomsky, Noam
1967 Recent contributions to the theory of innate ideas. Synthese 17: 2–
11.
Chomsky, Noam
1970 Remarks on nominalization. In Readings in English Transforma-
tional Grammar, R. A. Jacobs, and P. S. Rosenbaum (eds.), 184–
221. Waltham, Mass./London: Ginn and Company.
Chomsky, Noam
1981 Lectures on Government and Binding. Dordrecht: Foris.
Chomsky, Noam
1986 Knowledge of Language: Its Nature, Origin, and Use. New
York/Westport/London: Praeger.
Chomsky, Noam
1988 Language and Problems of Knowledge: The Managua Lectures.
Cambridge, Mass.: MIT Press.
Chomsky, Noam
1995 The Minimalist Program. Cambridge, Mass./London: MIT Press.
Chomsky, Noam
2004 Linguistics in the 21st Century. An Interview with Noam Chomsky.
[Interviewed by Naoki Fukui and Mihoko Zushi, November 22,
 Bibliography 337

2002], in The Generative Enterprise Revisited, Noam Chomsky,

Berlin/New York: Mouton de Gruyter.
Chomsky, Noam, and Morris Halle
1968 The Sound Pattern of English. New York/Evanston/London: Harper
& Row.
Clark, Eve V.
1998 Morphology in language acquisition. In The Handbook of Morphol-
ogy, Andrew Spencer, and Arnold M. Zwicky (eds.), 374–389. Ox-
ford/Malden: Blackwell.
Clark, Herbert H., and Eve V. Clark
1977 Psychology and Language: An Introduction to Psycholinguistics.
San Diego: Harcourt Brace Jovanovich.
Coates, Jennifer
1986 Women, Men and Language. A Sociolinguistic Account of Sex Dif-
ferences in Language. London/New York: Longman.
Coates, Richard
1987 Lexical morphology. In New Horizons in Linguistics 2, John Lyons,
Richard Coates, Margaret Deuchar, and Gerald Gazdar (eds.), 103–
121. Harmondsworth: Penguin.
de Cock, Sylvie
2000 Repetitive phrasal chunkiness and advanced EFL speech and writ-
ing. In Corpus Linguistics and Linguistic Theory, Christian Mair,
and Marianne Hundt (eds.), 51–68. Amsterdam/Atlanta: Rodopi.
Cook, Vivian J.
1988 Chomsky’s Universal Grammar. Oxford: Blackwell.
Coseriu, Eugenio
1973 Probleme der strukturellen Semantik. Tübingen: Narr.
Coulthard, Malcolm
2
1985 An Introduction to Discourse Analysis. London/New York: Long-
man. First edition 1977.
Coupland, Nikolas, and Hywel Bishop
2007 Ideologised Values for British Accents. Journal of Sociolinguistics
II/I: 74–93.
Cowie, Anthony P.
1981 The treatment of collocation and idioms in learners’ dictionaries.
Applied Linguistics 2 (3): 223–235.
Croft, William
2000 Lexical and grammatical meaning. In Morphologie. Morphology.
Ein internationals Handbuch zur Flexion und Wortbildung. An In-
ternational Handbook on Inflection and Word-Formation 1.
Halbband, Vol. 1, Geert Booij, Christian Lehmann, and Joachim
338 Bibliography

Mugdan in collaboration with Wolfgang Kesselheim, Stavros Sko-

peteas (eds.), 257–263. Berlin/New York: Walter de Gruyter.
Croft, William, and David A. Cruse
2004 Cognitive Linguistics. Cambridge: Cambridge University Press.
Cruse, David A.
1986 Lexical Semantics. Cambridge: Cambridge University Press.
Cruttenden, Alan
1986 Intonation. Cambridge: Cambridge University Press.
Crystal, David
1988 The English Language. Harmondsworth: Penguin.
Crystal, David
2
1997 The Cambridge Encyclopedia of Language. Cambridge/New
York/Melbourne: Cambridge University Press.
Crystal, David
2001 Language and the Internet. Cambridge: Cambridge University Press.
Crystal, David
2
2003 The Cambridge Encyclopaedia of the English Language, Cam-
bridge: Cambridge University Press.
Crystal, David
2006 English worldwide. In A History of the English Language, Richard
Hogg, and David Denison (eds.), 420–444. Cambridge: Cambridge
University Press.
Crystal, David, and Derek Davy
1975 Advanced Conversational English. London: Longman.
Culler, Jonathan
1976 Saussure. Glasgow: Fontana.
Daneš, František
1970 Zur linguistischen Analyse der Textstruktur. Folia Linguistica 1970,
72–78. [Also in Sprachwissenschaft: Ein Reader, Ludger Hoffmann,
(ed.), 591–597. Berlin/New York: Walter de Gruyter, 2000.]
Denison, David and Richard Hogg
2006 Overview. In A History of the English Language, Richard Hogg, and
David Denison (eds.), 1–42. Cambridge: Cambridge University
Press.
Dirven, René, and Günther Radden
1977 Semantische Syntax des Englischen. Frankfurt: Athenaion.
Dretzke, Burkhard
1985 Fehlerbewertung im Aussprachebereich. Hamburg: Buske.
Dretzke, Burkhard
1998 Modern British and American English Pronunciation. Paderborn:
Schöningh (UTB).
 Bibliography 339

Dufter, Andreas, Jörg Fleischer, and Guido Seiler

2009 Introduction. In Describing and Modelling Variation in Grammar,
Andreas Dufter, Jörg Fleischer, and Guido Seiler (eds.), 1–18. Ber-
lin/New York: Mouton de Gruyter.
Edwards, John R.
1982 Language attitudes and their implications among English speakers.
In Attitudes towards Language Variation: Social and Applied Con-
texts, E. B. Ryan, and H. Giles (eds.), 20–33. London: Arnold.
Ellis, Nick
2003 Constructions, chunking, and connectionism: The emergence of
second language structure. In The Handbook of Second Language
Acquisition, Catherine J. Doughty, and Michael H. Long (eds.), 63–
103. Malden/Oxford/Carlton: Blackwell.
Elyan, Olwen, Philip Smith, Howard Giles, and Richard Bourhis
1978 RP-accented female speech: The voice of perceived androgyny? In
Sociolinguistic Patterns of British English, P. Trudgill (ed.), 122–
131. London: Arnold.
Emons, Rudolf
1974 Valenzen englischer Prädikatsverben. Tübingen: Niemeyer.
Emons, Rudolf
1978 Valenzgrammatik für das Englische. Eine Einführung. Tübingen:
Niemeyer.
Engel, Ulrich
1977 Syntax der deutschen Gegenwartssprache. Berlin: Schmidt.
Enkvist, Nils Erik
1978 Coherence, pseudo-coherence, and non-coherence. In Reports on
Text Linguistics: Semantics and Cohesion, Jan-Ola Östman (ed.),
109-128. Åbo: Stiftelsens för Åbo Akademi.
Esser, Jürgen
1979 Englische Prosodie: Eine Einführung. Tübingen: Narr.
Esser, Jürgen
1993 English Linguistic Stylistics. Tübingen: Niemeyer.
Esser, Jürgen
2009 Introduction to English Text-Linguistics. Frankfurt: Peter Lang.
Fanselow, Gisbert, and Sascha W. Felix
1987 Sprachtheorie 2: Die Rektions- und Bindungstheorie. Tübingen:
Francke (UTB).
Faulhaber, Susen
forthc. Verb valency patterns – challenging semantics-based accounts.
Berlin/New York: Mouton de Gruyter.
340 Bibliography

Fillmore, Charles
1968 The case for case. In Universals in Linguistic Theory, Emmon Bach,
and Robert T. Harms (eds.), 1–88. New York: Holt, Rinehart and
Winston.
Fillmore, Charles
1976 Frame semantics and the nature of language. In Origins and Evolu-
tion of Language and Speech: Annals of the New York Academy of
Sciences, Stevan R. Harnad, Horst D. Steklis, and Jane Lancaster
(eds.), 20–32. New York: The New York Academy of Sciences.
Fillmore, Charles
1988 The mechanisms of “construction grammar”. In General Session and
Parasession on Grammaticalization, Shelley Axmaker, Annie Jas-
sier, and Helen Singmaster (eds.), 35–55. Berkeley: Berkeley Lin-
guistics Society.
Fillmore, Charles
2007 Valency issues in FrameNet. In Valency: Theoretical, Descriptive
and Cognitive Issues, Thomas Herbst, and Katrin Götz-Votteler
(eds.), 129–160. Berlin/New York: Mouton de Gruyter.
Fillmore, Charles, and Beryl T. Atkins
1992 Toward a frame-based lexicon: The semantics of RISK and its
neighbors. In Frames, Fields, and Contrasts: New Essays in Seman-
tic and Lexical Organization, Adrienne Lehrer, and Eva Feder Kit-
tay (eds.), 75–188. Hillsdale/Hove/London: Lawrence Erlbaum As-
sociates.
Fillmore, Charles, Paul Kay, and Catherine M. O’Connor
1988 Regularity and idiomaticity in grammatical constructions: The case
of let alone. Language 64: 501–538.
Firbas, Jan
1992 Functional sentence perspective in written and spoken communica-
tion. Cambridge: Cambridge University Press.
Firth, John Rupert
1968 A synopsis of linguistic theory: 1930–55. In Selected Papers by J. R.
Firth 1952–59, F. R. Palmer (ed.), 168–205: London/Harlow: Long-
man. First edition 1957.
Fischer, Olga, and Win van der Wurff
2006 Syntax. In A History of the English Language, Richard Hogg, and
David Denison (eds.), 109–198. Cambridge: Cambridge University
Press.
Fischer, Kerstin and Anatol Stefanowitsch
2006 Konstruktionsgrammatik: Ein Überblick. In Konstruktionsgramma-
tik: Von der Anwendung zur Theorie, 3–17. Tübingen: Stauffenburg.
 Bibliography 341

