-ANGUAGE IN SOCIETY

ialec s in CoGEl\: -RAL EDITOR:

Petcr Trudgill, Prof sso¡ of Linguistic Science,University of Reading

tactADV¡SORY EDITORS:

RaJph Fasold, Professor of Linguisties,Georgetown UniversityWilJiam Labov, Professor of Linguistics,University cf Pennsylvania

PETER TRUDGILL

Language and Social PsyehoJogyEdiled by Howard Giles and Roben N St Clair

2 Language and Social NetworksLes/ey Mi/roy

3 The Ethnography of CommunicationMürie! Savi/le- Troike

4 Diseourse AnalysisMichael Stubbs

5 Introduction to SocioJinguistiesVolume 1: The Sociolinguis ics of SoeieryRalph Faso/d

6 Introduetion to SoCioiinguisticVolume 1I; The Soeiolinguistics of LanguageRalph Fasold

7 The Language of Children and AdoleseentsThe Acquisition of Cornmunicative CompetcnceSuzanne Romaine

8 Language, the Scxes and SocietyPhilip M Smilh

9 The Language of AdvcrrisingTorben Vcstergaard and Kim Sc/¡roder

10 Dialects in ContaetPeter Trudgil/

11 Pidgin and Creole LinguisticsPeter Mühl/¡ausler

Ii

I

BASIL BLACKWELL

© Pcrer Trudgill 1986

First published 1986

Basil Blaekwell Ltd108 Cowley Road, Oxford OX4 lJF, UK

Basil Blackwell Inc.432 Park Avcnue South, Suite 1503,New York, NY 10016, USA

Al! rights reserved. Except for the quotation of short passages for the purpo es oferitici m and rcview, no part of this publieation may be reprodueed, sto red in aretrieval systern, or ransmirted, in any form Oí by any means, eiectronic,mechanical, photocopying, recording o. otherwise. without the prior pcrmissionthe publishcr.

Except in thc United States of America, this book is sold s bjeet to the conditionthat it shall not , by way of trade or otherwise, be le t, re-sold, hired out, orotherwise ci culated without the publi her's prior consent in any Iorrn oí binding orcovcr other than that in which it is published and without a similar conditionincluding his condition being imposed on the subsequent purchaser.

British Library Cataloguing in Publica/ion Data

Trudgil!, PeterDialects in contactol. Dialeetology 2. Gramrnar, Cornparative and genera!I. Title471'.2 P201

ISBN ~631-12691---DSB '~31-12733-X Pbk

Library of Congress Cataloging in Publica/ion Data

,..

Trudgill, Peter.Dialects in eontaet.(Language in society; ID)Bibliography: p.lncludes index.1. English Language-Dialects. 2. English Language-Social aspects.

3. Language in contact. 1. Title. Il. Series: Language in society (Oxford,Oxfordshirc); 10.PE 1711.D 1986427 85-30815IS8 ~31-12691-0ISB ~3l-12733-X (pbk.)

Typeset by Katcrprint Co. Ltd , OxfordPrinted in Great Britain byT.J. Press LId, Padstow

..

e ntents

Acknowledgements VI

Introduction VlI

1 Accommodation between Dialccts 1

2 Oialect Contact 39

3 Dialect Mixture and the Growth ofNew Dialects 83

4 Koinéization in Colonial English 127

f rences 162

ndex 169

Acknowledgements

This book has been a long time coming - too long, 1suspect, in the viewof thc publishers - and has been worked on in many different locations.1 am particularly grateful to colleagues and students who discussedtopics in the field of dialect contact with me at the Australian NationalUniversity, the Univcrsity of IIlinois, Stanford University, the Uni-versity of Texas at Austin, and the University of Toronto , as well as atthe University of Reading. 1 am also especially grateful to PhilipCarpenter, J. K. Chambers, Nikolas Coupland, Ralph Fasold, JcanHannah, John Harris, Richard Hudson, James Milroy, and LesleyMilroy, who read earlier drafts of the book and gave invaluable advice,only some of which fell on deaf ears. 1have also , J hope, oenefited fromdiscussions with and vital inforrnarion received from the following, towhorn 1 also express my thanks: Ian Bild, William Downes, JanetFletcher, Tina Foxcroft, Elizabeth Gordon, Jan Hancock, John Holm,Ernst Hákon Jahr, L. W. Lanharn, Tom Melchionne, Helge Omdal,Jarnes S. Ryan, Alison Shilling, Gary Underwood, Keith Walters, andJeffrey P. Williams.

1 would like to thank 'he following for permission to redraw andreproduce figures: Cambridge University Press (l.4, 1.5, 1.6); ProfessorJ. K. Charnbers (1.7); Mouton de Gruyter (1.1, 1.2, 1.3). 1 am alsograteful to the folJowing for permission to redraw and reproduce maps:Cambridge Univr-rsity Press (3.15); Craom Helm Ud (2.9, 2.10);Edward Arnold (4.4); Universiteitsforlaget, Oslo, Norway (3.4).J..

..

Introduction

This book is concerned \Vith the subject of linguistic change. It is,however, about only a very resiITc:ted set of all the possible types ofchange that can occur in language. It is concerned, in fact, with thosechanges that take lace durin or as a consequence of contacts betweenclosely related varieties of language. It deals wit 1 owan wny mutua y;ntelbglble lÍnguist¡c vanehes may mftuence one another, as welI as withthe social and geographical spread oí hnguistic forms frorn one dialect toanother. It also focuses on theway in which, in certain dialect mixturesituations, totally new dialects may be formed. These changes c1earlyform only a fraction of the changes that can occur in human languages,and 1 do not wish to overstate the importance of the role of dialectcontact in inducing change. Nevertheless, it emerges that a very greatdeal of information is already available in the literature on particularinstances of contact-induced change , and an extensive study of theseworks does suggest very strongly that dialect contact i5, in its way, asimportant aD area for investi ation as l;mguage contact.

e wor IS very much a study in SOCIO mguisncs. This is especially soin that it argues for the crucial importance, in the stud of dialectcontact, of human behaviour in face-to- ace interaction. Unlike manymteractlOaal sociolinguistic studies, however, it concentrates, in themanner of Labovian-style secular linguistics, on language form ratherthan on matters of greater concern to social scientists .

What 1 have tried to do in this book is to examine a numbcr of tneparticular instances of dialect contact described in the literature, both inmy own work and that of others, and draw fram them, as far as possible ,general conclusions about !he forces that appear to be at work durin~ processes involved in dlalect contact. My method has bcen toattempt explanations - usual!y very ad hoc - for developments that haveoccurred in one situation, and then to see if these can be generalized toother similar situations.

The ultimate goal of work of this type wilI be !O 1='1'-"'I\..1 \..Aa\..11% nuu •

VIII lNTRODUcnO

~U061LL ) p. (19~6) DJa/ects Ir') corcta.c,~\JCl YorK. I Easll .BlacklNell

1Accommodation between Dialects

..

2 A~LOMMODATION BET\VEEN DIALEcrs

very.well that, say '. ~merican English sidewalk corresponds to EnglishEnglish pavement, u ISperfectly possible that the American will eventu-ally s~art saying pavement, and/or that the English person will begin ~osay sidewalk - even though there is no strictly communicative point 10their doing so.

EX<l® ••why this.kind of thing..should happenis not immediatelv clea~.OneJJJeo: th~t seeks to ex lain these a arent] r unnecessary lingui~m?odicatlOns IS.that developed by the sociai I2svchologlst Howar IleLGlles (1973) wntes ot conversational situations thai- 'if the senderj

CIyaCriC situatíon wishes to gaín the receiver's approval, thcn he mayaJapt hls accent patlerns fowards that of thÍs persono re. reduce pro:nunclanon dlsslml]aríties.' Giles labels tRiS process--accent conver!j.:.ence' , and points out that the reverse process. 'accent divergenc~, maytake place instead if, for example, spc:1kers wish to dissociate them-seives from or show disapproval 01 otners. Thcse processes oí conver-gence and di\'ergence can clearly al so takc ~e at the grammatical andlexic:lI levels (though see Coupland, 1984). and are prcsumably part of awider pattern of behaviour modification under the influence of and inresponse to others. Scholars in fields such as communications andpsychology have, indeed, investigated this type of convergence/diverg-ence behaviour with reference to many other non-linguistic factors suchas body movernent, proximity, speech rhythm, speech speed, silence,gaze direction, eye contact, and so on. (A critical summary is providedby Gatewood and Rosenwein, 1981 in their papcr on interactionalsynchrony. See also Cappella, 1981; Dittrnan, 1962, 1972; Feldstcin,1972; Jaffe and Fe!dstein, 1970; Kendon, 1970; and Patterson, 1973,1983.) There is in this literature a strong sense that convergence of thistype is a universal characteristic of human behaviour-

In any case, behaviaural convergence is obv' usly a topic of ~turaland central interest for social psycholo ists and lan ua e rovides them\vit a very useful site for the study of this phenomcnon. Gi!es and hisCó-workers, as social psychologists of Ianguage , have cleveloped, usingla guaoe as data the theory alluded to above and labelled by thernaccom~~ theo~ This theorLfs>cuses ?n spe~ch, and dis~uss~

• ane attemp s~ ain wh[ speakers modtfy thcir lar~u~e I!!.,thejfresence of others in the way and to the extent .that they do. !~oexamines toe effects and CQsts of this type of modification..•.. "iles's initial (1973) paper looks mainly at convergence and diverg-ence in short-terrn contacts and in terms of adjustments up and downthe social dimension from high-prestige to low-prestige accents. Insituations where speakers with accents of different social status comeinto contact, the direction in which accommotlation will take place isoften problematical, and Giles and others have devoted considerable

~

ACCOMMODATIO:-l BETWEEN DIALEcrS 3

attention to exploring what factors are involvcd in determining whoaccornmodate-, to who; why speakers do it; to what cxtent they do it;and how it is perceived by others (see Giles et al., 1973).

