Dialect levelling: a two-dimensional process

Hinskens, F.L.M.P.
1998, Article / Letter to editor (Folia Linguistica, 32, 1/2, (1998), pp. 35-51)
Doi link to publisher: https://doi.org/10.1515/flin.1998.32.1-2.35

Version of the following full text: Publisher’s version

Downloaded from: http://hdl.handle.net/2066/139318
Download date: 2024-10-24

Note:
To cite this publication please use the final published version (if applicable).
Dialect Levelling:
A Two-dimensional Process*

Frans Hinskens

Abstract
Processes of dialect levelling reduce the linguistic 'autonomy' of individual dialects by leading to
their structural convergence with related varieties. This contribution presents research into proc-
esses of dialect levelling in a Limburg dialect of Dutch. After a concise overview of both the goal
of this paper and the definitions of some of the key notions (§ 1), the questions and related hy-
potheses guiding the research are presented (§2). A brief sketch of some of the most relevant
historical, socio-geographical and dialect-geographical aspects of the research area (§ 3) is fol-
lowed by an account of the considerations underlying the design of the study and an overview of
the main methodological Steps (§ 4). Some of the most important findings for each of the hy-
potheses are then illustrated and discussed. By far most attention is paid to the hypothesis central
in the present contribution, the one that says that dialect levelling does not necessarily lead to
structural convergence towards the Standard language. The findings and conclusions are com-
pared to those of related — older s well s contemporary — studies (§ 5). Finally, a follow-up
study is briefly sketched. This follow-up study was designed to deepen some of the insights and
to provide an answer to a question that emerged from the main finding relevant to the hypothesis
central in the present contribution (§6).

1. Introduction
In this paper I will demonstrate that the levelling out of cross-dialectal Variation
can be structurally independent of the levelling of Variation in the dialect-
standard language dimension. In doing so, I will also show how one research
project, and in particular some of its findings and conclusions, can lead to a new
research project.
As to terminology and definitions: dialect levelling is here defined s the
reduction of structural Variation — of both quantitative, internal Variation and
(either categorical or quantitative) differences between varieties of a language,
say, dialects. Dialect levelling thus makes (a) individual dialects more homoge-
neous, and (b) different dialects more similar and, consequently, diasystems
more homogeneous.
Conceptually, the relationship between language Variation, dialect levelling

Folia Linguistica ΧΧΧΠ/1-2 0165-4004/98/32-35 $ 2.-

(C) Mouton de Gruyter, Berlin — Societas Linguistica Europaea

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
36

and language change is äs follows. A process of language change that has not
come to completion in some respect leaves behind language Variation, either
internally (äs in e.g. lexically diffuse sound change) or between closely lan-
guage varieties (e.g. dialects or style levels). Variation, on the other hand, can
lead to linguistic change; especially quantitative internal Variation can be the
synchronic reflection of an ongoing process of language change. Now, dialect
levelling is the process which reduces language Variation. So dialect levelling is
a special type of language change.
In so far äs dialect levelling affects the differences between related language
varieties, it can lead to structural convergence. This can be convergence towards
surrounding dialects or towards the Standard variety — or both. Convergence
can be defined äs the becoming structurally more similar of languages or lan-
guage varieties, whereas divergence can be defined äs the becoming structurally
more dissimilar of languages or language varieties.

2. Research questions and hypotheses

In this contribution I will briefly sketch a research into processes of dialect lev-
elling. The aim of the research was to provide answers to three questions. Provi-
sional answers to these questions are given in the form of hypotheses which
have been tested quantitatively on the basis of fieldwork data.
question: hypothesis: Implementation:
\. what type of dialect levelling is a levelling can affect both Variation in
Variation does two-dimensional the dialect-standard language di-
it affect? process mension and Variation across related
dialects
More in particular: the levelling of differences between related dialects can be
independent from the levelling of Variation in the diaiect-standard language
dimension. In German dialectology the process of the levelling out of Variation
across dialects is sometimes referred to äs Ausgleich", the result of this process
can be the formation of what is sometimes called a 4 koine'. In German dialec-
tology, the process of the levelling out of Variation between a dialect and the
Standard variety which results from the convergence of the dialect towards the
Standard variety is referred to äs Abbau.
2. how does dia- dialect levelling is levelling is temporally, geographi-
lect levelling gradual in linguistic cally and internally gradual
proceed? äs well äs extra-
linguistic respects
3. why does dia- dialect levelling re- accommodation in the sense of more
lect levelling sults from and is dialect use in 'in-group' than in all
occur at all? hence foreshadowed *out-group' contact situations
in accommodation
Brought to you by | Radboud University Nijmegen
Authenticated
Download Date | 3/3/15 9:10 AM
 37

-group' contact is here defined äs contact between Speakers of the same dia-
lect, *out-giOiip' contact äs contact between Speakers of different dialects. Con-
sistent use of the features that set one's dialect apart from that of one's inter-
locutor can impede mutual comprehension, while the interlocutor may evaluate
it äs psychological divergence (Giles et aL 1987).
I will discuss some findings relevant to each of the above three hypotheses,
but I will pay by far most attention to the first one.