Frege, Gottlob
2000 Einleitung in die Logik. In Sprachwissenschaft: Ein Reader, Ludger
Hoffmann (ed.), 682–686. Berlin /New York: Walter de Gruyter.
First edition 1906.
Garman, Michael
1990 Psycholinguistics. Cambridge: Cambridge University Press.
Garside, Roger
1987 The CLAWS word–tagging system. In The Computational Analysis
of English: A Corpus-Based Approach, Garside, Roger, Geoffrey
Leech, and Geoffrey Sampson (eds.), 30–41, London/ New York:
Longman.
Giegerich, Heinz J.
1992 English Phonology: An Introduction. Cambridge: Cambridge Uni-
versity Press.
Giles, Howard
1970 Evaluative reactions to accents. Educational Review 22: 211–227.
Gilquin, Gaëtanelle
2007 To err is not all: What corpus and elicitation can reveal about the use
of collocations by learners. Zeitschrift für Anglistik und Ameri-
kanistik 55 (3): 273–291.
Gimson, A. C.
4
1989 An Introduction to the Pronunciation of English. London: Arnold.
First edition 1962. Revised version by Susan Ramsaran.
Gimson, A.C., and Alan Cruttenden
6
2001 Gimson’s Pronunciation of English. London: Arnold. Revised by
Alan Cruttenden.
Gläser, Rosemarie
2
1990 Phraseologie der englischen Sprache. Leipzig: Verlag Enzyklopä-
die.
Gleason, Henry Allan
1961 Revised edition. An Introduction to Descriptive Linguistics. New
York: Holt, Rinehart & Winston.
Gneuss, Helmut
1955 Lehnbildungen und Lehnbedeutungen im Altenglischen. Ber-
lin/Bielefeld/München: Schmidt.
Görlach, Manfred
1974 Einführung in die englische Sprachgeschichte. Heidelberg: Quelle
und Meyer.
Götz, Dieter
1971 Studien zu den verdunkelten Komposita im Englischen. Erlanger
Beiträge zur Sprach- und Kunstwissenschaft. Nürnberg: Verlag Hans
Carl.
342 Bibliography

Götz, Dieter
1976 Stilistik und Idiomatik im Englischunterricht. Dortmund: Lensing.
Götz-Votteler, Katrin
2008 Aspekte der Informationsentwicklung im Erzähltext. Tübingen: Narr.
Goldberg, Adele
1995 Constructions: A Construction Grammar Approach to Argument
Structure. Chicago: Chicago University Press.
Goldberg, Adele
2006 Constructions at Work. Oxford/New York: Oxford University Press.
Granger, Sylviane
1998 Prefabricated patterns in advanced EFL writing: Collocations and
formulae. In Phraseology: Theory, Analysis, and Applications, A. P.
Cowie (ed.), 145–160. Oxford: Clarendon Press.
Granger, Sylviane
2009 The contribution of learner corpora to second language acquisition
and foreign language teaching: A critical evaluation. In Corpora and
Language Teaching, Karin Aijmer (ed.), 13–32. Amster-
dam/Philadelphia: Benjamins.
Granger, Sylviane
forthc. From phraseology to pedagogy: Challenges and prospects. In
Chunks in the Description of Language: A Tribute to John Sinclair,
Thomas Herbst, Susen Faulhaber, and Peter Uhrig (eds.), Ber-
lin/New York: Mouton de Gruyter.
Granger, Sylviane, and Magali Paquot
2008 Disentangling the phraseological web. In Phraseology: An Interdis-
ciplinary Perspective, Sylviane Granger, and Fanny Meunier (eds.),
27–49. Amsterdam/Philadelphia: Benjamins.
Granger Sylviane, E. Dagneaux, F. Meunier, and Magali Paquot
2009 International Corpus of Learner English. Vol. 2. Louvain-la-Neuve:
Presses universitaires de Louvain.
Greenbaum, Sidney
1974 Some verb-intensifier collocations in American and British English.
American Speech 49: 79–89. First edition 1974. [Reprinted in S.
Greenbaum (1988), 113–124.]
Greenbaum, Sidney
1988 Good English and the Grammarian. London/New York: Longman.
Greenbaum, Sidney, and Randolph Quirk
1970 Elicitation Experiments in English: Linguistic Studies in Use and
Attitude. London: Longman.
 Bibliography 343

Grice, Paul
1975 Logic and Conversation. In Syntax and Semantics. Vol. 3: Speech
Acts, P. Cole, and J. L. Morgan (eds.), 41–58. New York: Academic
Press.
Gries, Stefan Th.
2008 Phraseology and linguistic theory: A brief survey. In Phraseology in
Foreign Language Learning and Teaching, Fanny Meunier, and Syl-
viane Granger (eds.), 3–25. Amsterdam/Philadelphia: Benjamins.
Gries, Stefan Th., and Anatol Stefanowitsch
forthc. Cluster analysis and the identification of collexeme classes. In Em-
pirical and Experimental Methods in Cognitive/Functional Re-
search, John Newman, and Sally Rice (eds.), 73–90. Stanford: CSLI.
Grimm, Anne
2008 “Männersprache” – “Frauensprache”? Eine korpusgestützte empi-
rische Analyse des Sprachgebrauchs britischer und amerikanischer
Frauen und Männer hinsichtlich Geschlechtsspezifika. Hamburg:
Verlag Dr Kovač.
Gut, Ulrike
2008 Nigerian English: Phonology. In Varieties of English 4: Africa,
South and Southeast Asia, Rajend Mesthrie (ed.), 35–54. Berlin/New
York: Mouton de Gruyter.
Gut, Ulrike
2009 Introduction to English Phonetics and Phonology. Frankfurt: Lang.
Haegeman, Liliane
1991 Introduction to Government and Binding Theory. Oxford/Cam-
bridge, Mass.: Blackwell.
Halliday, Michael A. K.
1970a A Course in Spoken English: Intonation. London: Oxford University
Press.
Halliday, Michael A. K.
1970b Language Structure and Language Function. In Horizons in Linguis-
tics, John Lyons (ed.), 140–165. New Harmondsworth: Penguin.
Halliday, Michael A. K.
2
1994 An Introduction to Functional Grammar. London/Mel-
bourne/Auckland: Arnold.
Halliday, Michael, Angus McIntosh, and Peter Strevens
1964 The Linguistic Sciences and Language Teaching. London: Longman.
Halliday, Michael A. K., and Ruqaiya Hasan
1976 Cohesion in English. London: Longman.
Handl, Susanne
2008 Essential collocations for learners of English: The role of colloca-
tional direction and weight. In Phraseology in Foreign Language
344 Bibliography