From the perspective of the linguist, however, it is clear that ~modation can a(so take place between acccnts that differ re io arather than socJally, and t at It can occur in the long term as well as inthe short termo In lona-term éontacfs, whO accommodates 10 »,ho is le;S:~matical, ;ince, in most cases where this phenomenon can beobselved, we are dealing with contact bctween speakers of differentregional varieties, and with reglOnally mobile individuals or minoritygroups who accommodate, in thc long term, 10 a non-mobile ma 'oritt at t ev ave come to (¡ve amon st. e pro em is then one of

etermining how spea ers accommodate, the extcnt to which theyaccommodate, and why sorne sítuations and so me individuals producemore - or different types of - accommodation than others, Long-termaccornmodation is therefore of I . - c; or the social s e5ut of considef3.bie interest to the Iinguis.L

Short-term accommodation

Work in accommodation theory on short-term accommodatíon betweenspeakers wlth socially different accents has roved to be most in si htful!rom a socIOpsychologlca perspectl~. It has been found, for example ,that hnglllstlc convergence In a socially downward directíon can lead, insome cultures, to speakers being evaluated as kinder and more trust-worthy than if thcy do not converge (Giles and Smith, 1979); and that, ifa person anticipates meeting another 'socially significant' person in theimmediate future , then the latter's speech (if, say, overheard) is per-ceived by the former as being more like the former's own speech thanwouid otherwise be the case. Many other exarnples could be given.

Frorn a linguistic point of view, however, work on accommodationtheory has until recently been less informative. This is not, of course,intended as a criticism of the work of social psychologists, since theirobjectives were obviously very different. However, it is apparent thatmany more insights, in addition to those already obtained, could begained by more linguisticaJly sophisticated analyses of the accornmo-dation process itself than those initially employed by Giles and hisassociates. In the wor~f thes~_~9.~i~l2.svc~ologists, for instance,jhe?egre~ ~f li.n~~o~.!!!odation indulged in by speakers is measu!}!dImpresslomstlcaE>'. Typically, tape recordings of speakers are played togroups of hngUlstically naíve subjects who are asked to assess them interms of accent 'broadness'. No actual linguistic analysis is involved at

1...

..

4 ACCOMMODATION BETWEEN DIALECTS

all. It is apparent, however, that dctailed linguistic analyses of iheaccommodation process would bring with thern a number oí bencfits forboth social psychologists and, especially, linguists. For example thcywould permit, amongst other things:

(1) An exact, rather than impressionistic, 9..uantification of degrce ~Jinguistic accommodationj

(2) An examination of which linguistic features are and are notchangcd during accommodation, togcther with cxplanations fortn1s;

(3) A study of whether accommodation is a uniform proces~ orwhethcr ljng~y dIfferent types of accommodation takeplace in the case of different speakers, different situations, ordifferent relationships;

(4) A study of .the limits of acccmmodatíon, what are the Iinguistic(as opposed to social and psychological) constraints 011accornmo-dation, and is it possibJe to accommodate totally to a new variety?

Quantification of accornmodation

We begin by dealing with the first of these benefits - that which arisesfrom exact rather than impressionistic quantification. We bear in mindin so doing that it has been one of the achievements of soeiolinguistics todemonstrate that the quantification of !inguistic phenorncna can revealhitherto unsuspected findings of considerable importance. Given thatthis is so, we must expect that exact quantification will provide ananalysis of the accommodation process more r~ing.J.han t1üitoIthesocial psychologist. -

We can illustrate this particularly clearly by examining those situ-ations in which social psychologists have been most interested, namelythose involving short-terrn' accommodation bctwcen speakers withsocially different accents. For example, Coupbnd (1984), in a pioneer-ing study of t·' '. . odat'on of an assistant in a travelagency to customers, in Cardiff, Wales, investigates three ardif{E"nglish variables. These are:

(1) Ihl vs. 0 in house, hammer etc.(2) [t] vs. [9] in better, city etc.(3) IUI vs. Inl in walking ; waiting etc.

Figures 1.1, 1.2, and 1.3, frorn Coupland (1984), show a very c1earcorrelation between the assistant's pronunciation and those of 51 of her

ACCOMMODATrON IlETWEEN D1ALECTS 5

clients grouped by social class. In faet, CoupJand writes that the percent-ages of variants in her s eech 'prove to b..: almost as good an indicator ofthe socioeconomic class and e uca iona ac groun of her interlocu-tor5 as the percentagl: of those forms in the ..8r0....!:!EL0fcl~~ts' ownspeech'. Indeed, Coupland's study as a whole is an exccllent example othe benefits of quantification in the study of accommodation.

lOO~

'~ 80~•..",.1:'"'"g 60~'"~B'O 40~~~~ 20r-- 16.7[

c1ients assistant

n n 88.7

n68.9

~353 I :'~31 fl'

lllilJ1 1 1 11 1 1 ILL.I ..I...-.--:';MO1-.1.1 ..to:::::J......-L.;-_II lIIN IlIM

occupational c1assIV V

Figure 1.1 Variable (h): comparison of clients' use and assistant's use; c\ientsby occupation (fram Coupland, 1984)

A further example of this type of quantification is the following. In hisinitial paper, Giles (1973) argues that the process of accommodationmay lead to circularity in the researeh of sociolinguists. In a comment onLabov's work in New York City (Labov, 1966), Giles suggests that itmay be the case that when they are intervíewing informants, socio-linguists expect the pronunciation of their inforrnants to correlate with,say, social class. The interviewing linguist therefore accomrnodates inanticipation, as it were, and speaks wlth a 'broader', more Tegióñálaccent when interviewing lower-class speakers than when recordinghigher-class inforrnants. The informants in the Íace-to-face situationtnen ac.commodat~ !Q. the intervICwer, producing the sort of languagetiíat \Vas expected and fu!filling the sociolinguist's erophecy. The resultsPrsome sociolinguistic surveys may tñerefore, aecording to Giles, besomewhat suspect.

~sing SGciGlinguists WGuld, J believe, wjsh to reject thi§hx.pothesls rather strongly. Certainly my own feeling concerning my

001-

6 ACCOMMODATION BETWEEN DIALECTS

I~O

OL-~O~ __ ~·~IL-1~~L_-L~L-~Ll~-LJl! 1Il N III M

occupational c1ass

60-

20f-

clients assistant

n n so.O

~'

n.1

2Q.O

IV v

Figure 1.2 Variable (t): comparison of clients' IS':: and assistant's use; clients¡,y \lccupation (from Coupland, 1984)

fiOL-~o~~I~-L __ ~-L __ ~~ __ ~~~Ll~

Ir 1lI N III Moccupational c1ass

1001-

c:ee 801-;;>"2'"] 601-~~.5:!o 401-<>0.0

'"c:~ 20~<Jo.

,..

clients assistantn n55.7

'Al

8~8 ~.7

7~0 73~68)2.

I~O

IV V

F:::ure 1.3 Variable (ng): comparison of clients' use and assistan"s use' clientsh .x'cupation (from Coupland, 1984) ,

~':r\ y of the En..&lih spo~en in Norwich (Trudgill. 1974) W h tao, mmodation did indeed take lac ... oda'ed lir ~. 1 h h . duci h l ~UIS-tI;.i V to my 111ormants rat er t an 10 ucmg t ern to accomm d_ . _ o ate to- ~ ---=--=-

ACCOMMODATION BETWEEN DIALECTS 7

me. As a native of the are a 1was investigating, moreover, 1had donethTs easily and in a relatively automatic, subeonscious way. (In carryingout linguistic interviews, as is well known, one wants as much a possibleto reduce the effeet oí the 'obse er's paradox' (Labov, 1972) and to putinformants at t ir ease. One has to behave, dress and speak in themanner most likely to produce rcJaxed conversation, and linguisticconvergence is part oí this overaJl pattern.) 1 h d, however, no directevidence to support ti is feeling, or o refute iles's hypothesis. Whatas required was a quantitative study oí the accornmodation process as

it had occurred during rny sociolinguistic interviews.To investigate t e ex ccommodation occurred, 1 there-

fore be an an anal sis of m own s eee a an m erviewer OiliTryorwieh tape recordings. Linguistic self-an Iysis by intervi~ers has

been carried out before, notably by Jahr (1979) in a paper in Norwegianentitled 'Er det sánn jeg snakker?' (Do 1 speak like that?) (see alsoJahr, 1978). Jahr analysed his use of a number oí syntactie variablesas an ir terviewer for Talmalsundersfjkelsen (investigation of spokenlanguage: abbreviated TAUS) i Oslo, He concludes that his syntax wasto a certain extent influenced by the sex of his informants and also bytheir svntax.Analysis of my own recordings revealed that aecommodation of a

rather dramatic phonological sort did indeed take piace. Fig re 1.4shows the seo res for the variable that were obtained by ten of theinformants in the Norwic study. (These inforrnants have been selectedfrom the total sample of 60 for the purpose of this study to give (t) seo resacross the whole range.) Figure 1.4 also shows the (t) scores obtained bymysel in interviews with eaeh of these informants.The variable (t) refers to the pronunciation of intervocalic and word-nal /tI as in better and bet and has three variants:

(t)-l = [t](t)-2 = [1]](t)-3 = [2]

Index scores are caJculated in such a way that they range frorn 0,indicating consistent use of the prestige pronunciation [t], to 200 forconsistent use of the low-prestige glottal-stop variant.Figure 1.4 dcmonstrates a remarkable degree of eoincidence between

my seo res and those of my informants. Clearly, accommodation hastaken place. lt is apparent, however, that the c10se approximation of the..!Wo lines as they slope 3.cross the graph has been produced by illi~ccommodating to my informants [ather than vice versa. For two of theinformants, Mrs W. and Mrs B., my scores are the lower, i.e. 1 did notuse so many glottal stops as th y did. These are the two informants who

Figure 1.5 shows that, although as we have just se en l.did accommo- •...dat to my informa~'n the ~ase of (t), 1di commodate to them

~y pronunciation of (a:) (or if did, any accommo ation was veryslight).

Without a detailed linguistic analysis, a finding of this sort vould nothave been possible. If no linguistic analysis had been carried out, wewould not have known for certain that, during accommodation betweenaccents that differ at a number of points, some features are modified andsome are noto Now that this fact has been attested, the very interestingquestion arises: why are some as ects of ronunciation altered..Qjlc·the accommodatlOn process while others remain unchanged?, If we area6Ie to make some progress towards answering this question, we may

8 ACCOM.1úDATION BETWEE. IALECTSACCOMMODATION BETWEEN DI LECfS 9

1401:01008000

40

never have been revealed by impressionistic measures f degree ofaccomrnodation. Figure 1.5 again compares my pronunciation ith thatof my Norwich informant , and relates to the variable (a:}z which dealswith the degree of fronting or backing of the vowel of the lexical set ofpart, path, ha ¡,banana, etc. There are three variants:

o__ e (a:)-l = [o:](a:)-2 = [Q:-9:](..)-3 = [a.](1)

0--inforrnant o• 0 •--o

20OL-~~~~~~~ __ ~ __ ~ __ ~ __ ~ __ ~Mrs Mrs Mr Mrs Mrs Mrs Mrs Mr Mrs MrsW1 B U H W2 G J L L

Figure 1.4 Variable (t): seleeted seores in author's Norwieh study (Trudgill,1974)

•

Indices are calculated in such a way that consistent use of the receivedpronunciation (RP) back vowel :1 o low- resti e frontvanant

.--. _._.~o"-....o-° author • _~

were lowest on the social scale and who used m st glottal stops. For theother eight info mants, the gr ph shows that 1 used more g ottal-stoprealizations of Itl than they did, including those informants with thehighest social class indices and lowest (t) scores. It is probable, 1 believe,that if 1had been modifying my pronunciation in such a way a to inducerny informants to produce pronunciations that would correlate withsocial class in the anticipated direction, then the cross-over pattern onthe graph would have be en reversed: 1would have had high 'r (t) indicesth n the working-class speakers, and lo ver scores than the middle-classspeakers.