3. Research area
The research was carried out in Rimburg, a small village in the south-east of the
Dutch province of Limburg, and hence in the extreme south-east of the Dutch
language area. Rimburg, which is a hamlet in the easternmost part of the village
of Ubach over Worms, and has slightly less than 1,000 inhabitants. Rimburg is
about 8 kilometers from both Heerlen and Kerkrade and it is located right near
the Dutch-German border, not far from the German city of Aachen (Aix-la-
Chapelle).
I will now go into some dialect-geographical detail. Rimburg is located im-
mediately east of the 'Benrath line', an isogloss bündle which is very important
in the history of Dutch and German; äs one of the spokes of the 'Rhenish Fan',
it is related to the High German Consonant Shift. Some 10 to 15 kilometers west
of the Benrath line there is a parallel isogloss bündle. These two bundles cut
southern Limburg into three slices; I will label these slices C (the westernmost
slice), B (the one in the middle) and A (the easternmost slice). See Map l.
Rimburg is located in the -type dialect area, i.e. in the area where the dia-
lects are spoken that are usually referred to äs Ripuarian dialects. The B area is
generally known äs the transition zone between the Ripuarian and the East-
Limburg dialects, and the C area is where the East-Limburg dialects are spoken
(Goossens 1965).
For a Dutch speaking outsider travelling from C through B to A (i.e. from
west to east), the dialects become less and less comprehensible. The reason for
this is the fact that — from the point of view of the Dutch Standard language —
the number of dialect features increases from west to east, i.e. from C through B
to A. Most of the features of the dialects in C also occur in the dialects in B and
A; the features in B also occur in the dialects in A — but not the other way
around. In other words, from west to east the dialect features accumulate. South-
Limburg hence has a terrace shaped dialect landscape.1

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
38

Map 1: The three dialect areas in the southern pari of

the Dutch province ofLitnburg

ca, 3 kms

A ; Ripuarian di aJects
B: tfansitton ione Ripuarian. S-ast - Lsmburg «iiatects
C, East-ümtsurg diatecis

In recent history, Southeast Limburg has been some sort of a 'natural labora-
tory' or maybe even a 'pressure cooker' of demographic, social and cultural
changes. These changes were brought about by the very rapid industrialisation
of this area in the first decades of this Century with the large-scale development
of coalmining. The industrialisation created Job opportunities, which led to con-
siderable Immigration — of foreigners (especially people from Eastern Europe)
and even more so of people from other parts of the country, mainly Limburgers.
The migration led, in turn, to urbanization. All coalmines were closed down
between 1966 and 1976, but the effects of the industrialisation remain. Now-
adays, with respect to the number of inhabitants, the Heerlen-Kerkrade agglom-
eration ranks among the largest in The Netherlands. Its average density of
population is about three times the national average.

4. Methodological aspects
Less than a few centuries ago, in the age when — at least in the Western world
— there was nothing corresponding to the modern concept of 'Standard lan-
guage', there were only what is nowadays referred to äs 'dialects'. What was
language change like in those days? Of course, there has always been language
change resulting from internal pressures, such äs phonetically motivated sound
change, change in response to some sort of structural imbalance because of, e.g.
asymmetries in vowel Systems, Sapir's 'dritV, the tendency to keep paradigms

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 39

regulär and transparent (e.g. to restore paradigmatic regularity aiter a sound

change had destroyed it), analogy, etc.
Another, though related, source of language change is the modular organi-
zation of grammar and phonology. The interaction of general rules, principles or
constraints trom different modules may cause internal tensions or imbalances
which, in turn, may lead to the emergence of new variants; cf. the notion of con-
flicting constraints in Optimality Theory. Consider the following example,
Taithfulness' constraints2 require the phonetic Output form to be maximally
identical to the underlying form (a Dutch word such äs recht, /rext/, 'straight,
real', is then realized äs [rext]), whereas phonological wellformedness con-
straints may, e.g., require syllables of a specific structure, which may in certain
cases trigger the simplification of consonant clusters (at least of adjacent obstru-
ents in final position), resulting in the Output form [rtx], If the two types of con-
straints are not mutually ordered (cf. Anttila 1997), this tension may lead to
quantitative Variation between the variant which satisfies the relevant faithful-
ness constraint (while violating the wellformedness constraint) and the variant
which satisfies the wellformedness constraint (while violating the faithfulness
constraint). Differences between dialects can sometimes be accounted for by the
different setting of parameters in otherwise stable rules (cf. Lenerz 1984;
Cornips 1994).
All the cases discussed so far concern purely internal motivations. As far äs
externally motivated linguistic change is concerned, the mixing and borrowing
resulting from the prolonged contact between dialects have probably always
been very prominent phenomena. According to Ferdinand Wrede (see, e.g,
1919:10-13) and Theodor Frings, two of the main exponents of 20th Century
German dialectology, Mischung and Ausgleich, i.e. mixing and levelling, are the
key mechanisms that destroy regularity and the alleged exceptionlessness of
sound laws in individual dialects. Mixing and dialect levelling probably played
leading roles in the shaping of the type of koine that most of our Standard lan-
guages basically are. Historically, the Dutch Standard language, for instance, is a
koine, a hotch-potch of some Flemish and Brabantic and many Hollandic dialect
features.
In short, the study of processes of dialect levelling ought to be located at the
crossroads of the study of language Variation on the one hand and language
contact — i.e. research into the structural consequences of language contact —
on the other.
These considerations played an important role in the design of my study of
dialect levelling in Rimburg. The main Steps were:

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
40

(i) · operationalization of the notion of dialect levelling: dialect levelling

may manifest itself äs a statistically significant decrease in the use of
features which distinguish a dialect from surrounding dialects or from
the Standard language (or both), from Older through Middle aged to
Younger Speakers;
• Implementation of the hypotheses (resulting in Operational hypothe-
ses'). Cf. § 2. above.
(ii) selection of the dialect features. The main criteria for selection were:
• systematic Variation of the degree of geographical spread of the dialect
features. Three types of dialect features were studied: features which
only occur in the dialects in A, features which occur in A and B, and
features which occur in A, B and C alike;
• various components of grammar. Some dialect features concern pho-
nology, some morphophonology, others morphology and yet others
morphosyntax;
• structural coherence. The dialect features were chosen such that at
least subsets of features are structurally related to each other, such äs
e.g. sandhi voicing, which is typical of the dialects in all three zones,
the derivational suffix Miy", which traditionally marks the A- and B-
zone dialects (whereas the C-zone Limburg dialects äs well äs the
Standard language have Mik') and the postlexical '-weakening proc-
ess (typical of the dialects in the A-zone).3 The ratio underlying this
criterion was the desire to get an insight in how dialect levelling may
affect not only isolated dialect features, but also mutually related
subareas of the phonology and grammar of the dialect (cf. Chambers &
Trudgill 1980:291);
• only dialects features with a reasonable average frequency of use were
chosen for analysis. Purely syntactic phenomena thus filtered them-
selves out.
(iii) drawing a sample of Speakers:
• the sample was stratified for age group. Three age groups were repre-
sented by 9 Speakers each: 20-30 years of age ('Younger' Speakers),
40-50 ('Middle' age group) and 60-75 ('Older' Speakers);
• the sample was homogenized for sex in that only men were included;
• obviously, the sample was homogenized for linguistic background in
that only native Speakers of the Rimburg dialect were included who
were born and bred in Rimburg;

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 41

• the sample was homogenized for socio-economic background. Socio-

economic background was operationalized through two variables: edu-
cational background and occupational level. Both were measured on
six-point scales. The homogenization was reached by including only
people whose score on the mean of these two scales is minimally 2 and
maximally 4. So in this respect there is Variation within the sample, the
Variation spanning exactly half of the scale — of course the half where
we find the modes, means and medians of the distributions.
(iv) the collection of the speech material: of every Speaker in the sample re-
cordings were made of three types of dialect use:
• elicitation through all sorts of formal tests, and
• spontaneous, conversational dialect use in situations of Mn-group' and
'out-group' contact.
(v) analyses: from recorded speech material to (quantitative) data.
The analyses consisted of several Steps. The end result consisted of fig-
ures representing the relative use of the dialect features that were chosen
for analysis; so to an important extent the analyses were quantitative in
nature: the end result were indexes ranging between 0 and 100. In the case
of most of the dialect features studied, the index values can actually be
read äs percentages. The index values were broken down for a ränge of
relevant linguistic conditions (or 'factors').

5. Some of the main Undings

I will look at the three hypotheses in the reverse order. The third hypothesis,
which says that dialect levelling results from and is hence foreshadowed in
accommodation, was tested by comparing accommodation with dialect level-
ling. Accommodation was traced by comparing 'in-group' dialect use with dia-
lect use in the k out-group' contact situations — i.e. on the basis of the two types
of relatively spontaneous conversational data. Accommodation was operation-
alized very restrictively, namely äs the type of convergence which consists of
the suppression of the usage of features which might make one's dialect less
understandable for Outsiders.4 Overall, I expected less dialect use in out-group
contact than in in-group contact. Compare this specific Implementation of con-
vergence5 to convergence in the social psychological sense; in social psychol-
ogy (cf. Thakerar et al 1982; Giles et al 1987), convergence is defined äs what
Speakers do to adapt to the speech of others in order to reduce differences.
One of the main findings is: the smaller the geographical spread of a dialect
feature, the larger the relative number of linguistic conditions in which one finds
accommodation. In other words, the more typical/unique a dialect feature is for
a speaker's dialect, the larger the relative number of linguistic conditions in
which the use of this dialect feature is suppressed in out-group contact situations