Learning and Teaching, Fanny Meunier, and Sylviane Granger

(eds.), 43–66. Amsterdam/Philadelphia: Benjamins.
Hansen, Barbara, Klaus Hansen, Albrecht Neubert, and Manfred Schentke
2
1985 Englische Lexikologie: Einführung in Wortbildung und Semantik.
Leipzig: Enzyklopädie.
Hansen, Klaus, Uwe Carls, and Peter Lucko
1996 Die Differenzierung des Englischen in nationale Varianten. Berlin:
Schmidt.
Harris, Zellig
1957 Co-occurrence and transformation. In Language 33.3, 283–340.
[Reprinted in Zellig S. Harris: Papers on Syntax, Henry Hiż (ed.),
143–210. Dordrecht/Boston/London: Reidel Publishing Company,
1981.]
Hasan, Ruqaiya
1984 Coherence and cohesive harmony. In Understanding Reading Com-
prehension, J. Flood (ed.), 181–219. Newark: International Reading
Association.
Haugen, Einar
1966 Dialect, language, nation. In American Anthropologist 68: 922–935.
[Reprinted in Sociolinguistics, J.B. Pride, and Janet Holmes (eds.),
97–111. Harmondsworth: Penguin, 1972]
Hausmann, Franz-Josef
1984 Wortschatzlernen ist Kollokationslernen. Praxis des neusprachli-
chen Unterrichts 31: 395–406.
Hausmann, Franz-Josef
1985 Kollokationen im deutschen Wörterbuch: Ein Beitrag zur Theorie
des lexikographischen Beispiels. In Lexikographie und Grammatik,
Henning Bergenholtz, and J. Mugdan (eds.), 118–129. Tübingen:
Niemeyer.
Hausmann, Franz-Josef
2004 Was sind eigentlich Kollokationen? In Wortverbindungen – mehr
oder weniger fest, Kathrin Steyer (ed.), 309–334. Berlin: de Gruyter.
Hausser, Roland
1999 Foundations of Computational Linguistics: Man-Machine Commu-
nication in Natural Language. Berlin: Springer.
Helbig, Gerhard
1970 Geschichte der neueren Sprachwissenschaft. Leipzig: VEB Biblio-
graphisches Institut.
Helbig, Gerhard
1992 Probleme der Valenz- und Kasustheorie. Tübingen: Niemeyer.
 Bibliography 345

Herbst, Thomas
1983 Untersuchungen zur Valenz englischer Adjektive und ihrer Nomina-
lisierungen. Tübingen: Narr.
Herbst, Thomas
1992 Pro-Nunciation: Zur Bedeutung einer guten Aussprache in der
Fremdsprache. Die Neueren Sprachen 91 (1): 2–18.
Herbst, Thomas
1994 Linguistische Aspekte der Synchronisation von Fernsehserien: Pho-
netik, Textlinguistik, Übersetzungstheorie. Tübingen: Niemeyer.
Herbst, Thomas
1996 What are collocations: Sandy beaches or false teeth? English Studies
77 (4): 379–393.
Herbst, Thomas
2005 Englische Grammatik ist nicht so kompliziert: Pro Minimalismus,
Lücke und Polysemie – Kontra Prototypik und Semantik in gramma-
tischer Terminologie. In Linguistische Dimensionen des Fremdspra-
chenunterrichts, Thomas Herbst (ed.), 11–28. Würzburg: Königs-
hausen und Neumann.
Herbst, Thomas
2007 Valency complements or valency patterns? In Valency: Theoretical,
Descriptive and Cognitive Issues, Thomas Herbst, and Katrin Götz-
Votteler (eds.), 15–35. Berlin/New York: Mouton de Gruyter.
Herbst, Thomas
2009a Valency: Item-specificity and idiom principle. In Exploring the
Lexis-Grammar Interface, Ute Römer, and Rainer Schulze (eds.),
49–68. Amsterdam/Philadelphia: John Benjamins.
Herbst, Thomas
2009b Introduction to the Erlangen Valency Patternbank. www.pattern-
bank.uni-erlangen.de
Herbst, Thomas
2010 Valency constructions and clause constructions or how, if at all,
valency grammarians might sneeze the foam off the cappuccino. In
Cognitive Foundations of Linguistic Usage Patterns: Empirical
Studies, Hans-Jörg Schmid, and Susanne Handl (eds.), 225–255.
Berlin/New York: de Gruyter Mouton.
Herbst, Thomas, and Susen Faulhaber
forthc. Optionen der Valenzbeschreibung: Ein Valenzmodell für das Engli-
sche. In Grammar and Corpora III: Conference Proceedings.
Tübingen: Gunter Narr.
Herbst, Thomas, David Heath, and Hans-Martin Dederding
1980 Grimm’s Grandchildren: Current Topics in German Linguistics.
London: Longman.
346 Bibliography

Herbst, Thomas, and Michael Klotz

2003 Lexikografie. Paderborn: Schöningh (UTB).
Herbst, Thomas, and Brigitta Mittmann
2003 How can American English and British English be recognised? –
Arguments in favour of shifting the focus in the study of language
varieties toward(s) use and perception. In Cultural Encounters in the
New World: Literatur- und kulturwissenschaftliche Beiträge zu kul-
turellen Begegnungen in der Neuen Welt, Harald Zapf, and Klaus
Lösch (eds.), 127–150. Tübingen: Narr.
Herbst, Thomas, and Susen Schüller
2008 Introduction to Syntactic Analysis: A Valency Approach. Tübingen:
Narr.
Herbst, Thomas, Rita Stoll, und Rudolf Westermayr
1991 Terminologie der Sprachbeschreibung. Ismaning: Hueber.
Heringer, Hans Jürgen
1970 Theorie der deutschen Syntax. München: Hueber.
Heringer, Hans Jürgen
1993 Basic ideas and the classical model. In Syntax: Ein internationales
Handbuch zeitgenössischer Forschung. Vol. 1, Joachim Jacobs, Ar-
min von Stechow, Wolfgang Sternefeld, and Theo Vennemann
(eds.), 298–316. Berlin/New York: Walter de Gruyter.
Heringer, Hans Jürgen
1996 Deutsche Syntax dependentiell. Tübingen: Stauffenburg.
Heringer, Hans Jürgen, Bruno Strecker, and Rainer Wimmer
1980 Syntax: Fragen – Lösungen – Alternativen. München: Fink (UTB).
Hjelmslev, Louis
1961 Prolegomena to a Theory of Language, transl. by Francis J. Whit-
field. Madison: The University of Wisconsin Press. First edition
1943.
Hjelmslev, Louis
1974 Prolegomena zu einer Sprachtheorie. München: Hueber.
Hock, Hans Heinrich
2
1991 Principles of Historical Linguistics. Berlin/New York: Mouton de
Gruyter.
Hockett, Charles
1954 Two models of grammatical description. In Word 10: 210–231.
Hockett, Charles
1958 A Course in Modern Linguistics. New York/London: Holt, Rinehart
& Winston. [Excerpts in Sprachwissenschaft: Ein Reader, Ludger
Hoffmann (ed.). Berlin/New York: Walter de Gruyter, 22000.]
 Bibliography 347

Hoffmann, Ludger (ed.)

2
2000a Sprachwissenschaft: Ein Reader. Berlin/New York: Walter de Gruy-
ter.
Hoffmann, Ludger
2
2000b Thema und Rhema. In Sprachwissenschaft: Ein Reader, Ludger
Hoffmann (ed.), 598–612. Berlin/New York: Walter de Gruyter.
Hoffmann, Sebastian
2004 Using the OED quotations database as a corpus: A linguistic ap-
praisal. ICAME Journal 28: 17–30.
Hoffmann, Sebastian, Stefan Evert, Nicholas Smith, David Lee, and Ylva Ber-
glund Prytz
2008 Corpus Linguistics with BNCweb: A Practical Guide. Frankfurt:
Lang.
Hogg, Richard
1992 Phonology and morphology. In The Cambridge History of the Eng-
lish Language. Vol. I: The Beginnings to 1066, R. Hogg (ed.), 67–
167. Cambridge: Cambridge University Press.
Holmes, Janet
1992 An Introduction to Sociolinguistics. London/New York: Longman.
Hopper, Paul
1987 Emergent Grammar. Berkeley Linguistics Society 13: 39–157.
Hopper, Paul, and Elizabeth Closs Traugott
2
2003 Grammaticalization. Cambridge: Cambridge University Press.
House, Juliane
2002 Communicating in English as a lingua franca. Eurosla Yearbook 2:
243–261.
Huber, Markus
2008 Ghanaian English: Phonology. In Varieties of English 4: Africa,
South and Southeast Asia, Rajend Mesthrie (ed.), 67–92. Berlin/New
York: Mouton de Gruyter.
Huddleston, Rodney
1984 Introduction to the Grammar of English. Cambridge: Cambridge
University Press.
Huddleston, Rodney, and Geoffrey K. Pullum (eds.)
2002 The Cambridge Grammar of the English Language. Cambridge:
Cambridge University Press. [CamG]
Hudson, Richard
1976 Arguments for a Non-Transformational Grammar. Chicago/London:
University of Chicago Press.
Hudson, Richard
1980 Sociolinguistics. Cambridge: Cambridge University Press.
348 Bibliography