The fact that 1have higher scores than most of the informants must be••ascribed to the factor of age. Glottal-stop realizations of (t) are increas-.;.

ing in freq' ency, nd younger speak rs typically scorc hig Jer than older.3?eakers, otter thmgs being equal. At the time of the mterview J wa-saged 24, and the ten inforrnants shown here were a!l older than that.

~t tbe influence of the sex oLtbe interlocutor ootcd..b.yJahr (1979) is probably at \Vor e. ~r?m the graph it !?O' s a if 1 may'well have een usmo a hi her roportlOn of low-prestl e lottal-srealizations of Itl when talkin 10 the two men t an to the ei I1t l' •

This is consonant with the findings of Shopen (ms. who has found thatin Australian English at least, both men and women use more higher~status pronunciations, on average, when talking to wornen than whentalking to men.

\Ve can argue , then, that linguistic analysis is a use fui tool in anyexamination of the processes in volved in linguistic accommodation. Thisis clearly demonstrated in figure 1.5, which presents a finding that would

200~o __ e

180r160140120

100

inforrnants/0_0o_o

(a:)

8\60r

~~L__-L__~ __~ L-__~ __~ __~oL"-....__ ~· __ ~.Mrs Mrs Mí Mrs Mrs Mrs Mrs Mr Mrs MrsWl B U H v 2 G J L L R

Figure 1.5 Variable (a:): seleeted scores in author's Nor 'ich study (Trudgill,1974)

..

10 ACCOMMODATION BETWEEN DIALECTS

also gain some insights into the mechanisms that come into p.lay indialect contact situations of the sort that we shall be observmg insubsequent chapters.

Explanations for modification

readingpassage

formalspecch

casualspecch

Figure 1.6 Variable (a:): by c1ass and style (fr '1 Trudgill, 1974)

But w y not? Simply to point out that (a:) is an indicator and (t) amarker is not to explain why this is so, or how this distinction arises inthe first place. Labov suggests in fact that~arkers are relatively high ina s eaker' consciousncss as comparr>d to mdlcators. (Variables which. ave an especiaJly high level of awarenes associated with them arecalled stereorypes.) The high leve! of a 'areness associated with'a ma '.QL

leads speakers to modify their pronunclahon ol It m sltuations (such as10rmal occasions) where they are monitoring t eir speech m~t c10sely

ACCO IMODATION BETWEEN DIALECTS ] 1

(though see Bell, 1984). The same explanation obviously works for theaccommodation process: in contact with speakers of other lang agevari ties, speakers modify those features of their own varieties of whichthey are most aware.

This leads, of course, to a further question: wh exactl are s eakersmore aware of sorne variables than ot ers? Our earlier Norwichresearch (see Chambers and Trudgill, 1980 sugge ted that in theabsence of cer ain factars, at least one of which must be pre$eñGlinguistic variable wIiI normally be an mdlcator. In fhe case oi Norwlch,neast, the factors which !ead to greater awareness and thus to anindicator becoming a mark:E are the following:

(1) Greater awareness attaches to forms which are o ertly stig a-tized in a articular communit . Very often, this oyen stigmatiz-ation is beca use there is a high-status variant of the stigmatizedform and this high-status variant talli s with the orthographywhile the stigmatized variant does not. Examples of this in or-wich English include 0 vs. Ihl in hammer etc., and Inl vs. IUI inwalking etc.

(2) Greater awareness also attaches to forms tha1 are currentlinvolve m mguistic change.

(3) S ea ers -e_also more aware of variables whose var" t are,2honetically radically different.

(4) Increased awareness is al so attached to variab at are involvin e maintenance of honological contrasts. Thus, in Norwich,items from the lexical set of huge, cue, music, view, tune may bepronounced with either lu:1 or Iju:/. The latter pronunciationimplies a contrast in minima! pairs such as Hugh:who, dew:do,feud:food etc. The former, on the other hand, involves a loss ofthis contrast.

Long-term accornrnodation

\Ve are thus able to argue that, <1ltting accommodation to speakers whoare members of the same immediate speech community, speake~

ñi'Odil'Y their pronunclatlOll of Iingulsdc vanables that are marke..§.wlthm the commumty. 1his 15 5ecause oi {he sal ence which attaches to·marker~ and indeed turns variables in o arkers in the first place. Thi;salience is, in turn, due to factors such a tose we have just outlined -to do with stigmatization, linguistic chan¡:¡;, phonetic distance, andphonological contras:.,(see Timber ake, 1977; Kerswill, 1985) .

1•

..

12 ACCOMMODATiON BETWEEN DlALECTS

The next qucstion we would like to ask concerns the xtent to whichspeakers accommodating to othcr speakers from otj¡"¡,'speech communi-nes wIlI oét!ave in li,e saJile way (see Knops, 1981). ls!Í the case tfiat,~commodabng to vanefies that are regionally different from th irown, speakers wil! also modif fcalures that for them are in some wasalient ( t IS not immediately clear tnat t .ey wi 1, since accommodatioDbeyond the speech community will often be a rather mrrerent. rom ac ommodation within it. Accommodation \••.ithin the s eech corn-mumty, as 111my Nor lch interviews, inyolve a tering the frequency of

usage oí partIcular yanants ot yana51es oyer which the speaker aireadas control. Accommo atIon eyon t e speech communit on thTJ-

~ther hand, may we 111volve the adoption of tata y n w fea ures '?!.pronuncianon.

We move now to nn investigatíon of this issue, with particular refer-ence to the question of whether It IS salient linguistic fea mes that aremodified in all types of accent convergence. We do this by examining anot uncommon type of long-term, ex ra-speech cornmunity accornmo-dation in the English-speaking world, namelj accommodation byspeakers of English English to American English as a result of resldencein the United States.

In carrying out this investigation, we are naturally cone rned toestablish exactly what are the features of American English that aremost prominent in t e consciousness of English English speakers,for whatever reason. This, in faet, is a rela ively simple task as faras phonology is concerned. Obviously the mas salient features ofAmerican En lish ronunciation, for trlglish pea le, are preciselytose \'\'1 ich are reproduce during ÍtmtatlOll. ost speakers o ng ishEnglish do not of course spend much of their lives imitating mericanEnglish, but there are a number of speech events where this doeshappen, such as the tellina of jokes invoh'in Americans, and thep!aying of American roles by ng IS 1 actors Pcrhaps, however, themost obvious site for thestudy of the imitatíon of American English byEnglish English speakers lies in the linguistic behaviour of Britísh pODsinger.s-

It can readily be noted that singers of this type of mus'c observe to aremarkable extent a number of tules concerning the way in which the\vords of pop and rock songs should be pronounc~. The strength withwhich these rules apply varíes considerably from singer to singer andtime to time. but it is clear that most such singers employ differentaccents when singing than when speaking. lt is also clear thar, whateverthe s~eaking a~cent, the si~g~ng accent is one whi ? Í.s influenced byAmencan English pronuncianon. The process that 1 l. volved in thisphenomenon, moreover . iously imitation and not accommodation.

ACCOMMODATION BETWEEN DlALECTS 13

In modifying their accents as they do, singers render their pronunciationless like that of their British audiences, not more.

Analysis of the pronunciation used by British pop singers, from tht;.late 1950s to the late 1970s shows that the followin as ects of

merican English oronuncia ion are widespread, normal or even com-~lsory (see Trudgill, 198'»:

(1) Word such as life, my tend to be ~un with. a m~nophthon-gal vowel of the t pe [a-], alth ugh III spoken English Englishthey are most usually pronounced with a diphthong of the type[al-m-Al] etc.

(2) Words such as girl, more tend to be pronounced with an Irl evenby those Engli h English speakers (the majority) who do not havenon-prevocalic Irl in their speech.

(3) Words such as body, top may be pronounced with unrounded [o]instead of the more usual British [o].

(4) lt is not usual to pronounce words such as dance, last with the la:1that is normal in the speeeh of south-eastern England. lnsteadthey are pronounced with the lrel of cal (as in the north ofEngland, although the pronunciation is usually [re] rather thannorthern [aJ). In addition, words such as hal] and can't, which arepronounced with la:1 by most northern English speakers, mustalso be pronounced with lre/. Thus:

cat dance halfsouth-eastern England te! = [re] la:1 la:1northern England lrel = [a] lrel la:1pop-song style lrel = [re] lrel lrel

(5) The pronunciation of intervocalic Itl in words like better as [t] or[2], which are the pronunciations most often used by most Britishspeakers, is generally not used. In pop singing, a pronunciation ofthe type [r-Q] - a voiced alveolar flap - has to be employed.

Q!hcr features of American English pronunciation do occur but th~are less freguent and less wide pread. Clear y, the above fiye features.§he most common in British pop-singíng style because it is the~pronunciations which are most saliently characteristic of Americanaccents for the sin ers and, presumabl , other British or at least

n Iish eo ie. As to w ritish ingers should want to imitateAmericans, see Trudgill, 1983, chapter 8.)