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 42

(see Hinskens 1992:431-445 for details). This mirrors the patterns of dialect
levelling, which will be briefly discussed below. So it appears that 1. conver-
gence in the social psychological sense can be a pre-condition for structural
convergence, and 2. dialect levelling can be a long-term effect of sustained,
frequent convergence on the part of the Speakers.
Accommodation and dialect levelling should be understood in the light of
the continuous struggle between what de Saussure referred to äs "la force
d'intercourse et Tesprit de clocher" (part III, eh. 4 of the Cours), i.e. between
the tendencies towards unification on the one hand and those towards particu-
larism and cultural fragmentation on the other.
Hypothesis 2 says that dialect levelling is gradual in linguistic äs well äs in
extra-linguistic respects. Many of the findings I will refer to in this connection
are based on the data from the elicited dialect use. As far äs dialect levelling
goes, the main findings are summarized in Table 1; see Hinskens (1992: Ch. 6)
for a detailed account.
This table summarizes the findings for levelling in the sense of the loss of
each of the various dialect features on three different levels. The most general
level is what I call the overall level; a fc+' signifies that the loss of the feature
concerned is not confined to any specific linguistic condition(s), but takes place
throughout the aggregated data. 'Loss conditions' indicates the proportion of
linguistic conditions (or 'factors') studied under which a dialect feature appears
to be undergoing loss. For each dialect feature small sets of linguistic conditions
form what is here referred to äs linguistic dimensions; e.g. two of the seven
conditions which have been distinguished in the analyses of the dialectai
'-weakening process concern the question whether [y H ] is part of a lexeme or a
bound morpheme. As will be obvious, these two conditions complement each
other; therefore they form what I refer to äs a linguistic dimension (or 'factor
group' in the variationist terminology). The column for 'Loss dimensions' indi-
cates the proportion of the linguistic dimensions studied which turn out to have
a significant effect on the loss.

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 43

Table 1: An overview of (befindings regem!ing apparent time loss ofthe dialectfeatures

in the elicited materiell (V = verbal morphology)6
spread dialect feature example LOSS?
overall conditions dimensions
A Ach- laut allophone ος - οχ + 10/14 0/6
γ'-weakening Y*RaV3 - JR3V3 + 7/7 0/3
i-lowering bme - be.na 0/6 0/3
dorsal fricative deletion 1ις(0 ~ li:3t + 6/6 0/3
[s] in diminutive suffix me:3cjk9 - me:3gska + 3/4 0/2
Α,Β R-deletion e:3Rt ~ e:t9 5/13 2/5
n-deletion ναη ~ να 1/9 0/3
derivational suffix -Ίιγ'' i:Rlik ~ i:3Rli9 + 5/6 1/3
K preterite suffix
weak verbs magda - makat -
V prefixless past participle γ " 3ko3m3 - ko3ma +
Vsubjunctive wol ~ w01 +
V strong/irregular
- weak conjugation VROX - VRO:3K3t +
A,B,C t-deletion lopt - ΙΛΡ 0/11 0/5
sandhi voicing dat in ~ dad in 0/30 0/14
derivational suffix -kd3* WAR3mt3 ~ weR3md3 1/2 O/I
absence of inflectional Ιιογο d0r - hui3Y ' dy:I3R + 3/6 0/3
schwa
noun pluralization moeizs - my % s + 5/25 1/10
V strong/irregular
~ weak conjugation ZEt'9 "* ZQt -
V stem 2, 3 sg.pres.
indicative lopt - LOps -
oblique pro no uns wi - wem 0/6 0/3
expletive element ER - 3t / [zero] + 2/2 l/l
In all cases of dialect levelling, the process turns out to be gradual
(a) extra-linguistically:
• in (apparent) time, in that the degree of use of the dialect features de-
creases step-by-step from the Older to the Middle to the Younger age
(this is not expressed in Table 1). The levelling process also occurs
gradually
• in space: the smaller the relative geographical distribution of a dialect
feature the earlier and/or faster it is levelled out. Among other things 7
this is evident from the fact that the ratio of dialect features showing
overall loss to / dialect features investigated decreases with increasing
geographical distribution:
A: 4/5
A,B: 4/7
A,B,C: 3 / 9
(b) linguistically, i.e. internally, the levelling process is usually gradual too.
Two examples will serve to illustrate this:

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
44

• '-weakening, i.e. the weakening of the voiced palato-velar fricative into the
palatal glide, an -type feature:
A B/C StandardDutch gloss
H
(1a) ja's ~ y Vs y a-s yas 'gas'
(Ib) JRa:s~y"Ra:s y'Rais yras 'grass'
jla:s~y"la:s y'lais ylas 'glass'
The loss of the weakening rule appears to proceed faster betöre consonants (and
the only consonants we find in the dialect in this type of onset clusters are li-
quids — cf. (Ib)) than before vowels (la); see Table 2.