Hudson, Richard
1984 Word Grammar. Oxford: Blackwell.
Hudson, Richard
1993 Do we have heads in our minds? In Heads in Grammatical Theory,
Greville G. Corbett, Norman M. Fraser, and Scott McGlashan (eds.),
266–291. Cambridge: Cambridge University Press.
Hundt, Markus
1992 Einstellungen gegenüber dialektal gefärbter Standardsprache. Stutt-
gart: Steiner Verlag.
Hunston, Susan and, Gill Francis
2000 Pattern Grammar: A Corpus-Driven Approach to the Lexical
Grammar of English. Amsterdam: John Benjamins.
Jackendoff, Ray
2
1997 The Architecture of the Language Faculty. Cambridge,
Mass./London: MIT Press.
Jakobson, Roman, C. Gunnar Fant, and Morris Halle
1951 Preliminaries to Speech Analysis: The Distinctive Features and their
Correlates. Cambridge, Mass.: MIT Press.
Jespersen, Otto
1909 A Modern English Grammar on Historical Principles. Part I:
Sounds and Spellings. Heidelberg: Carl Winter’s Universitätsbuch-
handlung.
Johansson, Stig
1991 Times change, and so do corpora. In English Corpus Linguistics,
Karin Aijmer, and Bengt Altenberg (eds.), 305–314. London/New
York: Longman.
Jones, Daniel
9
1960 An Outline of English Phonetics. Cambridge: Heffer and Sons. First
edition 1918.
Jones, Daniel
1973 Some thoughts on the phoneme. In Phonetics in Linguistics: A Book
of Readings, William Eric Jones and John Laver (eds.), 168–179.
London/New York: Longman. First edition 1944.
Kaplan, R. M., and Joan Bresnan
1982 Lexical-functional grammar: A formal system for grammatical rep-
resentation. In The Mental Representation of Grammatical Rela-
tions, Joan Bresnan (ed.), 173–281. Cambridge, Mass.: MIT Press.
Kastovsky, Dieter
1982 Wortbildung und Semantik. Düsseldorf/Bern/München: Schwann-
Bagel, Francke.
 Bibliography 349

Kastovsky, Dieter
1992 Semantics and vocabulary. In The Cambridge History of English
Language. Vol. I: The Beginnings to 1066. R. Hogg (ed.), 290–408.
Cambridge: Cambridge University Press.
Kastovsky, Dieter
2006 Vocabulary. In A History of the English Language, Richard Hogg,
and David Denison (eds.), 199–270. Cambridge: Cambridge Univer-
sity Press.
Katz J. J., and J. A. Fodor
1963 The structure of a semantic theory. Language 39: 170–210.
Kay, Paul
2005 Argument structure constructions and the argument-adjunct distinc-
tion. In Grammatical Constructions: Back to the Roots, Mirjam
Fried, and Hans C. Boas (eds.), 71–98. Amsterdam: Benjamins.
Klann-Delius, Gisela
1999 Spracherwerb. Stuttgart: Metzler.
Klotz, Michael
2000 Grammatik und Lexik: Studien zur Syntagmatik englischer Verben.
ZAA Studies. Tübingen: Stauffenburg.
König, Ekkehard, and Volker Gast
2007 Understanding English-German Contrasts. Berlin: Schmidt.
Kohler, Klaus J.
1977 Einführung in die Phonetik des Deutschen. Berlin: Schmidt.
Kortmann, Bernd
2005 English Linguistics: Essentials (English edition). Berlin: Cornelsen.
Kortmann, Bernd
2008 Synopsis: Morphological and syntactic variation in the British Isles.
In Varieties of English 1: The British Isles, Bernd Kortmann, and
Clive Upton (eds.), 478–495. Berlin/New York: Mouton de Gruyter.
Krug, Manfred G.
2000 Emerging English Modals: A Corpus-Based Study of Grammaticali-
zation. Berlin/New York: Mouton de Gruyter.
Labov, William
1966 The Social Stratification of English in New York City. Washington:
Center for Applied Linguistics.
Lakoff, George
1987 Women, Fire, and Dangerous Things: What Categories Reveal
about the Mind. Chicago/London: University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Volume 1: Theoretical Prereq-
uisites. Stanford, CA: Stanford University Press.
350 Bibliography

Lass, Roger
1984 Phonology: An Introduction to Basic Concepts. Cambridge: Cam-
bridge University Press.
Lass, Roger
1992 Phonology and morphology. In The Cambridge History of the Eng-
lish Language. Vol. II: 1066–1476, Norman Blake (ed.), 23–155.
Cambridge: Cambridge University Press.
Lass, Roger
1994 Old English: A Historical Linguistic Companion. Cambridge: Cam-
bridge University Press.
Lass, Roger
1999 Phonology and morphology. In The Cambridge History of the Eng-
lish Language. Vol. III: 1476–1776, R. Lass (ed.), 56–186. Cam-
bridge: Cambridge University Press.
Lass, Roger
2006 Phonology and morphology. In A History of the English Language,
Richard Hogg, and David Denison (eds.), 43–108. Cambridge: Cam-
bridge University Press.
Lea, Diana
2007 Making a collocations dictionary. Zeitschrift für Anglistik und Ame-
rikanistik 55 (3): 261–271.
Leech, Geoffrey
2
1981 Semantics. Harmondsworth: Penguin.
Leech, Geoffrey
1983 Principles of Pragmatics. London/New York: Longman.
Leech, Geoffrey
1991 The state of the art in corpus linguistics. In English Corpus Linguis-
tics, Karin Aijmer, and Bengt Altenberg (eds.), 8–29. London/New
York: Longman.
Leech, Geoffrey
1992 Corpora and theories of linguistic performance. In Directions in
Corpus Linguistics: Proceedings of Nobel Symposium 82, Jan Svart-
vik (ed.), 105–122. Berlin/New York: Mouton de Gruyter.
Leech, Geoffrey, and Jan Svartvik
2
1994 A Communicative Grammar of English. London/New York: Long-
man.
Leech, Geoffrey, Marianne Hundt, Christian Mair, and Nicholas Smith
2009 Change in Contemporary English: A Grammatical Study. Cam-
bridge: Cambridge University Press.
Lees, Robert B.
5
1968 The Grammar of English Nominalizations. Bloomington: Mouton &
Co. First edition 1960.
 Bibliography 351

Lehmann, Hans Martin, Peter Schneider, and Sebastian Hoffmann

2000 BNCweb. In Corpora Galore. Analyses and Techniques in Describ-
ing English, J. Kirk, (ed.), 259–266. Amsterdam/Atlanta: Rodopi.
Leisi, Ernst
5
1969 Das heutige Englisch: Wesenszüge und Probleme. Heidelberg: Win-
ter. Revised edition: Ernst Leisi, and Christian Mair (81999): Heidel-
berg: Winter.
Leisi, Ernst
1973 Praxis der englischen Semantik. Heidelberg: Winter.
Lenker, Ursula
2010 Argument and rhetoric: Adverbial connectors in the history of Eng-
lish. Berlin/New York: De Gruyter Mouton.
Levinson, Stephen C.
1983 Pragmatics. Cambridge: Cambridge University Press.
Lieven, Elena
2008 Learning the English auxiliary: A usage-based approach. In Trends
in Corpus Research (Finding Structure in Data) TILAR Series. Vol.
6, Heike Behrens (ed.), 61–98. Amsterdam/Philadelphia: Benjamins.
Lieven, Elena, Heike Behrens, J. Speares, and Michael Tomasello
2003 Early syntactic creativity: a usage-based approach. Journal of Child
Language 30: 333–370.
Lightfood, David
1988 Syntactic change. In Linguistics: The Cambridge Survey. Vol. I.
Linguistic Theory: Foundations, Frederick J. Newmeyer (ed.), 303–
323. Cambridge: Cambridge University Press.
Lipka, Leonhard
1976 Topicalization, case grammar, and lexical decomposition in English.
Archivum Linguisticum VII (2): 118–141.
Lipka, Leonhard
3
2002 English Lexicology. Tübingen: Narr.
Lorenz, Gunter
1999 Adjective Intensification – Learners versus Native Speakers: A Cor-
pus Study of Argumentative Writing. Amsterdam/Atlanta: Rodopi.
Luick, Karl
1903 Studien zur englischen Lautgeschichte. Wien et al.: Braumüller.
Luick, Karl
1940 Historische Grammatik der englischen Sprache. Vol. 1. II. Abtei-
lung. Leipzig: Tauchnitz.
Lutz, Angelika
2002a Sprachmischung in der deutschen und englischen Wortbildung. In
Historische Wortbildung des Deutschen, Mechthild Habermann, Pe-
352 Bibliography

ter O. Müller, and Horst Haider Munske (eds.), 407–437. Tübingen:

Niemeyer.
Lutz, Angelika
2002b When did English begin? In Sounds, Words, Texts and Change,
Teresa Fanego, Belén Méntez-Naya, and Elena Seoane (eds.), 145–
171. Amsterdam/Philadelphia: Benjamins.
Lutz, Angelika
2004 The first push: A prelude to the Great Vowel Shift. Anglia 122 (2):
209–224.
Lutz, Angelika
2009 Celtic influence on Old English and West Germanic. English Lan-
guage and Linguistics 13 (2): 227–249.
Lyons, John
1968 Introduction to Theoretical Linguistics. Cambridge: Cambridge
University Press.
Lyons, John
1977 Semantics I. Cambridge: Cambridge University Press.
Lyons, John
3
1991 Chomsky. London: Fontana.
Maclay, Howard
1971 Overview. In Semantics: An Interdisciplinary Reader in Philosophy,
Linguistics and Psychology, Danny D. Steinberg, and Leon A. Jako-
bovits (eds.), 157–182. Cambridge: Cambridge University Press.
MacWhinney, Brian
1998 Models of the Emergence of Language. Annual Review of Psychol-
ogy 49: 199–227.
Mair, Christian
1997 Einführung in die anglistische Sprachwissenschaft. Darmstadt: Wis-
senschaftliche Buchgesellschaft.
Mair, Christian
2008 English Linguistics: An Introduction. Tübingen: Narr.
Marchand, Hans
2
1969 The Categories and Types of Present-Day English Word-Formation:
A Synchronic-Diachronic Approach. München: Beck.
Mathesius, Vilém
1975 A Functional Analysis of Present Day English on a General Linguis-
tic Basis. The Hague/Paris: Mouton.
Matthews, Peter H.
1974 Morphology: An Introduction to the Theory of Word-Structure.
Cambridge: Cambridge University Press.
Matthews, Peter H.
1981 Syntax. Cambridge: Cambridge University Press.
 Bibliography 353

Matthews, Peter
1982 Do Languages Obey General Laws? Cambridge: Cambridge Uni-
versity Press.
Matthews, Peter
1990 Language as mental faculty. In An Encyclopaedia of Language, N.
E. Collinge (ed.), 112–138. London: Routledge.
Matthews, Peter H.
1993 Grammatical Theory in the United States from Bloomfield to Chom-
sky. Cambridge: Cambridge University Press.
Matthews, Peter H.
2007 Syntactic Relations. A critical survey. Cambridge: Cambridge Uni-
versity Press.
McQueen, James M., and A. Cutler
1998 Morphology in word recognition. In The Handbook of Morphology,
Andrew Spencer, and Arnold M. Zwicky (eds.), 406–427. Ox-
ford//Malden: Blackwell.
Mey, Jacob L.
1993 Pragmatics: An Introduction. Oxford, UK/Cambridge, Mass.:
Blackwell.
Meyer, Matthias L. G.
2009 Revisiting the evidence for objects in English. In Exploring the
Lexis-Grammar Interface, Ute Römer, and Rainer Schulze (eds.),
211–227. Amsterdam/Philadelphia: John Benjamins.
Mindt, Dieter
1988 EDV in der Angewandten Linguistik: Ziele – Methoden – Ergebnis-
se. Frankfurt: Diesterweg.
Mittmann, Brigitta
2004 Mehrwort-Cluster in der englischen Alltagskonversation: Unter-
schiede zwischen britischem und amerikanischem gesprochenem
Englisch als Indikatoren für den präfabrizierten Charakter der
Sprache. Tübingen: Narr.
Moon, Rosamund
1987 The analysis of meaning. In Looking Up, John Sinclair (ed.), 86–
103. London/Glasgow: Collins.
Mugglestone, Lynda
2009 The Oxford English Dictionary. In The Oxford History of English
Lexicography. Vol. I., A. P. Cowie (ed.), 230–259. Oxford: Oxford
University Press.
Mukherjee, Joybrato
2002 Korpuslinguistik und Englischunterricht: Eine Einführung. Frank-
furt: Lang.
354 Bibliography

Mukherjee, Joybrato
2009 Anglistische Korpuslinguistik. Berlin: Schmidt.
Nesselhauf, Nadja
2004 Learner corpora and their potential for language teaching. In How to
Use Corpora in Language Teaching, John Sinclair (ed.), 125–152.
Amsterdam/Philadelphia: Benjamins.
Nesselhauf, Nadja
2005 Collocations in a Learner Corpus. Amsterdam: John Benjamins.
Neuhaus, Hans Joachim
1971 Beschränkungen in der Grammatik der Wortableitungen im Engli-
schen. Dissertation: Saarbrücken.
Newmeyer, Frederick J.
2001 The Prague School and North American functionalist approaches to
syntax. Journal of Linguistics 37: 101–126.
Noël, Dirk
2007 Verb valency patterns, constructions and grammaticalization. In
Valency: Theoretical, Descriptive and Cognitive Issues, Thomas
Herbst, and Katrin Götz-Votteler (eds.), 67–83. Berlin/New York:
Mouton de Gruyter.
Ogden, Charles K., and I. A. Richards
10
1956 The Meaning of Meaning. London: Routledge & Kegan Paul Ltd.
First edition 1923.
Palmer, Frank
1971 Grammar. Harmondsworth: Penguin.
Palmer, Frank
2
1981 Semantics. Cambridge: Cambridge University Press.
Palmer, Frank
1994 Grammatical Roles and Relations. Cambridge: Cambridge Univer-
sity Press.
Paquot, Magali
2009 Exemplification in learner writing: A cross-linguistic perspective. In
Phraseology in Foreign Language Learning and Teaching, Fanny
Meunier, and Sylviane Granger (eds.), 101–119. Amster-
dam/Philadelphia: Benjamins.
Pike, Kenneth L.
1943 Phonetics: A Critical Analysis of Phonetic Theory and a Technic for
the Practical Description of Sounds. Ann Arbor: The University of
Michigan Press.
Pinker, Steven
1991 Rules of language. Science 253: 530–535.
 Bibliography 355

Pinker, Steven, and Alan Prince

1994 Regular and irregular morphology and the psychological status of
rules of grammar. In The Reality of Linguistic Rules, S. D. Lima, R.
L. Corrigan, and G. K. Iverson (eds.), 321–351. Amsterdam: Benja-
mins.
Plag, Ingo
2003 Word-Formation in English. Cambridge: Cambridge University
Press.
Plag, Ingo, Maria Braun, Sabine Lappe, and Mareile Schramm
2007 Introduction to English Linguistics. Berlin/New York: Mouton de
Gruyter.
Prince, Ellen F.
1981 Toward a taxonomy of given-new information. In Radical Pragma-
tics, Peter Cole (ed.), 223–255. New York/London: Academic Press.
Pullum, Geoffrey K.
2009 Lexical categorization in English dictionaries and traditional gram-
mars. In Classification in Linguistics and Lexicography. Zeitschrift
für Anglistik und Amerikanistik 57 (3): 255–273.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik
1972 A Grammar of Contemporary English. London: Longman. [GCE]
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik
1985 The Comprehensive Grammar of the English Language. Lon-
don/New York: Longman. [CGEL]
Quirk, Randolph, and C. L. Wrenn
2
1957 An Old English Grammar. London: Methuen.
Radford, Andrew
1993 Head-hunting: On the trail of the nominal Janus. In Heads in Gram-
matical Theory, Greville G. Corbett, Norman M. Fraser, and Scott
McClashan (eds.), 73–113. Cambridge: Cambridge University Press.
Roach, Peter
4
2009 English Phonetics and Phonology: A Practical Course. Cambridge:
Cambridge University Press.
Robins, Robert H.
1967 A Short History of Linguistics. London and Harlow: Longman.
Robins, Robert H.
1971 General Linguistics: An Introductory Survey. London: Longman.
Robins, Robert H.
2000 The development of the word class system of the European gram-
matical tradition. Foundations of Language 2: 3–19. First published
in 1966. [In Sprachwissenschaft: Ein Reader, Ludger Hoffmann
(ed.), 452–469. Berlin/New York: Walter de Gruyter, 2000.]
356 Bibliography

Rosch Heider, Eleanor

1971 ‘Focal’ color areas and the development of color names. Develop-
mental Psychology 4 (3): 447–455.
Rosch Heider, Eleanor
1972 Universals in color naming and memory. Journal of Experimental
Psychology 93 (1): 10–20.
Rosch, Eleanor
1973 On the Internal Structure of Perceptual and Semantic Categories. In
Cognitive Development and the Acquisition of Language, Timothy E
Moore (ed.), 111-144. New York/San Franciso/London: Academic
Press.
Rosch, Eleanor
1975 Cognitive Representations of Semantic Categories. Journal of Ex-
perimental Psychology 104 (3): 192–233.
Russell, Bertrand
1919 Introduction to Mathematical Philosophy. London: George Allen
and Unwin.
Sampson, Geoffrey
1980 Schools of Linguistics. Stanford: Stanford University Press.
Sampson, Geoffrey
2001 Empirical Linguistics. London/New York: Continuum.
Saussure, Ferdinand de
1916 Cours de linguistique générale, Charles Bally, and Albert Séchehaye
(eds.). Paris/Lausanne: Payot.
Saussure, Ferdinand de
1964 Lettres de Ferdinand de Saussure à Antoine Meillet (4 janvier 1894).
In Société Genevoise de Linguistique, Cahiers Ferdinand de Saus-
sure, 21, 1964, Genève: Librairie Droz: 93–130. First edition 1894.
Saussure, Ferdinand de
1983 Course in General Linguistics, translated and annotated by Roy
Harris, London: Duckworth.
Scheler, Manfred
1977 Der englische Wortschatz. Berlin: Schmidt.
Scherer, Günther, and Alfred Wollmann
1972 Englische Phonetik und Phonologie. Berlin: Schmidt.
Schmid, Hans-Jörg
2005 Englische Morphologie und Wortbildung: Eine Einführung. Berlin:
Schmidt.
Schmid, Hans-Jörg
2008 New words in the mind: Concept-formation and entrenchment of
neologisms. Anglia 126: 1–36.
 Bibliography 357