Why these features should be sa ient in this way is less easy to.establish, but an exarnination of the rcasons suggested above for thegrowth of markers in Norwich English does give us some clues. Thefactor that has to do with ongoing linguistic change is not likely to be of

14 ACCOMMODATION BETWEEN DIALECTS

relevance. But the other factors - those that have to do wi h phoneti .distance, stigmatization, and phonological contrast - do provide sumepointers. In particular, we can note a sirnilarity between the last two: thevariable affected by stigmatization often involve, as the commeñ1above abo.'t orthography suggest , phoflological rather than phoneticvariation and alternation or contrast between surface phonemes. Tne(ñg) variable, for instance, involves alternation between one phon !1 eIUI and anot er In/; whilc (h) involves alterna ion bet veen a phon e/hJ and i s absence, just as the Hughiwho (yu) variable involves alterna-tion between zero and Ij/. Note then that the salience of r.on-prevocalicIrl in American English for English listeners may have to do with thefact that this dif erence between the two varietic al o concerns~-ence of a phoneme versus its absence, whil the salicnce of l<elin danceal so involves alternatÍon betwecn two phoncmes l<el;:md la:1 rathcr thana purely phonetic differcnce. The other three features are less easv toaccount for, but notice that American 101 in hot does sound like En~lishla:1 in hcart (see be ow), and that Itl = [9] involves loss of phone miccontras! between Itl and Id/. The phonetic distance bet wecn [a-] and, sa y,[DI] is rather striking and may aiso be of relevance hcre (see point 3below).

We may al so note, as further evidence for the importance of phone-mic alternation in leading to salience in cross-dialectal irnitation. thatthere are a number of other American features that could ha 'e been~alient but do not seem to be. Analysing linguists, for instance , mightcontrast the longer, cIoser realizations of l<elcommon in rnanv varietiesof American English, as in bad [ba-?d], with the more open, shortervariants found in England, as in [bed]. Irnitation by pop singers, how-ever, typical!y does not involve this feature which is, from an EnglishEnglish point of view, purely phonetic.

.If we then accept imitation as a good guide to the degree of salience ofA~,1 pronunclatlon features for English listeners, we can nowmove n o an examination of whether it is in f~ these salient featuresw lC are a so, as \Ve might want to predict. accommodated to whenEng IS nglish s eakers come into contact wlth Amencans.

e ata on which this examination is based is perhaps a littleun usual. There are two main data sets. The first consists of notes madeby myself on the segi. -cntal phonology of native speakers of EnglishEnglish who have be en or are living in the United States. These no esare bascd on informal observations of speakers mostly engaged inacademic occupations, and were often made, 1 confess, at conferencesand during lectures. There are, of course, dangers to be aware of .working with data fror i such a restricted social base but the tes. . ' no esare numerous enough - and contain sufficient observations on. non-

ACCO IMODATION BETWEEN DIALECTS 15

academics - for me to be ieve that this is not a serious cause for concern.The second set consists of observation of what happened to my ownspeech when, as a native speaker of English English, 1 spent ayearliving in the USA. There are of course obvious worries about informal,untaped investications of one's own spee~h. 1 at empted, howe~er., toensure that the qata was as 'clean' as possible by nonng pronuncianonsemployed y me in a relatively unconsci?us ~ay and that .1, as it w~~e,caught myself saying unawares. ~Ma~y linguists are, 1 ?ehevc, fa.m liarwith the phenomenon of realizing tnat they ave said something oflinguistic interest only after they h ve said it.) Lingui stics colleagueswere also kind enough, frorn time to time, to point out Arnericanisrns in

my speech. .'My notes show that i1 is indeed the features smgled out by PO? smgers

_ ~d for the most part no other features - that are modIfied duringacco11'modation. A comparison of imitation by pop singers with accom-modation by expatriates, as far as the fi le main features are COIcerned,sho NS the foll wing:

(1) lai/: [al] > [a·] as in life. This feature o British pop-singing styleis not in imitation of Americans as a whole, but rather of Southernersand/or Black. Manv American Blac s have monophthongal realiz-ations of /ail. This pronunciation is also very widespread in the speech ofWhi es in the American South, In so me areas, such as parts of Virginia,it occurs before voiced consonants or word-finally only, while in otherare as of the South it is found in al! environments.) Inde d, m nyAmerican sinzers who have a di hthonzal pronunciatior of lail in their~ ~speech also adopt the monophthong when singing, in imitation o: Blacksand/or South rners. During mv time in the USA 1 was not in the Southor in close contact with Blacks or Southerners. It is therefore n.Q!

surprisi '5 that 1, like most other Eng ish visitors to America, did fill,tacquire this reature. Nor did an)' of my othe; ¡nf rmants.

p'(2) Ir/: 0 > Ir/le as in cart. During my stay in the USA there we~no signs at al! of any acquisition of non-prevocalic Ir/;, 1did occasionallypronounce Ir/ in this position, but this was done deliberately and con-sciously to avoid confusion between, say, 130b and Barb, rny Englishpronunciation of the latter often being taken by Amcricans as theformer. My notes in fact suggest that the vast majority of non-rhoticadult English English s eakers in the A do not ac uire this featureunti t ey ave een in America for a considerable period (say ten ycarsOr so), if at all. Those that do acquire it certainly acquire otherAmerican English features first. and acquire it, too, in an inconsistentand/or lexically conditioned andJor not entirely accurate manner (se echapter 2). One of my inforrnants, resident in the USA for ten years,

16 ACCOM. OOATIO BETWEE~ DIALEas

as consistent 10 pronouncing Irl only in the vords [or, where, here, andare, and with a couplc of very rare exception p nounced non-pre-vocalic Ir/ nowhere else, not even in aren 't.

It seems then, that Irl, though apparently salient, is not readilaccommodated to. s . ,. othcsis inc t? Theanswer appcars to 'be that, while salience (as indicated by imitatio ) isindeed crucial, as we have argu d, it is not the whole story. Salientfeatmes \ViII be accommodated to unless o her factors inter ene toÜ iav. inhibit or even prevent accommodation. n IS particular casethe i actor wou ear o e a onotactic constraint, Ion-rhotic English English accents, obviously, have a p onotactic rule whichpermits Irl to occur only before a vowei, and prevents its ocurrenc pre-consonantally and pre-pau ally.

Now there is plenty of evidence available to indicat that phonotacticconstraints of thi type are very strong, and e use e nsiderable difficultyin foreign language learning. Broselow (1984), for inst nce, argues thatin second-language acquisition 'syllable structure re trictions are par-ticulariy susceptible to transfer'. and suggests the following syll blestructure transfer hypothesis:

When the target language permits syliable structures which are no!permittcd in the native language, 1 arners : ill make errors whichinvol e altering these structures to those which wou!d be perrnit-ted in the native language.

Thus, English speakers who have no troublc at a1l pronouncing l J 1:1

sing e c. have considerable difficulty \ ith word-initial [1J] in. say, Bur-mese, and con -ert forms such as Nkomo [rjkomo] found in Africinla guages to forms such as [onkoumou] that conform to English pt-terns.

There seems to be no reason at all why thesc difficulties should :.".1¡

also apply to second-dialect acquisi io and to the eccommodacoaprocess. 1 can certainl a test that if 1 \ ant to ronounce sa ar. :lS

Ipartl I find it ver hard to do so in t e ow of con" rsation, and i: isort y 01 note that even those British pop singers who appear to re

trying hardest to imitate American singers ncverthcless rarely achie.ean Irl- pronunciation rate of higher than 50 per cent (see Trudgill, 19~':)even though, one assumcs, thcy are usually performing songs that tz.evhave rehcarscd ano sung many times before. We can clairn, theref :-;.

. tI at ~lthough non-prevocalic Irl is indeed a salient feature of Amen::w~ngJ¡s 1 ~or Eoglish peop e, the phanotactic constraint present in t!1!;;

non-~hotlc acccnt prevents them from accornmodating to Am ri.::mEnghs on thls partIcular feature.

ACCOMMODATION BETWEE' lALECfS 17

(3) 10/: 101 > laJ. A& in there was no trace of any tendency 'n rnyspe cl1 to madify the pronunciatian of hot, top etc. from [hot] to r ot],This is more dif cult to explain, since the chang could be interpreted asbeing a purely phon tic one invoIving no phonotactic constraints. It ispossibIe, however, that the answer to the question of why this modifi-cation was not made lies instead in the notion, wel! known to students ofdialectology, of homonymic clash, English English already has a vowelof the low back unrounded [aJ type in the exical set of heart, park,calm, half cte. It is true that this voweJ in my speech, approximately [a: J,is not absolutely identical with the voweI many Americans have in hot,top etc., approximately [q-~J.But it is close e ough to ea se confu-sion, as in the e se of my Barb being interpre ed as American EnglishBob, mentioned above. Certainly, if I try to say hot i the Ame icanmanner, it feels to me as if I were saying heart. The wholesale adoptionof the Am rican vowel wouId thus have led to the Ioss of contrastbetween pairs su has:

hot heartPO! partcod card etc.

Just as the possibility of the loss of contrast can prevent the occurrenceof sound changes, o apparently can it infiuence accommodation. Inot ier vords, it is precisely the same characteristics of laI a make itsalient, and th refore a candidate for accommodation, that delay(although not prevent - mergers do occur!) its accommodation. Asimilar phenomenon occurs in the Engli h of Belfast (J. MiIroy, per-sonal communication), where speak rs accommodating upw rds do notgenera Jy change tei! [e:] to [el] in lane etc. because [el] already occurs asa realization of ¡al! in Une etc.

There are, however, other factors one should per aps considero Forinstance, t relationship etween English English 101 and US Englis10/ is not entirely straightforward. In rr any varieties of U English ,some words which in English English have 101 actually have 1::11ratherthan 10/: lost, long, off etc. Other words which have 101 in EnglishEnglish have I¡J in US English: oj, what, was, etc. Successful accommo-dation would therefore be a somewhat complex process.

Secondly, Labov has suggested (personal communication) that afurther inhibiting factor in my own case may be that [a] is al so aconservative , rural, low- tatus pronunciation i orfolk, the Engliscounty of which I am a native.

The rcst of my data indicates, in any case, that while Eng . h Englishspeakers d in fact accommo ate on thi feature more readily than they

1

,!..•.

..

18 ACCOMMODATION BETWEEN DlALEcrS

. . f ccommodationdo on Irl, it still takes sorne considerable time be ore abegins.

(4) la:1 < lrel in 4E.nce, last etc. The data indicates that this is a changewhich English English speakers do make reasonably early on, if they aregoing to accommodate in the long term to US Eng1ish. Even speakersfrom the north of England, moreover, can be perccived to acco-: rno-date on this feature. First, the vowel in words such as hal] and can'tchanges from la:1 to lrel and, secondly, the phonetic realization of lrelchanges from [aJ to [Le-re·], e.g. [Iast] > [la-st] last (southern EnglishEnglish [Io:st]). In my own specch there was some trace of accornrno-dation on this feature (see below), though much of it was at a relativelyconscious level and occurred only in certain situations.