Table 2: Average use ofthe "-weakening rule in two linguistic conditions

by the Speakers ofthe three age groups

before liquids before vowels

Older 41.82 52.94
Middle 3.13 15.09
Younger 1.67 2.19

As is evident from the distribution ofthe means, the Speakers of the Middle age
group have practically given up the weakening rule betöre liquids, while still
applying it in more than one out of seven cases betöre vowels. The fact that the
weakening rule is levelled out earlier and faster before consonants than before
vowels can be explained on the basis of wellformedness constraints on syllable
structure, specifically sonority distance.
• R-deletion, i.e. the deletion of postvocalic R preceding a coronal obstruent in
monosyllabic words, a lexicalized rule ofthe A- and B-type dialects. The data in
(2) exemplify the traditional Situation:
A/B C / Standard Dutch gloss
(2a) koi^l ko^Rt 'cord, string'
so:3t soi'Rt 'sort' N
ba:t baiRt 'beard'
(2b) kot kDRt 'short'
4
hats haRt heart'
As can be seen in Table l, R-deletion is not subject to loss on an overall level.
However, there is a significant apparent time decrease in the use of this deletion
rule in five specific linguistic conditions. One of these conditions concerns
words which contain a short vowel — in words with a long vowel the rule does
not undergo levelling and the effect of the interaction between the linguistic
dimension ('factor group') length of the preceding vowel on the one hand and
the age group of the Speakers on the other is highly significant (F=3.65 df=2,24
p=.041). The explanation of this directionality has to do with syllable weight: a

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 45

syllable consisting of a long vowel plus r plus an obstruent in the coda, äs in the
words in (2a), has three moras and is therefore superheavy. In such a constella-
tion, R-deletion has a wholesome effect. This is different in words with a short
vowel — like the ones in (2b). In such words, R-deletion would lead to a change
from a heavy to a light syllable.
In the data from the elicited dialect use, there is no overall age effect on
word-final t-deletion, i.e., there is no statistically significant difference in the
Proportion of t-deletion between the Speakers of the three age groups. But in the
data from the two types of conversations, there is: to teil from the apparent time
increase in the use, the nature of this effect is that the rule gets applied more and
more often — the mean for all Speakers in the sample being approximately 80
out of 100. Is this a manifestation of dialect levelling? I think it is, äs it makes
the dialect sound more like the surrounding dialects, in which t-deletion is also a
very prominent feature.
On a more specific level, the following can be said: about a Century ago, a
grammar was published of a group of Ripuarian dialects of German — just
across the border (Münch 1904), and well over a Century ago, a grammar of the
local dialect of Heerlen appeared (Jongeneel 1884). Both grammars are quite
detailed and precise. 't-apocope' is mentioned in both of them. As far äs gram-
matical aspects go, both grammars point out that the deletion erases the 3rd
person sg. present indicative suffix [t], not however the 2nd person pl. suffix [t].
This is different in the present-day Rimburg dialect; my data show that t-
deletion also affects the 2nd person pl. suffix [t] — äs in (3a) — although sig-
nificantly less often than the 3rd person sg. morpheme — äs in (3b).
(3a) y;^R wmlct ~ y:3R wmTc k
you (pl.) work'
4
y:JR Jbpt -y^Jlop you (pl.) sleep'
(3b) he:11 WIR% - he:9 wiR^k *he works'
^Ikfjik Jli:3pt ~ Qt kfjik fli:3p 'the child sleeps'
So obviously the rule has been generalized and is in the process of loosing some
of its grammatical constraints.
In order to test hypothesis /, which Claims that dialect levelling is a two-
dimensional process, the question needs to be answered whether levelling of
cross-dialectal Variation is always a function of levelling of Variation in the dia-
lect-standard language dimension.
As can be seen in Table l, dialect levelling in the sense of the loss of dialect
features occurs in 14 out of 21 cases. In three out of these 14 cases, the loss is
only manifest in certain linguistic conditions äs yet — hence not on the overall
level. In four out of 14 cases, the levelling does not lead to convergence towards
the Standard language; in three of these four cases, there is even divergence vis-
a-vis the Standard language going on. I will briefly discuss these four dialect
features.

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
46

γ'-weakening — cf. (la, b). The weakening rule is given up; the then un-
weakened voiced fricative [y H ] is palato-velar, s in the surrounding Limburg
dialects, rather than velar-uvular, [y], s in the Standard language. Moreover, the
latter is more and more subject to devoicing (cf., e.g. Van de Velde 1996:102-
104).
Ich-Laut ~ Ach-Laut allophony (the [a back] specification of the preceding
vowel spreads to the fricative — s far s the Dutch language area goes, this is a
feature of the Α-type dialects):
A B/C Standard Dutch gloss
(4a) kRi9 kRi£ ~ kReg -('oorlog') 'war'
vle£ vle$ vlix 'fly'; 'fly!' (imperative)
fc
vley'3 - vleJQ vley'a vliya flies'; 'to fly'
(4b) dax da£ dax *day'
zax zag - (*zei';'zeg!') 'said';'say!'(imperative)
zox ζος ζαχ 'saw'
zay H a -('zeggen') 'to say'
klay's klaya 'tocomplain'
Traditionally, in syllable-fmal position (voiceless realization), the A-dialects
alternate between an ach-Laut allophone which is identical to the Standard pho-
neme /x/, and an ich-Laut allophone which is identical to the /ς/ phoneme in the
B/C dialects; in intervocalic position (voiced realization), the ach-laut in the A-
dialects ([κ]), is both articulatorily and acoustically almost identical to the Stan-
dard variant /y/, while the ich-Laut ([g]) is again identical to the phoneme oc-
curring in the B/C-dialects.
In the levelling process the ach-Laute disappear, hence the B/C-type dia-
lects, which have the palato-velar realization throughout, win — and not the
Standard variety, which does not have this allophony either; instead it has the
velar-uvular realization throughout. In the latter respect, and from an abstract
structural point of view, the levelling process could be said to lead to a partial
convergence t o the Standard System however.
The non-palatalization of the epenthetic /s/ in the diminutive suffix (the un-
derlying form of which is -ke in Limburg dialects); /s/ occurs when the stem
ends in a velar consonant. This feature sets the Rimburg dialect apart from all
immediately surrounding dialects:
Rimburg dialect B Standard Dutch gloss
(5) dijisko dijijko dingetje 'thing'-DIM
kAkJTco koekje 'cookie'
:Jk3 kraagje 'collar'-DIM