Schmid, Hans-Jörg
forthc. English Morphology and Word-Formation. Berlin: Schmidt.
Schmied, Josef
1991 English in Africa. London/New York: Longman.
Schmied, Josef
2008 East African English (Kenya, Uganda, Tanzania): Phonology. In
Varieties of English 4 Africa: South and Southeast Asia, Rajend
Mesthrie (ed.), 150–163. Berlin/New York: Mouton de Gruyter.
Schneider, Edgar
2007 Postcolonial English: Varieties around the World. Cambridge: Cam-
bridge University Press.
Schneider, Edgar
2008a Synopsis: Morphological and syntactic variation in the Americas and
the Caribbean. In Varieties of English 2: The Americas and the Car-
ibbean, Edgar Schneider (ed.), 763–776. Berlin/New York: Mouton
de Gruyter.
Schneider, Edgar
2008b List of features: Phonology and phonetics. In Varieties of English 1:
The British Isles, Bernd Kortmann, and Clive Upton (eds.), xix–
xxiv. Berlin/New York: Mouton de Gruyter.
Schröder, Konrad
2005 Einige unmaßgebliche Gedanken zur grammatischen Terminologie
und zum Grammatikunterricht des Schulfaches Englisch: A some-
what personal account. In Linguistische Dimensionen des Fremd-
sprachenunterrichts, Thomas Herbst (ed.), 1–9. Würzburg: Königs-
hausen und Neumann.
Schubert, Christoph
2008 Englische Textlinguistik: Eine Einführung. Berlin: Schmidt Verlag.
Searle, John R.
1969 Speech Acts: An Essay in the Philosophy of Language. Cambridge:
Cambridge University Press.
Searle, John R.
1971 What is a speech act? In The Philosophy of Language, John R.
Searle (ed.), 39–53. London: Oxford University Press.
Searle, John R.
1979a A taxonomy of illocutionary acts. In Expression and Meaning: Stud-
ies in the Theory of Speech Acts, John R. Searle, 1–29. Cambridge:
Cambridge University Press.
Searle, John R.
1979b Indirect speech acts. In Expression and Meaning: Studies in the
Theory of Speech Acts, John R. Searle, 30–57. Cambridge: Cam-
bridge University Press.
358 Bibliography

Sells, Peter
1985 Lectures on Contemporary Syntactic Theories. Stanford: Center for
the Study of Language and Information, Stanford University.
Siepmann, Dirk
2005 Collocation, colligation and encoding dictionaries. Part I: Lexico-
logical aspects. International Journal of Lexicography 18 (4): 409–
444.
Siepmann, Dirk
2007 Collocations and examples: Their relationship and treatment in a
new corpus-based learner’s dictionary. In Zeitschrift für Anglistik
und Amerikanistik 55 (3): 235–260.
Siepmann, Dirk
2008 Phraseology in learners’ dictionaries: What, where and how? In
Phraseology in Foreign Language Learning and Teaching, Fanny
Meunier, and Sylviane Granger (eds.), 185–202. Amster-
dam/Philadelphia: Benjamins.
Sinclair, John (ed.)
1990 Collins Cobuild English Grammar. London/Glasgow: Collins.
Sinclair, John
1991 Corpus, Concordance, Collocation. Oxford: Oxford University
Press.
Sinclair, John
1994 Trust the text. In Advances in Written Text Analysis, Malcolm
Coulthard (ed.), 12–25 London/New York: Routledge.
Sinclair, John
2004 Trust the Text. London/New York: Routledge.
Sinclair, John, and Anna Mauranen
2006 Linear Unit Grammar: Integrating Speech and Writing. Amster-
dam/Philadelphia: Benjamins.
Söll, Ludwig, and Franz-Josef Hausmann
3
1985 Gesprochenes und geschriebenes Französisch. Berlin: Schmidt.
Somers, Harold L.
1987 Valency and Case in Computational Linguistics. Edinburgh: Edin-
burgh University Press.
Stefanowitsch, Anatol
2007 Linguistics beyond grammaticality. Corpus Linguistics and Linguis-
tic Theory 3 (1): 57–71.
Stefanowistch, Anatol
2008 Negative entrenchment: A usage-based approach to negative evi-
dence. Cognitive Linguistics 19 (3): 513–531.
 Bibliography 359

Stefanowitsch, Anatol, and Stefan Th. Gries

2003 Collostructions: Investigating the interaction between words and
constructions. International Journal of Corpus Linguistics 8 (2):
209–243.
Steinberg, Danny D.
1982 Psycholinguistics: Language, Mind and World. London/New York:
Longman.
Stockwell, Robert
2002 How much shifting actually occurred in the historical English vowel
shift? In Studies in the History of the English Language: A Millen-
nial Perspective, Donka Minkova, and Robert Stockwell (eds.), 267–
281. Berlin/New York: Mouton de Gruyter.
Strang, Barbara
1970 A History of English. London: Methuen.
Stubbs, Michael
1996 Text and Corpus Analysis. Oxford/Cambridge, Mass.: Blackwell.
Stubbs, Michael
2009 Technology and phraseology: With notes on the history of corpus
linguistics. In Exploring the Lexis-Grammar Interface, Ute Römer,
and Rainer Schulze (eds.), 15–31. Amsterdam/ Philadelphia: John
Benjamins.
Taylor, John R.
3
2003 Linguistic Categorization. Oxford: Oxford University Press.
Tesnière, Lucien
1959 Éléments de syntax structurale. Paris: Klincksieck. [Excerpts in
Sprachwissenschaft: Ein Reader, Ludger Hoffmann (ed.). Ber-
lin/New York: Walter de Gruyter, 22000.].
Tomasello, Michael
2003 Constructing a Language. Cambridge, Mass./London: Harvard Uni-
versity Press.
Traugott, Elizabeth Closs
1992 Syntax. In The Cambridge History of the English Language. Vol. I,
Richard M. Hogg (ed.), 168–289. Cambridge: Cambridge University
Press.
Traugott, Elizabeth Closs
2002 From etymology to historical pragmatics. In Studies in the History of
the English Language: A Millennial Perspective, Donka Minkova,
and Robert Stockwell (eds.), 19–49. Berlin/New York: Mouton de
Gruyter.
Trier, Jost
1934 Das sprachliche Feld: Eine Auseinandersetzung. Neue Jahrbücher
für Wissenschaft und Jugendbildung 10: 428–429.
360 Bibliography

Trudgill, Peter
1974 The Social Differentiation of English in Norwich. London: Cam-
bridge University Press.
Trudgill, Peter
1975 Accent, Dialect and the School. London: Arnold.
Trudgill, Peter (ed.)
1978 Sociolinguistic Patterns of British English. London: Arnold.
Trudgill, Peter
1986 Dialects in Contact. Oxford: Blackwell.
Trudgill; Peter
3
1995 Sociolinguistics. Harmondsworth: Penguin. First edition 1974.
Trudgill, Peter, and Howard Giles
1978 Sociolinguistic and linguistic value judgements: Correctness, ade-
quacy and aesthetics. In Functional Studies in Language and Litera-
ture, P. Coppieters, and D. L. Goyvaerts (eds.), 167–190.
Gent/Antwerpen/Brüssel: E. Story Scientia p.V.B.A. Scientific Pub-
lishers.
Trudgill, Peter, and Jean Hannah
5
2008 International English: A Guide to Varieties of Standard English.
London: Hodder Education. First edition 1982, London: Arnold.
Ullmann, Stephen
1972 Semantics. Oxford: Basil Blackwell.
Ungerer, Friedrich
1999 Englische Grammatik heute. Stuttgart/Düsseldorf/Leipzig: Klett.
[EGH]
Ungerer, Friedrich, and Hans-Jörg Schmid
2
2006 An Introduction to Cognitive Linguistics. London: Pearson Long-
man.
Upton, Clive
2008 Received Pronunciation. In Varieties of English 1: The British Isles,
Bernd Kortmann, and Clive Upton (eds.), 237–252. Berlin/New
York: Mouton de Gruyter.
Vennemann, Theo
1977 Konstituenz und Dependenz in einigen neueren Grammatiktheorien.
Sprachwissenschaft 2: 259–301.
Vennemann, Theo
1998 Germania Semitica: +plōg-/+pleg-, +furh-/+farh-, +folk-/+flokk, +felh-
/+folg-. In Deutsche Grammatik: Thema in Variationen. Festschrift
für Hans-Werner Eroms zum 60. Geburstag, Karin Donhauser, Lud-
wig M. Eichinger (eds.), 245–261. Heidelberg: Winter.
 Bibliography 361