This feature would seem to be a very obvious candidate for changeduring accommodation, since it involves a very simple modification.English English speakers aiready have the vowel lrel in their inventory,and it would therefore be a very simple matter to substitute this for la:1and say /dans/ rather than Ida:ns/. Southern English English hasromance /roumzens/, so why 110tIdrensl? It has ant lrent/, so why not plant/plsent/?

It is therefore not easy to explain the delay that occurs in the acqui-sition of this feature amongst those English English speakers whoaccommodate lo US English. Introspection, however, suggests a socio-psychological explanation, at least in my own case. Since this explan-ation stems from introspecrion, it mal' not be applicab!e in other cases,although informal discussions have indicated that other people mayhave the same experience. The explanation lies in the faet that the vowelleel in this lexical set is toa salient an American feature. It is not adoimmediatel because It sounds, and fee1s, toa American. The eIS .00 strof)3. ( y t IS is, it is har to say, but note again thatalfernation between phonemes is involved: see below.)

Other similar phenornena can be noted, even if they have not yetbeen studied in any systematic way. In England, 'Northerners' arestereotyped by 'Southerners' as saying butter etc. as /buta/ rather thanIbAt'J/, and as saying dance Id<ensl rather than Ida:ns/. 'Southerners', onthe other hand, are stereotyped by 'Northerners' as saying Ida:nsl ratherthan Idrens/, while the ~ronuneiation of butter appears to be of relativelylittle significance and is rarely commented on. It is therefore interestingto note that Northerners moving to the South and aceommodating toSouthern speeeh usually modify butter /buto/ to Ib/\t;:,1 or at least t/boto/, but mueh less rarely modify /dams/ to Ida:ns/. Many Northerners,it seems, would rather drop dead than say Ida:ns/: the stereotype thatthis is a Southern form is again toa strong.

ACCOMMODATlON BETWEEN DlALEcrS 19

The argument given above for suggesting that the modification of la:1to te! (and therefore also vice versa) should be an easy one to makebecause of the prior existence of the required phoneme in the system,may in fact be precisely the explanation for why these changes are notmade. If differences between two aceents involve simply the incidence ofa particular phoneme in a given lexical set, then that difference wil! bevery highly salient - and maybe too salient - since speakers are condi-tioned to tune in to features that are phonemic in their own variety.English English speakers are highly aware of US English lrel in dancebeca use they themselves have leel in romance. Southern Eng!ish Englishspeakers are highly aware that Northern English English speakers saybutter Ibutdl because they themselves have lul in Iput/. Northernspeakers are highly aware that Southern speakers say /da.ns/. becausethey themselves have la:1 in ealm, hal], car, banana ete. On the otherhand, they are not so aware of the Southern butter Ib/\!'JI pronuneiationsinee they have no such vowel as IN.

(5) It/: [tJ > [9L My notes indieate that this is a feature which isaccommodated to very early on by many spcakers of English English inNorth America. It is also a modification that took place relative.!yrapidly in my own speeeh - not consistently, but to a considerableextent. This is not difficult to aceount for, especially since the inhibitingfactors we have discussed in (2)-(4) above appear not to be present.First, the cha!1ge is a purely phonetic one involving no phonologicaleomplication~. Intervocalic Itl simply becomes realized as [9J. Secondly,no homonymic clash is involved. For example, in my own speech latterand ladder rernained distinct as [Ireg;:,]and [leed;:,]. (This, of course, is notwhat happens in many genuine American aceents, where the contrastbetween Itl and Idl is neutralized intervocalically, both bei.rg realized as[9]: see above.) Thirdly, the flap [9J is actually already available in mynative accent. (lt is al so common in London varieties of English, as amore formal alternative to [2] for intervocalic It/, and is widespread insouth-western and Welsh (see above) varieties, especially rural dialects,as the most usual realization of this consonant.) In many East Anglianvarieties, there is a phonotactic constrain: (which does not occur in, forexample, London English) whereby a glottal stop may not oceur bothbefore and after an unstressed III or 1;:,/. Thus, while gel is [gE2] and ir is[ltj, and get him is [gE?¡m-gE?:Jm], get ir cannot be *[gE?!?J. In casessuch as these the pronunci+ on has to be [gt:<;!!?J(or the more formal[gttI2-gEtIt]). The fact that tne phone is rcasonably widespread alreadyin some varieties of English English has the consequence that it is nottoo strongly stereotyped as being American. The fact that it is alreadyavailable in my own speech in intervocalic position meant that there was

20 ACCOMMODATION BETWEEN D1ALEC.TS

no difficulty in my extending it to al! intervocalic positions. Fin~lIy, it isa1so worth noting that the pron -nciation o: intervocalic Itl 111 manyBritish English accents - iudecd i.icrcasia :'. in al! acccnts except thoseof tne north-west, the west midlands, the,~ '..' -west and most of W~les,and high-status accents everywhere _ ,.> become problem,aheal.Sp~akers can either select the variant [t; . 1'::11 is sociai.ly ~arkc~ 7sbeing careful, formal, posh, upper class '- r", vi [?], which IS sociallymarked as being careless, informal, rough, lo-ver c1ass etc. The use oíthe flap [9] is a convenient way out of having to select a pronunciationvhich is socially markcd in one way or another. (For most speakers, [:;>]as a realization of word-final Itl is not nearly so salient and occurs muchmore frequently and higher up the social scale than the more conspi-cuous intervocalic ItI.)

The overall picture, then, is that the majority of English Englishspeakers accommodating to American English follow exaetly the sameroute. There is no way, of course, of predicting how fastand how farindividuals wil! accommodate, if indeed they accommodate at all. This,we can assume, will depend on a number of factors, including person-ality type. What we can say is that ifthey accomrnodate, they wil! almostcertainly accommodate phonologica!ly by acquiring features in a eertainorder. The order is:

.!

(1) -ltI- > -[9J-(2) la:1 > lrel in dance ete.(3) [o] > [a] in top ete.. {C(4) 0> Irlj_ #

Thus, English people resident in the USA who pronounce top as [thop]will al so certainly have at least some tendency to pronounce dance etc.with lrel, while the reverse is not necessarily the case. Al! my informants,in fact, conform to this pattern, with accommodation to a given featureimplying accommodation also to those features lower on the hierarchy,but no. necessarily to higher features. (One apparent exception to thispattern was an Englishwoman who had lived in the USA for over tenyears and who had non-prevocalic Irl and 101as [o] but who did not havelrel in the lexical set of dance. It emerged, however, that most of hertime in America had been spent in eastern New England where, as inEngland, dance has the vowel of pa c-id not of pat.)

Our hypothesis is therefore confirmed, if in modified formo Accom-modation does indeed take place by the modification of those aspects ofsegmcntal phonology that are salient in the accent to be aceommodatedto. This salience is revealed by what happens during imitation, and canmost likely be mainly accounted for by the involvement of phonemic

..

ACCOMMODATION BETWEEN D1ALECTS 21

contrasts and alternations. There are, however, a number of factorswhich intrude to delay or prevent, to different extents, the acquisition ofpart.cular salient features. The factors include phonotactic constraints inparticular, but also the possibility of homonyrnic clash and strength ofstereotyping. These factors produce, in two-accent contact, a hierarchyof features such that those with thc fewcst or weakest inhibiting Iactorsare aceommodated to first, regardlcss of the actual speed of accomrno-dation of a given individual.

In any examination of the routes followed by individual speakers duringaccommodation, there is another important factor that we have todiscuss. This is a factor which has been of little interest to socialpsychologists but must be of relevance to linguists: the need to 0eünderstood. We are concerned here, of course, with interactionoetween ieIated varieties where mutual mtelhglblhty is not \ls!!ally asenous or long-term problem. 1.tcan, however, be a short-term nroblem"in some cases, and speakers iñ" this sort of situation rapidly ac ire anawareness that so me features are 1 ely to cause mterlocutors moretrou61e than others (see Haugen"S 1'966 dlScussion of intra-Scandinavian~ÜiñiñuhIca fíon).

This point, and its iruluence on accommodation, has been investi-gated by Shockey (ms.) in her examination of long-ter m accommo-dation by middle-class Americans living in England to English English,the reverse of the process we have been discussing above. She observesthat the speech of long-term American residents in England is charac-terized by three main modifications:

(1) The pronunciation of loul as in boat becomes fronted from [o-u]to [su], a feature of modern RP. Whether this aspect oí the RPaccent is sa!ient for American speakers to the same extent ascertain other more phonemic features is not clear, as it re?rescntsa modification that is purely phonetic. As such, however, it issubject to no inhibiting factors. (There are, of course, a numberof areas of the USA where front or central realizations such as[0U-SU] occur, particularly in Philadelphia, along th- central eastcoast, and in the inland south, but Shockey's informants all carnefrom the midwest or California and did not have this featurenatively.)

(2) The pronunciation of the vowel of hot, top etc. as rounded [o], asin most British accents, rather than as the unrounded [o] typicalof most American accents. This, of course, is the reverse of the

22 ACCOMMODATION ETWEEN DIALECTS

process that occur during accommodation in the opposite direc-t.on, suggesting that the contrast be ween [o] and [a], which isinde d phonetically one of the sharpest differences between thetwo varietie , is salient for both sets of speakers. Degree ofphonetic distance betwecn phones must sur Iy e a factor con-tributing to salience (see above; and Thelan er, 1979, p. 108).However, unlike the change in the reverse direction, f e changeby American speakers from [o] to [o] produces no Iikelihood ofhomonymic clash.

(3) The intervocalic flap [Q] is modified t [t] in the set of latter and to[d] in the set f ladder. Shockey has some interesting data on thisfeature frorn recordings of her own speech:

percentage [Q] Itl Id!after si. months in England 100 100after three years in England 66 77

She points out that even after three years her scores are higherthan those of her informants (se e below), and suggests thataccommodation must be a slow, ongoing process which is notcompleted for a number of years. Note also that Shockey wasmuch slower in losing flaps than 1 and my English informantswere in acquiring them. This points to another factor whic. mustbe of importance in influencing the rate of accornmodation onparticular featur s: the relative naturainess of a phonetic/ hono-logical change. The voicing of intervocalic voicel ss stops, as inmoving frorn British to American English, is a very well-attested,natural and ph netically rnotivated type of sound change. Thereverse process, as in moving from American to English English,whereby voiced stops becorne voiceless in intervocalic contexts, isneither natural nor we!! known as a Iinguistic change. It istherefore not surprising if English-to-Arnerican ae ommodationtakes place much earlier with resp ct to this feature thanArnerican-to-English aecommodation.

ow Shockey's al alysis of tape-recorded interviews with her infor-mants shows that all of them are variable with respect to this feature,and interestingly tha , as in her own speech, Itl and Id! are affecteddifferently:

percentage ftapsinformant 1inforrnant 2informant 3inforrnant 4

F aps have been reduced, as a result

1."