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 47

The Rimburg dialect is in the process of adopting the morphophonemics of the

surrounding B-type dialects. So here, again, the B-type dialects win — and not
the Standard language.
The System of noun pluralization undergoes levelling, but the levelling does
not affect the uniquely dialectal non-concatenative, stem-internal morphology,
such s Umlaut — e.g.
A/B/C Standard Dutch gloss
4
(6) apsl: εροί appel : appels apple' : 'apples'
kap : koep kop : koppen 'head' : 'heads'
or a change of the tone contour from High-Low-High (i.e. 'circumflex') to
High-Low ( k grave') 8 , e.g.
(7) evR3m : ε^ιΛη arm : armen 'arm' : fcarms'
ftev : Jtev steen : stenen k
stone' : 'stones'
These four cases show that dialect levelling does not necessarily lead to conver-
gence towards the Standard language. By and large, the concept of Standard lan-
guage is a 19th Century invention; the spread of Standard languages throughout
speech communities and the rise of mass literacy are even younger. So i t may
well be that cross-dialectal levelling is the older and historically the main or
even only type of dialect convergence. As Mate' s 1914 a study by Terracher of
the dialects of the region of Angouleme (France) brought to light that "l'agent
destructeur de la morphologie des patois n'est pas le fran^ais, mais les parlers
limitrophes" (Pop 1950:106).
But cross-dialectal convergence is certainly not a thing of the past;9 neither
is it a uniquely European phenomenon. For instance, in their recent study of
/oy/, the dialect variant of the Standard English diphthong /ai/ in the dialect of
English spoken on Ocracoke Island, off the coast of North Carolina, USA,
Wolfram & Schilling-Estes (1995:710) found that "Speakers may add vernacular
features from surrounding dialects to their speech", in this case the Southern
vernacular variant /a:/.
The Southeast-Limburg dialect landscape has in fact preserved an older in-
stance of cross-dialectal convergence. The dialects in the A-zone have under-
gone dorsal fricative deletion with compensatory lengthening,10 which occurs
neither in the dialects in the B- and C-zones nor in the Standard language. In a
subset of lexical items, B-zone dialects do show vowel lengthening but no dor-
sal fricative deletion — e.g.
A B C / Standard Dutch gloss
(8) na:t na:$(t) ηας(ί) / naxt 'night'
li:"t Iei9(t) h9(t)/lixt * light'

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
48

This seems to be a consequence of partial convergence between adjacent dia-

lects, specifically of the B-dialects towards the A-dialects.11 Convergence is one
of the processes that lead to the formation of transition zones.
For specific dialect features, cross-dialectal convergence can lead to either
complete or partial similarities between dialects — cf. the types of structurally
intermediate varieties which have been labelled 4 mixed' and 'fudged' lects.
Mixed lects are lects which, for a given dialect feature, combine variants from
two different 'pure' lects, to use Chambers & TrudgilFs (1980:132-37) termi-
nology.12 Fudged lects also combine the variants from two different 'pure* lects,
but they have an additional third variant which is a compromise between the
two 'pure' variants. For dorsal fricative deletion and compensatory lengthening,
B-dialect variants such äs those in (8) constitute a case of 'fudging'; they have
the preservation of the dorsal fricative in common with the C-dialects, while the
vowels are long, just äs in the A-dialects.

6. Conclusions and discussion

In sum, there is ample evidence in favour of each of the three hypotheses. The
answers to the questions underlying the three hypotheses may constitute build-
ing blocks for the construction of a fully-fledged theory of dialect levelling.
Hypothesis l, which says that dialect levelling is a two-dimensional process
in that it can affect both Variation in the dialect-standard language dimension
and Variation across related dialects, has a direct bearing on the question re-
garding the role of Standard varieties in the convergence and divergence of dia-
lects. It turns out that in some cases, the levelling of cross-dialectal Variation
occurs independently from the Standard language. In some of those cases, dia-
lect levelling even constitutes crossdialectal convergence and dialect-standard
language divergence at the same time. In most of these cases, the dialect change
appears to be structurally independent from the Standard language — but is it
also sociolinguistically independent from the Standard language? More in par-
ticular, the question is whether this type of dynamics would also have occurred
in case there had been either no Standard language or another Standard lan-
guage.
For the Rimburg case one can only speculate äs to the first scenario (no
Standard language). However, there are possibilities to study the second one, i.e.
a different Standard language. This different Standard language would be Ger-
man. As a matter of fact, at present a research project is being prepared at the
Department of General Linguistics and Dialectology of the University of Nij-
megen in which in the neighbouring German dialect of Übach-Palenberg (just
across the Dutch-German border) the vitality of a number of dialect features
will be studied which this dialect has in common with the Rimburg one. Tradi-
tionally the Übach-Palenberg dialect is very similar to the Rimburg one. This