Vennemann, Theo, and J. Jakobs

2000 Sprache und Grammatik. Darmstadt: Wissenschaftliche Buchgesell-
schaft. First edition 1982. [Excerpts in Sprachwissenschaft: Ein Re-
ader, Ludger Hoffmann (ed.). Berlin/New York: Walter de Gruyter,
2
2000.]
Viereck, Wolfgang, Karin Viereck, and Heinrich Ramisch
2002 Dtv-Atlas Englische Sprache. München: dtv.
Weinreich, Uriel, William Labov, and Marvin I. Herzog
1968 Empirical foundations for a theory of language change. In Directions
for Historical Linguistics: A Symposium, W.P. Lehmann, and Yakov
Malkiel (eds.), 95–188. Austin/London: University of Texas Press.
Wells, John
1982 Accents of English, 3 vols. Cambridge: Cambridge University Press.
Werlich, Egon
1975 Typologie der Texte. Heidelberg: Quelle & Meyer.
Werlich Egon
1976 A Text Grammar of English. Heidelberg: Quelle & Meyer.
Wiik, Kalevi
1965 Finnish and English Vowels: A Comparison with Special Reference
to the Learning Problems Met by Native Speakers of Finnish Learn-
ing English. Turku: Annales Universitatis Turkuensis Series B/94:
Turun Yliopisto.
Wimmer, Heinz, and Josef Perner
1979 Kognitionspsychologie. Stuttgart: Kohlhammer.
Wotjak, Gerd
1971 Untersuchung zur Struktur der Bedeutung. München: Hueber.

Dictionaries:

A Dictionary of English Affixes: Their Function and Meaning

2007 by Gabriele Stein. München: Lincom.
Advanced Learner’s Dictionary of Current English
2
1963 by A. S. Hornby, edited by E.V. Gatenby, and H. Wakefield. Lon-
don: Oxford University Press. [OALD2]
A First Dictionary of Linguistics and Phonetics
1980 by David Crystal. London: André Deutsch.
Anglizismen-Wörterbuch: Der Einfluß des Englischen auf den deutschen Wort-
schatz nach 1945
1993 by Broder Carstensen [continued by Ulrich Busse]. Berlin/New
York: de Gruyter.
362 Bibliography

A Valency Dictionary of English

2004 by Thomas Herbst, David Heath, Ian Roe, and Dieter Götz. Ber-
lin/New York: Mouton de Gruyter. [VDE]
Cambridge English Pronouncing Dictionary
17
2006 by Daniel Jones. Edited by Peter Roach, James Hartman, and Jane
Setter. Cambridge: Cambridge University Press [EPD17].
Collins COBUILD English Language Dictionary
1987 edited by John Sinclair. London: Collins.
Duden Aussprachewörterbuch
4
2000 by M. Mangold, and Dudenredaktion. Mannheim: Dudenverlag.
Duden: Deutsches Universalwörterbuch
4
2001 by Wissenschaftlicher Rat der Dudenredaktion: Anette Klosa, Kath-
rin Kunkel-Razum, Werner Scholze-Stubenrecht, and Matthias
Wermke. Mannheim/Leipzig/Wien/Zürich: Dudenverlag.
Duden Oxford Großwörterbuch Englisch: Deutsch-Englisch / Englisch-Deutsch
2
1999 edited by Dudenredaktion, and Oxford University Press. Mannheim/
Leipzig/Wien/Zürich: Dudenverlag.
English Pronouncing Dictionary
15
1997 by Daniel Jones, edited by Peter Roach, James Hartman, and Jane
Setter. Cambridge/New York/Melbourne: Cambridge University
Press (see also Everyman’s English Pronouncing Dictionary and
Cambridge English Pronouncing Dictionary).
Erlangen Valency Patternbank
2009– by Thomas Herbst and Peter Uhrig. Available online at:
http://www.patternbank.uni-erlangen.de.
Everyman’s English Pronouncing Dictionary
13
1967 by Daniel Jones. London: Dent & Sons/New York: E.P. Dutton &
Co. [EPD13]
Everyman’s English Pronouncing Dictionary
14
1977 by Daniel Jones, edited by A. C. Gimson. London: Dent & Sons.
[EPD14]
Framenet http://framenet.icsi.berkeley.edu/
Longman Dictionary of Contemporary English
1978 edited by Paul Procter. Harlow: Longman. [LDOCE1]
Longman Dictionary of Contemporary English
4
2003 edited by Della Summers. Harlow: Longman. [LDOCE4]
Longman Dictionary of Contemporary English
5
2009 edited by Michael Mayor. Harlow: Pearson Longman. [LDOCE5]
Longman Pronunciation Dictionary
3
2008 by John Wells. Harlow: Pearson-Longman. First edition 1990.
[LPD3]
 Bibliography 363

New Shorter Oxford English Dictionary on Historical Principles

1993 edited by Lesley Brown. Oxford: Clarendon Press. [NSOED]
Oxford Advanced Learner’s Dictionary of Current English
6
2000 by A. S. Hornby, edited by Sally Wehmeier. Oxford: Oxford Uni-
versity Press. [OALD6]
Oxford Advanced Learner’s Dictionary of Current English
7
2005 by A. S. Hornby, edited by Sally Wehmeier. Oxford: Oxford Uni-
versity Press. [OALD7]
Oxford Collocations Dictionary for Students of English
2002 edited by Jonathan Crowther, Sheila Dignen, and Diana Lea. Ox-
ford: Oxford University Press.
Oxford Dictionary of Current Idiomatic English. Vol. 1: Verbs with Prepositions
and Particles
1974 by Anthony P. Cowie, and R. Mackin. London: Oxford University
Press.
Oxford Dictionary of Current Idiomatic English. Vol 2: Phrase, Clause and Sen-
tence Idioms
1983 edited by Anthony Paul Cowie, Ronald Mackin, and Isabel R.
McCaig. Oxford: Oxford University Press. [ODCIE2]
Oxford English Dictionary
2
1989 edited by John Simpson, and E. S. C. Weiner. Oxford: Clarendon.
[OED2]
VALBU – Valenzwörterbuch deutscher Verben
2004 by Helmut Schumacher, Jacqueline Kubczak, Renate Schmidt, and
Vera de Ruiter. Tübingen: Narr.
Wörterbuch zur Valenz und Distribution deutscher Verben
2
1973 by Gerhard Helbig, and Wolfgang Schenkel. Leipzig: Enzyklopädie.
First edition 1969.

Sources used:

AE Authentic example
BNC British National Corpus
HM http://www.henry-moore.org/pg/henry-moore-research/henry-
moores-life/1898–1925 (last accessed 14 April 2010).
NF 125 Jahre Reederei Norden Frisia. By Heinz Busching (1996). Nor-
den: Soltau-Kurier-Norden.
NW Nice Work. By David Lodge (1989). Harmondsworth: Penguin. First
published 1988.
PH Patrick Heron. By Mel Gooding (1994). London: Phaidon.
364 Bibliography

PW Painting the Warmth of the Sun. By Tom Cross (1995). Tiverton:

Westcountry Books/Cambridge: Lutterworth Press.
QE Quoted example
SP Pygmalion. By George Bernard Shaw (1916). Harmondsworth:
Penguin.
TI Tate St Ives. Tate Gallery St Ives. Barbara Hepworth Museum and
Sculptore Garden. An Illustrated Companion. By Michael Tooby
(1993). London: Tate Gallery Publications.
TS The Shining Sands. Artists in Newlyn and St Ives 1880–1930. By
Tom Cross (1994). Tiverton: Westcountry Books/Cambridge: Lut-
terworth Press.
VDE A Valency Dictionary of English (see bibliography).
WG Patrick Heron: Paintings 1970–1984. By James Beechey (2004).
Waddington Galleries London.
ZD Zennor in Darkness. By Helen Dunmore (1994). Harmondsworth:
Penguin.
Index

abbreviation 105 collocation 40, 129, 131–134, 139,

ablaut 91, 105, 206, 327 206, 293
accent 4, 55, 82, 119, 305–311 competence 28–32, 112–115, 200,
acoustic phonetics 45 209, 216, 219, 309
acronymisation 105, 108, 109, 112 complement 107, 147, 149, 150, 152,
adjective 40, 99, 101, 102, 109, 110, 153, 156, 175, 179, 184–191, 193,
150, 153, 155, 156, 158, 159, 162– 196–198, 214, 215, 227, 313
164, 187, 195, 253, 325 complementarity 245, 246, 251
adjunct 152, 184, 185, 193, 198 complementary distribution 57, 60,
adverbial 107, 135, 147, 152, 153, 184 85, 89, 317, 318
affixation 103, 105, 108 componential analysis 248–251, 260
affricate 50, 59 compounding 102–104, 108
allomorph 84, 87, 89, 90 concordance 37
allophone 57, 71, 79, 317 consonant 54, 58, 66, 69, 71, 81, 82,
alveolar 49, 50, 53, 59, 70 317, 321, 322
antonymy 245, 246 construction grammar 30, 42, 138,
approximant 50, 59, 77 139, 141, 191, 205, 207, 210–212,
argument structure construction 192, 217, 329
212, 214, 215, 218, 219 contextually optional 190
articulatory phonetics 46 conversational implicatures 270
aspiration 45, 51, 57, 58, 69, 79 converseness 246, 251, 252
assimilation 73, 75, 316 conversion 103, 105, 109–113
back vowel 51, 319 co-operative principle 268–270, 280
back-formation 104, 105, 111–113 corpus 9, 33–42, 125, 129, 130, 206
bilabial 49, 58, 59 cranberry morpheme 94
blending 108, 112 denotation 220, 223, 224, 228, 229,
blocked morpheme 94 231
blocking 114, 117, 122 dental 49, 53, 59
bound morpheme 85, 103, 217 determinative 155, 166, 169
cardinal vowel 51, 52, 61 determiner 96, 103, 110, 155, 166,
case grammar 107, 171, 176, 178, 167, 169, 170, 194, 195, 289
179, 183, 204, 282 diachronic 14, 16–18, 36, 93, 100,
clipping 109, 112 111, 235
coherence 283, 285–287, 291, 293, dialect 3, 4, 206, 303–305, 308
295 distinctive feature 58, 60, 61, 68, 249
cohesion 283–288, 291–293, 295 elision 73, 75, 316
366 Index

familiarity marker 105 lexicalization 108, 117, 119–121

foreign language teaching 4, 22, 23, loan word 8, 322
25, 40, 69, 131, 134, 136, 137 manner of articulation 50
fortis 51, 53, 60, 61, 66, 69, 71, 79 meaning 14–16, 19, 23, 56, 58, 67, 71,
free morpheme 85, 87, 102 81, 84, 92, 94, 98, 107, 119, 131,
fricative 50, 53, 59, 60, 70, 77 132, 134, 136, 152, 153, 176, 177,
front vowel 317, 319 204, 211, 213–215, 220–226, 229–
functional morpheme 85 240, 242, 247–253, 258–261, 263,
generative grammar 20, 28, 35, 201, 265–267, 288, 289, 294, 298, 312,
202, 204, 207, 210, 212 324, 326, 328
glottal 50, 59, 69, 70, 71, 82, 304 minimal pair 58, 60
gradation series 91 modifier 106, 108, 156, 299
grammatical morpheme 85–87 morph 87, 88, 90
grammaticalization 120, 328, 329 morpheme 67, 83–90, 92, 94, 97, 102,
Great Vowel Shift 6, 318–322 108, 115, 120, 143, 161, 220
head 102, 106, 108, 135, 154–156, morphological conditioning 90
167, 169, 191, 193–198, 226, 235, morphological process 86
276, 290 nasal 48–50, 53, 59, 321
homonymy 202, 233–238, 325 nasalized 49, 77, 78
hyponymy 240, 247, 293 New Englishes 4
idiolect 3 nonce formation 115, 118, 119
idiom 134, 136–138, 207, 212, 268 object 107, 142, 145, 147–153, 173,
idiom principle 136–138, 207, 212 176, 179, 184, 204, 214, 215, 326
i-mutation 91, 318, 319, 322 obligatory 124, 146, 152, 156, 184,
incompatibility 251 190
intonation 44, 54, 55, 80, 81, 230, open-choice principle 136, 137
297, 298 optional 75, 110, 122, 146, 152, 184,
intuition 27, 31, 32, 35, 72, 205 190, 290, 291
language acquisition 26, 28, 30, 31, oral 46, 48, 49, 53, 77, 301
39, 93, 112, 124, 170, 200, 208– palato-alveolar 50, 59
210, 215–219 paradigmatic 19, 20
langue 14, 29, 30 parole 14, 29, 30
lateral 46, 50, 59, 77 particle 135, 136, 160, 168, 169, 188,
lenis 51, 53, 60, 61, 69 197
lexeme 97, 98, 100, 101, 109, 118, performance 28–32, 39, 200, 201,
119, 127, 131, 136, 233–235, 239, 206, 274, 277, 279, 299
250, 293 phone 44, 48, 67–71, 74, 76, 79, 104,
lexical field 239, 247, 248, 293 233
lexical morpheme 85–87 phoneme 57–60, 63–65, 67–72, 74,
lexical unit 73, 98, 100, 121, 186, 188, 76, 78–82, 84, 90, 97, 220, 249,
191, 233–235, 237–240, 242, 243, 310, 316–318
247–249, 251–324 phoneme merger 318
 Index 367

phoneme shift 318 subject 107, 142, 144–149, 152, 153,

phoneme split 318 174, 176, 177, 184, 189, 191, 193,
phonological conditioning 67, 89 198, 204, 295
phonotactics 65 suffixation 104, 112–114, 162
phrase 102, 126, 127, 136, 143, 146, suprasegmental element 54
152–157, 162, 165–167, 169, 171, synchronic 14, 16–18, 31, 91–93, 111,
172, 174, 176, 179, 185, 187, 188, 113, 236, 237
194–198, 201, 203, 204, 215, 224, synonymy 241–243, 247, 293
226, 227, 288–290, 299 syntagmatic 19, 20, 251
place of articulation 49, 321 theme 213, 289, 295–297
plosive 50, 53, 59, 60, 66, 68, 69, 321 token 96, 216
polysemy 215, 234–238, 266, 325 type 96
portmanteau morph 87 unique morpheme 94
post-alveolar 50, 59 usage-based 30, 31, 124, 215, 217,
pragmatics 25, 228, 267, 268, 274, 219, 303
281 valency 20, 37, 110, 138, 139, 141,
predicate 144–147, 152, 155, 174, 146, 151, 168, 176, 178, 179, 183–
189, 192, 193, 198, 290, 295 193, 195, 197, 204, 206, 212–215,
prefixation 104, 114 218, 290, 291, 313
productivity 113, 114 variety 3, 4, 21, 22, 27, 34, 42, 49, 64,
reference 155, 220, 221, 224–228, 149, 217, 288, 293, 298, 303, 306–
262, 275, 276, 288, 289, 291, 292, 308, 312, 313, 315, 316
299 velar 50, 53, 59, 317
relational opposition 246 verb 20, 32, 39, 91, 103, 107, 109–
rheme 295, 296 113, 120, 135–138, 142, 145–148,
Sapir-Whorf-hypothesis 255, 256 151–153, 155, 156, 158–162, 164,
semantic role 148, 176, 178–180, 191, 165, 168–170, 173, 174, 176, 177,
198, 204, 326 180, 181, 183–191, 193, 197, 204,
semantics 19, 25, 188, 228, 229, 231, 212–215, 217, 218, 274, 290, 327,
234, 239, 241, 247, 249, 252, 256, 328
257, 266–268, 270, 285 voiced 47, 49, 53, 59–61, 68–70, 80,
semi-vowel 47, 74, 77 82, 85, 90, 317, 318
signifiant 14, 15, 220 voiceless 49, 53, 59–61, 68–70, 80,
signifié 14, 15, 220 81, 85, 90, 317
sound change 107, 318, 319 vowel 47, 51–54, 60–65, 68–71, 73–
speech act 25, 29, 140, 224, 228, 271, 75, 77–80, 82, 88, 90, 91, 309–
274, 277, 278, 280–282 310, 317, 319, 321, 322
standard 6, 11, 24, 61, 93, 238, 302, weak form 75, 82
303, 305–309 weak verb 91, 92, 322
stress 54, 55, 75, 102, 103, 120, 297, word class 12, 23, 95, 96, 105, 110,
298, 311 142, 147, 157–170, 187, 188, 197,
strong verb 91, 92, 327 198, 237, 247, 293
368 Index

word formation 25, 83, 94, 99–108, zero-derivation 103, 104, 109–111
110–115, 117–121, 131, 134, 136, zero-morph 88, 89, 104, 109
256, 293, 322 zero-morpheme 88, 104, 109