/tI Id!17 6137 5841 6739 68

of accommodation to English

ACCO MODATION BETWEEN DlALECTS 23

English, from a presurncd original score of 100 per cent in both cases,but the reduction is much greater in the case of Itl han in th case of Id/.

Shockey rightly makes the point, in attempting to explain this fact,that students of accomm dation must recog iz that, in a d·::o to thesociopsychoiogical factor hicn lie at the root oí accommodation uchas the desire not to be too 1 ieren , e eSlre to e 1l1te I ible is alsoan Important factor. Amencan an Tl.IS nglish, particularly themore s an ar varieties, are very readily mutually intelligible, but diffi-cuities o a ise f om tim to time. Shockey points out t at cornprehen-sion of TV programmes fram across the Atlanti e ofter reíie on context.It is, moreover, in situations where no context is provided (and wherethe Iistener has not had time to work out which variety t~ e speake isusing) that misunderstanding occurs. These situations are often servieeeneounters. Shockey reports that vowel differences have led to herreceiving cherrics (EngEng [éE1IZ)) in Engl nd whe she asked forcarrots AmEng [ksrots], EngEng [kreratsj). She a!so reports, howev r,that it is the flapping of i tervocalic Itl which seems t cause Britishlisteners the greatest compre, ension difficulties. Flapping of Id/, on theother hand, i much less of a probiem beca use of the close phone icsimilarity of American [9] and Engiish [d]. The desire to make one elfmore easily understood is therefore at least art responsibl for the

1 erentlal modific tion durin accommodation of Id! and IU.i ere IS also evide ce for the obvious e ect of cornprehension as a

factor in accommod ... : n to American English y speakers of EnglishEnglish. 1 can attest that one factor that witho t doubt preeipitated theintroduetion of flaps into my own sp ech in America was the number ofpeople who thought, for example, if only for a second, that 1 wan ed apizza rather than that my name was Peter. And, while 1 did notgenerally cha g la:1 to I~! in the exical set of dance ete., 1 did end upsaying words such as g/ass, hal], and bathroom with I~! i, serviceencounters in s ops, bars, and restaurants, in order to avoid exchangesof the type below:

Waiter: Would you care for another ott of wine?Author: A ha f bottle, pIease.Vaiter: Coffee?

The problem was of course that the la:1 in half sounded to the waitermore like his own vowel in coffee than the expccted /~I vowel of half.

The accommodatio process

\Ve ~ave argued that, ~t east in contact between American and EnglishEnglish, accornmodation follows a fix route. Jf it is the ea e that

24 ACCOMMODATION BETWEEN DIALEcrs ACCOMMODATION BETWEEN DlAlECTS 25

Nordenstam fin~ it is at th o· 1 level that accommodation ¿'begins rst. his is als obviously the case with EñgIish/ Americanae ommodation. It is also clear why this is the case. Lexical differencesare highly salient, and are readily apparent te all speakers of mévarieties concemed without an lin UIStlCtraining or analysls. They arealso mostly non-systematic, and susce tI o emg eamed one at atime. rucia y, t e can aJso cause severe an o VIOUS,comprehenslOndifficu tieso Indeed, in both Scandinavia and the English-speaking worldthere is a fund of folk knowledge about le .ical differences which isshared by most adults. It is widely known in Britain, for instance , thatcertain lexical iterns and phrases are to be avoided when talking toAmericans, e.g. rubber (EngEng 'eraser', USEng 'condom'); te knockup (EngEng 'to awaken y knocking', USEng 'to mak pregn~' ,. It issirnilarly widely known in Scandinavia that e.g. ro .g p"'eans 'pe ceful' inNorwegian but 'arnusing' in Swedish. There are also, of course, manyother differences that are not known, but these are gene rally soon le rntwhen the new variety is encountered (unless ambiguity is possible, e.g.pavement (USEng 'roadway', Engf.ng 'sidewalk'j).

Jn T rdenstar;1's study, lexical accommodation is followed by mor-phological accornmodatlOn. This is not the case with En lisrJAmericanacc mmo ation, of course, where phono oglca accommodation comes~ext. E"nglish Engli h speakers in the USA, for instance, may end upusing forms such as gotten and dove (for dived), but this is usuallypreceded by at least so me phonological modifications. We can probablyascri e the situation described by Nordenstam to the far greatersalience , due in turn to greater frequency, of morphological differe .cesbetween Norwegian and Swedish, and/or to the relative lack of phono-logical uniformity within and differentiation between Swedish andNorwegian.

At a number of points. Nordenstam's data s ows that h r Swedishsubjects do indeed follow a regular and common route towards Nor-wegian during morphological accommodation. The majo ity of herinformant , as the implicational scale of tab!e 1.1 shows, acquire Nor-wegian-style pronouns in the following order. First, Swedish jag fju! 'I'is replaced by Norwegian jeg /jei/. Secor-dly, Swedish dom 'they'is replaced by Norwegian de /di:/. Thirdly, Swedish honom 'him' isreplaced by hamo And finally, Swedish ni 'you (plural)' gives way to derefde:r:J/. (Many of the other pronominal forms are identical or verysimilar, such as vi 'we", hon (Swedish)/hun (Norwegian) 'she'.) In the88-eell table, only four are 'ineorrectly' ordered, although i must beconceded that eight of the informants show no accommodation at a11,soperhap we should say four out of 56. It is difficult, in view of the

rcgularities of th., sort are to be found in othcr accomm dation situ-ations, then this opens up the possibility not only that 'e will be ableto make sensible generalizations about the accommodation proc ss a awbole, but also that it might be po sible, gi en a comparison of twovarieties, to predict what for 1 accommodation between the \ViII take.If this is so, then .i.!Jnight even be possible o predict and ·cxplain whirofeatures will survive, or not, in dialect contact and dialect mixturesituations also (see chapter 3).

Further evidence on the regula ity of the accommodation processcomes from the work of Nordenstam (1979). Nord nstam has examinelong-term linguistic accommodation by Swedish wornen living inBergen, Norway, to Norwegian. This is a situation somewhat compar-able to that of British speakers residing in the USA. Swedish andNorwegian have a very hrgh degree of mutual intelligibility, and Swedesdo not for the most part need to modify their speech greatly whencommunicating with Norwegians in order to be unc!erstood. However, itis clear that t e degree of intelligibility (see Haugen, 1966) depends on anum er of factors - the variety of SwedishINorwegian spoken, thedegree of edueation, the d gree of willingness to ccrnmunicate , and soon - and is probably somewhat smalI r than that between at leaststandard American and English English. It is al so apparent that the faetthat Norwegian and Swedish are two autonomous, separa te languages -and are perceived as such by their speakers - is of some consequence.Some of the Swedes studied by ordenstam, for exam le, were clearlyattempting to keep the two languages apart and become bilinguai,rather than introduce Norwegian features piecerneal into their Swedish.This does not normalIy happen within the English-speaking world,except at times in the case of bidialectal children, since there is noperception that, say, American and English standard English are di -ci etely autonornous varieties and that they therefore ought to be keptapart. Rather, the autonomy is shared (see Chambers and Trudgill,1980). .

Nordenstam's study is mainly lexical and morphological, and indee9jjis at these two levels that the two languages differ most. (Syntacticalffcrences are very few, and pronunciation differences between thetwo, though clear enough to most Scandinavians, are probably nogreater tha differences within the two languages.) This contrasts withdifferences between English and American English, where there arehardly any morp ological differences (and what there are are mostlyt ndencies rather than absolute differences); a number of importantsyntactic and phonological differenees; and a very considerable numberof le .ical differences (se e Trudgill and Hannah, 1982).

,.f>

..

26 ACCOMMODATION BETWEEN DlALECTS ACCOMMODATlON BETWEEN UIAL.c"".J _.

similarity or identity of t e other forms in thc system, to at nmpt toexplain this ordering. But the salience of the first-person singular is notentirely un xpected, particularly since the phonetic fo m of the' wedishjag could be interpretccl by Norwegians in some contexts as ja 'yes',while the delay in acquiring dere could well be due to the fact that ni isthe polite prono un of address in Swedish.

farligt > del el' [arlig, 1 . B rgen Norwegian, moreover: plural ad jecti~estake -e nly in attributive position. Thu accornmodation from Swedishrequires dom (ir fina> de er fin.

Table 1.2 Norwegian and Swedish adjecti al agreement

Adverb Neuter adj. Pred. a j, pl.-igl-igt -igl-igl fíJl-e-a

Fanny N SJenny N N NKatarina S N NBodil N NEva N S NBlenda N N NCharlotte N NHenny S N SCarin N S NStina N N NBarbro N S NLisbeth N N NAlma N S NNancy S S NEr a N N SElien N S SInez S S SHelen N N S10na N N S

Nina S S NLinda N SLena S S S

Source: Nordenstam, 1979.

Taole 1.1 Norwegian nd Swedish pronouns

jegljag deldom hamlhonom derelni

Fanny N N NJenny N N NKatarina N NBadil N N N --srEva N > .lL S¡.

Blenda N N S ~Charlotte N N S NHenny N N S -S--Carin N S SStina N S ~ SBarbro N S -S- - SLisbeth S S SAlma S S SNa ley Ji S S SErna S S S SEl en S S S SInez S S S SHelen S S S SMona S S S SNina S S S SLinda S S S SLena S S S S

Source: Nordenstarn, 1979..,•

At a number of other points, on the other and, it is difficuit to findany regularity at all. This can be illustrated by table 1.2. Both Nor-wegian and wedish express adjectival agreement by suffixing -t toneuter adjectives. lural adjectives tal e Swedish -a, Norwegian -e. Thesuffix -a/-e also occurs in th definite singular. e.g. den store mannen 'thebig man'. The neuter forms of adjectives also function as adverbs. Thereis, however , a clifference concerning adjcctives with the common ending-ig, e.g. Norwegian [arlig 'dangerous', fattig 'poor', etc. In S vedi hthese are treated like any other adjective. In Norwegian, on the otherhand, they do not take neuter -t; thus accommodation involves del ár

Table 1.2 shows that those speakers who have accornmodated most toN rwegian in tabl 1.1 are aIso for the most part those who haveaccommodated most here, and vice -ersa. However, there is no way inwhich tabIe 1.2 can be reordered into anything approaching an irnplica-tional sca!e. There is no regularity h, re. It is perfectly possible, ofcourse, that we are grouping together three features which should notbe grouped together, but there are in Nordenstam's work a nu ber ofother points at which the same type of phenomenon occurs. In fact,muc of her data suggests quite strongly that, while there are constraintsand regularities in linguistic accornrnodation, there is also, as in childlanguage acquisition and in second-Ianguage learning, plenty of room..