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 49

new research project will not only make it possible to give a better answer to the
question regarding the structural and sociolinguistic role of the Standard lan-
guage in processes of dialect levelling. The outcomes will also be relevant to
two other issues. One concerns the relative weight of internal factors. The other
one is the question regarding the effect of state borders on the supposed erosion
of old dialect eontinua.13

Address of the Author

Frans Hinskens
Department of General Linguistics and Dialectology
University of Nijmegen
Erasmusplein l
NL-6525 HT Nijmegen (The Netherlands)
e-mail: F.Hinskens@let.kun.nl

Notes
* The work on this paper has been made possible by a fellowship of the Royal Netherlands
Academy of Arts and Sciences. The study of the Rimburg dialect, which is central in this
contribution, was supported by ihe former Foundation tbr Linguistic Research, which was
funded by the Netherlands Organisation for Scientific Research. I wish to thank the partici-
pants at the first workshop of the ESF Network on social dialectology, July 1996 in Berg en
Dal (Nijmegen), s well s the editor for discussion and comments, which led to a number of
significant improvements of this text.
1 Staffellandschqft, in the German terminology — see Bach (1934:61); (1950:144-145); Hard
(1972). Cf. also Weijnen (1977:18-19).
2 Cf. McCarthy & Prince (1995).
3 See Hinskens (1992), sections 5.3.2, 5.3.8, 5.3.15, 6.3.2, 6.3.8 and 6.3.15 for details regard-
ing the relationship between these dialect features, and section 6.4.4 for a short overview of
several other cases of structural relationship between dialect features and their consequences
for the actual process of levelling of these features.
4 Cf. the operationalization of the notion of accommodation in § 2 above. A case of the oppo-
site suppression of the usage of features which make one's dialect less understandable for
Outsiders, i.e. of linguistic divergence, has been reported in Labov's 1963 study of the cen-
tralization of the diphthongs in the English dialect spoken on the island of Martha's Vine-
yard, off the coast of Massachusetts, USA.
5 Linguistic, rather than merely psychological convergence, to be specific. The latter concerns
a Speaker's intention and conviction regarding his or her interactional behavior in general.
6 l consistently take s a starting point the varieties to the west that do not have the features
concerned. Likewise in the examples: whereas the C- and B-zone dialects have, e.g. [ος], the
A-zone dialects have [ox], etc., whereas the C-type dialects have [van], the B- and A-zone
dialects have [να], etc., whereas the Standard language has [lopt], dialects in the C-, B- and
A-zone dialects have [Up], etc. The glosses of the examples are, respectively, 'eye', 'to dig',
'inside', 'light', 'stomach-DIM', 'earth', Of, 'honest', 'made-PRET', 'come-PAST PART',
'would like to-l./3.SG.', 'asked', 'walks*, 'that in', 'warmuY, 'high door', 'mice', 'put-
PRET 1./3.SG', Valk-2.SG', 'whom', 'there'.
7 See Hinskens (1992), § 6.4.2 for additional approaches.
8 The Rimburg dialect, iike most dialects spoken in Limburg and the neighbouring Rhineland,
is a pitch accent system. Pitch accent Systems are a special type of tone language in which
the type of tone contour (i.e. intonational contour) with which a syllable is realized under
certain conditions, can be lexically or morphologically contrastive. See Hinskens 1992,
§ 2.3.5 for a more elaborate account and some illustrative examples.

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
50

9 See Auer & Hinskens (1996:13-15) for some of the most recent references.
10 These telescoped processes are no longer productive; äs a matter of fact they have been
lexicalized — cf. Hinskens (1996).
11 This analysis is corroborated by the fact that, in the course the 19th and the first half of the
20th Century, the center of prestige shifted from Cologne (some 75 km to the east), which for
many centuries had been an important center especially in economical and cultural respects,
to the north-western part of the Netherlands (roughly the triangle Amsterdam - The Hague -
Utrecht). Tliis shift followed the change in the geo-political position of most of the present
province of Limburg, which was incorporated into the kingdom of the Netherlands no earlier
than 1839. See Hinskens (1992:§3.2.1) for a brief sketch. Traditionally, dorsal fricative dele-
tion with compensatory lengthening occurs in the Cologne dialect (Münch 1904:37, 92, 95-
96, 158) and in most local dialects of German east of Cologne, including the neighbouring
dialects of Aachen and Herzogenrath.
12 Recall the sketch of the considerations regarding Mischung and Ausgleich in older German
dialectology presented in § 4 above.
13 See Hinskens (1997:75-78) for a more extensive sketch of the aims and methods of this new
project.