Irregulari y in accommodation

ACCOMMODATION BETWEEN DlALECfS 29

Table 1.3 Main consonantal and vocalic modifications

Englis (Reading) Australian

1 -ltI- better [th 1 [rJ2 lail high [~!] [a·i]3 loul low [:}u] [re·U]

4 leiJ face [El1 [re·i]5 li:1 see [ti ] [dl1

6 IAI but [d] [2/]7 10:1 part [g:] [a:1

8 lu:1 boot [t¡:] [\1 ~]9 [E1 bed [E] [e]

10 [cu] how [~u) [E·U]

11 -ltI get 2J [th]

12 lrel bat [re] [E]

13 IEdl there [E:] le:}

14 Ii/ David Ir! Idl

15 Irl hit [11 [iJ

So urce: Trudgill, 1982.

Table 1.4 Richard

Month ·itl· /ail /oul íciJ li:/ iN la:! lu:/ Id /aul -ltI /3!f IE';}I fII II/

1 A AB AB B B B B B B B B B B B B

2 A AB AB AB A A"u AB B B B B B B B B

3 A A AB AB AB AB AB AB AB B B B B B B

4 A A A A AB AB AB A AB AB B B B B B

5 A A A A A A AB A AB A A A(B) B B B

6 A A A A A A AB A AB A A A(B) B B B

A: AustralianB: BritishAB: both formsA(B), B(A): one instance of form in parenthesesSo urce: Trud;ill, 1 82.

28 ACCOMMODATIO· RETWEE DIALECfS

for individual strategies. This is qui e comforting, in a way, but disturb-ing for o' r hypothesis that accommodation takes place by means of afixed route.

It could be cIaimed, oí course, tilat morphelogy and honolo!! areIke!y to ehave differentiv in accommodation. Unfortunately for our

fixe -route h. ·pothesls, however, t ere is somc evidenee that even inphonology reg arity is not (he whole story. For xample, we have dataon long-te m linguistic aceommodation by children which shows vcrydearly the extent to which individual routes can be followed. Theevidence is alJ the more striking because it comes from the Iinguisticbehaviour of twins. The data is as fo lows.

Debbie and Rie ard were born and grew up in Britain. At the age ofseven they went with their parents fron Reading, in the south ofEngland, wh ....re they had iived for a number of years, to Australia,where they stayed for one year before r turning horne. In Australia,recordings were rnade of their speech at monthly intervals for sixmonths by Inge Rogers of Macquarie University, and these reco dingswere subsequently kindiy rnade available to me.

The recordings make it possible to carry out a longitudinal study ofthe accomrnodation process through which the twins adapted theirReading phonologj to that of Australian English. (Doubtless lexi-cal accomrnodation occurred also. Gramrnatical differences betweenAustralian and English English are so few as to be irnpossible to study inthis way.)

ogers (1981) showed that the twins quite rapidly acquired the dis-tinctively Australian high-rising statement intonation. My ownresearches (Trudgill, 1982) investigated their accommodation at thelevel of segmental phonology. The main consonantal and vocalie featuresmodified by the twins •.íuring the six-rnonth period were as in tab e 1.3.

Table 1.4 shows Richard's development over the six-mo th periodoNote the very regular pattern, and the almost entirely perfect impli-cational scaling. Table 1.5 shows the long-terrn aeeommodation byRichard's twin sister Debbie. The contrast is quite striking. First,Debbie has been much less regular than Richard. Secondly, the routesthe two <hildren have followed to acquiring an Austra ian accent have inmany respects been rather different. After six months they sound, atleast to a non-Australi: .1, very Australian, but they have got to thisstage via different paths. Moreover, even though shc got off to a slower

start, Debbie has acquired some Australian features that Ri~hard hasnoto The extent of this differenee is iIlustrated in table 1.6, which showsthe first month of acquisition by both children of eaeh feature.

It is of course possible to attempt to aceount for the different rate ofaccommodation by the children by noting the sex difference and obse.n:-ins that, during their stay in Australia, the children's friends and acti 1-

ties differed quite considerably - as did their personalities.The diff rent routes they followed during aceommodation, how ver,

are more troubling. The fact that t e order of aequisition of Australia

..

30 ACCOMMODATION BETWEEN D1ALECfS

Table 1.5 Debbie

Month -tü- lail loul leif li:1 ItJ 10:1 tU'1 Id lauJ -/ti lorJ i€~1 ff/ III

1 B B 13 B(ó) B B B B B t,B B B B B B

2 B(t,) B 13 B B B B(A) B B B Be) B B B B

3 B B A B B B AB B B A B B B B B

4 B AB A B B B A B AB A B B B B B

5 B AB A A B B A B A A B B A A(B) B

6 B(A) A A A A A A@ A A (13) B(A) AB A A AB

A: AustralianB: BritishAB: both IorrnsA(B), B(A): one instance of form in parenthescsSource: Trudg: 1, 1982.

ACCOMMODATION BETWEEN DIALECTS 31

Richard, we may retreat furthcr to a position which confines the hypo-thesis to adult , or per .aps more probably t p st-adolescents. ClearIy,accommodation by chiidren may be a very di ferent kind of phenome-non frorn accommodation by adults, This is par icularIy so giv~n theenormously greater lingnistic . exibility of young childre ,especially upto the age oí approximately eight (see below). The speed of accommo-dation is greater, and of course so is thc degree. (Note, in fact, the verylarge number of features accommodated to by the twins as comp. rewith the íour main features we noted for English adults in the USA.)This suggests that thc constraints that delay accommodation by adults,and which thereby lead to the ordering of the acquisition of features, arenot, for children, constraints at all - or at least not eriously so.Therefore, the same phenomenon of ordering does not occur. Just asyoung children are not inhibited by, say, phonotactic constraints inJearning a foreign language, so they are equaJly uni hibited in acquiringa different di .ect. They therefore have much more freedom and scopefor accommodation, and are much Iess likcly to conform to the same

fixed pattern.

Table 1.6 Mont of acquisition

Feature Key word Debbie Richard

1 better 12 high 3 13 lo iV 3 44 [ace 4 1

5 s e 3 2

6 but 4 37 part 5 28 boot 59 bed 5

10 how 6 211 get14 6 212 bat 6 313 there 6 514 David 615 hit 5

Source: Trudgill, 1982.

Limits to accommodation

This discussion of accommodation by young children leads us to anotherimportant and in eresting question, especially since the role of youngchildren rnay be vital in dialect mixture and in new-dialect formation,which we shali be looking at in la er chapters. The question is: what arethe limits on accommodation? Specificaliy, is total accommodation to a"new variety possible in the 10n term?

ow the obvious place to look, if we are concerned with the limits onlong-term accommodation, is precisely the linguistic ehaviour of youngchildren. As we have just noted, children are \Vell known tn b cmore rapid and com tete accommodatOís t, an adu ts. The explanationfor L is may in part be sociopsychological, ut is .rnost certainly mostlylinguistic, and is concerned with the nature of brain developli1ent andthe human I nguage faculty.

In any case, the conventional wisdom is that ~ng children, unlikSadults, are indeed capable of accommodating to a y to the speech o.0eir peers, as eb le an Ric ard seemed to be at least well on theirway to oing. t is a matter of common observation, and has often becnnoted by Labov and others, that children use the dialect and accent oftheir friends, and not those of their parents or teachers. Indeed thismust necessarily have been the case for regionally distinct diaJects tohave survived in the face of geographical mobility .

".•.featl:res was s?mewhat different for the two children is obviousJy aconsiderable difficu ty for the fixed-route hypothesis ju t Nd' ., . ,s as was or-~enstam s data. Obviously, 111 both these cases there seems to be c1earpotent:al for different peakers to adopt different strategies of accorn-modation.f The fixed-route 1ypothesis can, nevertheless, be defended. In therace of Nordenstam's data we are able to retreat to a position whichconfines the hypothesi to phonology. In the case of Debbie and

..

As can be seen, the index is computed in such a way that a speakeradhering strictly to the phonological rule of Canadian Raising, as mostCanadians do, will score O. On the o her hand, any s eaker who cc.i-sistently violated the rule and had open first elements in voiceless e viren-ments, and vice versa, would,' C ambers's calculations, seo 100.

Figure 1.7 shows the raising indcx scores in three speech styles ob-tained by six Toronto adults in tape-recorded interviews. Of particular

Figure l.7 Index scores for Canadian Raising in three speech styles, Taranta(from Chambers, 1980)

intere 't to us are the scores obtained by Mr J. Clearly, Mr J. isdoi g something wrong as far as Canadian Rr ising is concerned. NowChambers indicates that otherwise Mr J. speaks perfectly normalToronto Englis , So why should he have trouble with Canadian Rais-ing? The answer turns out to be that Mr J, was born in New York Cityand moved to Toronto only at the age of 11, Since that time he hasaccornmodated totally to Canadian English - except at this one point,where he mostly gets thir-gs right, but not entirely. W can suggest thatit is the difficulty of mastering the correct phonological constraintsin olved in Canadian Raising that have prevented Mr J. from acquiringthe Toronto allophonic pattern complete y correctly.

Howe ver, it can easily be ar ued that, if we are interested in the limitson accommod~y children. then the age of 11 is simply too late, rbelieve we can agree with this up to a point. Labov (1972) has argued

32 ACCOMMODA TION BETWEEN DIALECTS

There are, of course, qualifications that must obviously be made atthis point. A number of children of parent who speak a varie y differ-ent fro~ that?f the area in which they are living become bidialectal, andspeak like their parents as well as like their pcers; and .attitudinal factmay retard or lirni ccommodatio" Moreover, isolated individuals _~xtrem~ 'Iames' in Labov's sense (1972), wch as the Nathan B. dis-cussed In some detail in Labov (1966 - may be relativelv '

(ou)-O(ou)-l

peer roup pressurc o con orm SiOce they do not have a eer rou , Forexample, j ew roox turne up one inforrnant in his survey of theEnglish spoken in the Merseyside area of England w o had a consider-able nurnber of Scottish features in his speech even though he had livedal! hs life in the Merseyside crea, The explanation for this was that theinformant's mother was Scottish an , crucially, that the family belongedto a c1osed, isolationist religious sect. Other linguists have similaranecdotes - and 1 make no apology for cmploying anecdotes since if, ona particular topic, we have many of them and : ey all point in the samedirection, then we cannot ignore them. 1 have heard recently, forexample, of a child born and raised in Iowa who had a strang foreignaccent; and of a child who had !ived all his life in Florida but whohado a noticeable New York City accent. These people, however, areobviously exccptions. In general we can accept that, at least in mostwestern cultures, children are known, in normal circurnstances, to adaptat least to anoextent to the speech of their peers. (This is not necessarilya cultural un¡ ers~l, however: se Kazazis, 1970 for an important studyof the role of farnily and local pride in inhibiting change in Greece.)