Bibliography
Anttila, Arto. 1997. "Deriving Variation from Grammar." In: F. Hinskens, R. van Hout & W. L.
Wetzels (eds.), Variation, Change and Phonological Theory. Amsterdam - Philadelphia:
Benjamins, 35-70.
Auer, Peter & Frans Hinskens. 1996. "Convergence and Divergence of Dialects in Europe. New
and not so New Developments in an Old Area." Sociolinguistica 10, 1-30.
Bach, Adolf. 1934. Deutsche Mundartforschung. Ihre Wege, Ergebnisse und Aufgaben. Eine
Einführung. Heidelberg: Winter. First edition; second edition 1950.
Chambers, Jack & Peter Trudgill. 1980. Dialectology. Cambridge: Cambridge University Press.
Cornips, Leonie. 1994. Syntactische variatie in het Algemeen Nederlands van Heerlen. PhD.
Diss. University of Amsterdam.
Giles, Howard, A. Mulac, J. Bradac & P. Johnson. 1987. "Speech Accommodation Theory: The
First Decade and Beyond." In: M. McLaughlin (ed.), Communication Yearbook, 10. Beverly
Hills: Sage.
Goossens, Jan. 1965. "Die Gliederung des Südniederfränkischen." Rheinische Viertel Jahresblätter
30, 79-94.
Hard, Gerhard. 1972. "Ein geographisches Simulationsmodell für die rheinische Sprach-
geschichte." In: E. Ennen & G. Wiegelmann (eds.), Festschrift Matthias Zender, Studien zu
Volkskultur, Sprache und Landesgeschichte. Bonn, 25-58.
Hinskens, Frans. 1992. Dialect levelling in Limburg. Structural and sociolinguistic aspects. PhD.
Diss. University of Nijmegen. A revised and abridged Version has been published in 1996 by
Niemeyer, Tübingen.
Hinskens, Frans. 1996. "Nacht in Kerkrade. Dorsale fricatief deletie en woord-finale t-deletie in
Ripuarische dialecten van het Nederlands." In: Roeland van Hout & Joep Kruijsen (eds.),
Taalvariaties. Toonzettingen en modulaties op een thema, ter gelegenheid van de zestigste
verjaardag vanToon Hagen. Dordrecht: Foris, 127-142.
Hinskens, Frans. 1997. "De Nederlandse-Duitse staatsgrens als belangrijke toekomstige iso-
glossenbundel." In: Wim de Geest (ed.), Recente studies in de contactlingu'istiek. Bonn:
Dümmler, 65-79.
Jongeneel, J. 1884. Een Zuid-Limburgs taaleigen; proeve van vormenleer en woordenboek van de
dorpsspraak van Heerle, met taal- en geschiedkundige inleiding en bijlagen. Heerlen: Wey-
erhorst.
Labov, William. 1963. "The Social Motivation of a Sound Change." Word 19, 273-309.
Lenerz, Jürgen. 1984. Syntaktischer Wandel und Grammatiktheorie. Eine Untersuchung an
Beispielen aus der Sprachgeschichte des Deutschen. Tübingen: Niemeyer.

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
 51

McCarthy, John & Alan Prince. 1995. "Faithfulness and Reduplicative Identity." In: J. Beckman,
L. Walsh & S. Urbanszyk (eds.), Papers in Oplimality Theory. UMOP 18, GLSA, Amherst,
249-384.
Münch, Ferdinand. 1904. Grammatik der ripuarisch-fränkischen Mundart. Wiesbaden: Sandig.
Photo-mechanical reprint, 1970, Bonn: Bouvier.
Pop, Siever. 1950. La dialectologie. Apercu historique et methodes d'enquetes linguistiques.
Premiere partie: dialectologie Romane. Louvain: Chez l'auteur.
Saussure, Ferdinand de. 1985. Cours de linguistique generale. Publie par Charles Bally et Albert
Sechehaye, avec la collaboration de Albert Riedlinger. Edition critique preparee par Tullio
de Mauro, postface de Louis-Jean Calvet. Paris: Editions Payot.
Thakerar, J., Howard Giles & Jenny Cheshire. 1982. "Psychological and Linguistic Parameters of
Speech Accommodation Theory." In: C. Fräser & K. Scherer (eds.), Advances in the Social
Psychology of Language. Cambridge: Cambridge University Press, 205-255.
Van de Velde, Hans. 1996. Variatie en verandering in het gesproken Nederlands (1935-1993).
PhD Diss. University of Nijmegen .
Weijnen, Anton. 1977. The value of the Map Configuration. Special Issue of the Mededelingen
van de Nijmeegse Centrale voor Dialect- en Naamkunde.
Wolfram, Walt & Natalie Schilling-Estes. 1995. "Moribund Dialects and the Endangerment
Canon: the Case of the Ocracoke Brogue." Language 71.4, 696-721.
Wrede, Ferdinand. 1919. "Zur Entwicklungsgeschichte der deutschen Mundartforschung."
Zeitschrift für deutsche Mundarten, Jg. 1991, 3-18.

Brought to you by | Radboud University Nijmegen

Authenticated
Download Date | 3/3/15 9:10 AM
Brought to you by | Radboud University Nijmegen
Authenticated
Download Date | 3/3/15 9:10 AM