However, we now have some evidence to indicate that while thispiece o~ conventional wisdom is broadly speaking correct, the truepicture ¡S actually a little more cornplicated. The fact is that recentresearch has made available some studies which shov that there arelinguistic limits on the degree of phono!ogica accornrnodation achiev-able even in the case of young children. There are three studies that wecan mention here.

~I-o lOle

(1~ Chambers (1980) examines changes that are taking place in theEnghsh of Tor?nto a~ ~ar ~s the nature of Canadian Raising is con-cerned, Canad:an R3Isl!1g ,IS the characteristic of Canadian Englishwhereby the diphthongs /al/ and /au/ have mid-central first elementsbefore voiceless consonants and open first elernents elsewhere. as in outloud rxut laud] and night time [nxrt tal m] (for further discussion seechapt~r 4), In his, study Chambers shows, amongst other things,' t' atthere IS now considerable franting of the first element of /au/, He alsostudies the degree ~f adhe~e?ce by speakers to the Canadian Raisingpattern by constructmg a rarsmg index for his informants as follows:

..

ACCOMMODA TlON BETIVEE 'DlALECfS 33

before voiceless consonantse.g. out[EU-eu-Au][reu-au-ou)

elsewheree.g. loud, now[au=au=cu][eu=eu+xu]

Mr J

=====:MrsT~MrH_________ c:::;;;;;¡¡¡¡,,===:;;::..-- MrsB}MrsJMrTo

word lis!style

reading pas agestyle

interviewstyle

IjH:/~/tdItdIH:¡~/u:1lu:1lu:/~/ullulIAUI

tune ete.do etc.boot etc.school etc.road etc.pul ete.OlVn etc.

34 ACCOMMODATION DET\VEEN DIALECTS 1 ACCOMMODATION BETWEEN DIALECT· 35

that, while children younger than eight appear to be certain to accorn-modate totally, there can be no assurance that, after the age of eight,children will become otally integrated into a new speech community. 1would also add that, after the ag of 14, one ean be fairly sure that theywill not. The problem years are eight to 14, with the degree of inte-gratio depending on many different social and individual factors.

MEQ> lu:1 MEou> IAUImoan mownnose knowsrose rowssole soul ete.

(2) Some pioneering work in this field by a .ne (1976, 1980) ha:irvíicated that there is a close correlation betweeJl how Id speakers arewhen they move to a n w area and the degree to whieh they accommo-

ate succe y. ore 111eres 1I1gy, owever, her work also showsthat, in some respects, even children of ei ht ears oId may be toa oId toaequire certain linguistie features during 10ng-term aeeommo ation.

Payne's researeh shows that eh¡/dren from New York Clty fáiñilieswho have moved to Philadelphia aeeommodate almost totally to thePhiladelphia sound system after residing there for a while, with theyounger children accommodating more rapidly than the older. Closelinguistic analysis, however, of the ty? we were advocating earlier inthis ehapter, shows that there may be some inadequacies to this accorn-modation. The children now sound as if they come from Philadelphia,but this overall impression masks the fact that they have aetually failedto master a few fine phonological dc.ails. Where the modifieation to bemade is purely phonetic, there are no problems for the children. Fore: ample, the istinctively Philadelphian phonetie realizations of thevowels lou/ as in boat, lu:1 boot, lau/ out, lail bite, andc/oi/ boy are allreadily acquired. However, in some cases where the modifieationsrequired are more complex phonologieally, difficu ties may arise. TheNew York City children, for instance, show no tende cy to merge thevowels of ferry and [urry , as Philadelphia speakers do (and see furtherbelow).

However, it is probably also relevant for hat fol ows that the situationis further e mplicated by the interaction of this contrast with othervowels, especially IH:I and lu/, and o her lcxica set . Anyone wishi g toacquire native-Iike Norwich pronunciation has to note the exi tence ofat least seven different lexical sets (see furthe chapter 3):

(3) Ciearly, then, the more complex the aru->mmodatjon linguistic;..ally, the earlier the child has to be in in order to adapt suceessfully. Justhow ear y spea ers have to begin to aequire certam linguistic ormsturns out, however, to be rather surprising in at least some instanees. Infact, astonishingly eno gh, .there is some evidence to suggest that c!I:.tain t nes of honological differentiation ma never be aceommodatedto successfully, however oun a s eaker may be. le evidence i asfollows.

In the English of Norwich (se e Trudgill, 1974) the originally distinetMiddle English vowels Q and ou have been pr served as distinct, as theyhave also in a number of other (mainly geographically peripher 1) are asof Britain. The distinetion in Norwich English is as follows:

They must learn, that is, that do, for example, can be pronouneed onlyIdH:I = [d3H], while boot can be pranounced either ¡bü:tl or Ibu:tl =[buu'P].

Now, r search that I have carried out into Norwich English (see alsoTrudgill, 1982) indicates that even people who were born and braughtUt' in NOTWich and who otherwise have períect .ocal accents do notcorrectly master the lu:/-lAu/ distinction between moan, mown etc. iftheir parents come from somewhere else, i.e. if their parents do not havea Norwieh accent. (In some ea-es, it seems to be necessary for only themother to have had a non-r orwich aeeent for the distinction not to bemastered. And in one case, the distinction had not been mastered by aspeaker both of whose parents did have a Norwich accent but whohimself had lived away frorn Norwich until the age of eight, bearing outLabov's point above.)

In investigating this phenomenon, inforrnants from Norwieh ageu3~0 were used, since it is possible (see chapter 2) that younger peopleare now losing e lu:/-/Aul distinction as a result of infiuence of theLondon area and frorn RP. And although the researeh was promptedinitially by o servations of natural speeeh, the main evidenee carne fromtests where informants were required to repeat a sentence in 'a proper

G vic accent'. This was neeessary ecause the RP prestige accent, aswe have seen, does not rnake the phonologieal distinetion in question,and 'correction' towards the RP norm is sometimes indulged in by(especially soeially upwardly mobile) Norwich speakers. Absenee of the

..

36 ACCOMMODATION BETWEFN D1ALEcrs

distinction fro their actual speech does not therefore necessarily meaní at they hay not mastered it eorrectly.

Test sentences were of the form: Norwieh City seo red an own goal.When asked to repeat this sentence in a 'proper orwich accent',informants ad no difficulty at al! in cornprehending what was requiredof them, and produced a rendering that was as basilectal as they couldmanage. A!I focused attention on producing City as [si?IÍ] a id iany onproducing Norwieh as [nnnj] or [ncroj] rather than [nnnj]. The point ofinterest for this research, however, was of course the pronunciation ofown goal. Of the ten informants with Norwich parents, all produced thecorrect Norwich pronunciation of own goal /xun gu:l/. Of the ten withnon-Norwich parents, none produced the correct response. In everyother respect their phonetics was perfect, but they all produced fAungxub', with the exception of one inforrr+nt who had some awareness oíthe issue and reported that he was not sure whether goal should be Igu:Uor IgAul/, but that he was 'pretty certain' that it was /gxul¿ (Interest-ingly, while for example (t) as in bettc: is, as we sa-v above (p. 10), asah~n~ variable in Iorwich English and therefore subject lo stylisticvanation, the vowel lu:1 in goal, moan etc. has at least until veryrecen.tly been subject to very little variation on the part of (particularlyworking-class) speakers, and has not bcen at all salient for localspeakers. It is, on the other hand, a feature which non-Iocais oftenc~mment . n, since .t~e contrast between e.g. London [<e'-u] and Nor-\\~lch (uu 11~very striking, and Norwich speakers moving away from thecity are quickly made aware of t iis fact.)

It therefore appears to be the case that, probably because of thecomplex w.ay in which the Norwich phonological system differs frorn?ther, English systems at .this point, speakers are not capable of acquir-mg tne correct underlying phonological distinction unless they areexposed to it from the very beginning, before they hemselves have evenbe~un to s~eak .' Expos~~e to it in the speech o their peers frorn the ageof tour or nve lS, surpnsmg as this may seem, not sufficient.

This finding. frorn Nor:wich English tallies with a finding of Payne(1976: made In fact before '11y own investigations, a!though 1 wasregrettably not aware of this fact). However, her results are perhapss!J.ghtly less surp~ising than the Norwich results, sinee she was dealingwith a new hOUSIng are a with very many in-migrants, while my infor-mant ere all almost entirely surrounded in their early years by localpeoplc: She notes that the linguistic change whereby le! is being raisedphonetically to [¡::g-eg 1 causes particular problems for her New YorkCity family children in Philadelphia. The progressive raising of l<el from[re.] through [E;)] even as far as [ro1 is taking pla e (se e Labov, 1982), at

ACCOMMODATION BETWEEN D1ALECTS 37

least in 1 rban areas, throughout the north-eastern United States, includ-ing Chicago, Detroit, Cleveland, Buffa o, Boston, New York, andPhiladelphia. lt is also spreading from one phonological environment toanother. In Buffalo, for example, raised vowels occur in all environ-ments, while in New York City raising is confined to vowels that occurbefore ImI, Inl, Is/, 18/, /fI, Isl, Ibl, Id/, Ig/. In Philadelphia, the changc hasno progressed so far, and raised vowels occur only before Im/, Inl, Isl,18/, If/, and Id/. (There is also lexical diffusion in Phiiadelphia in the caseof the j--1dl environment, with some words such as bad having clesevowels, others such as dad having non-raised variants.) ayne showsthat her informants ha e variable success in acquiring the correct hila-delphia pattern of l<el -raising, success diminishing with age of arrival inPhiladelphia, and that a11the originally New Yorl City children showsome tendency to have non-Philadelphia raised variants of te! in smash,bag , dad, grab . The only children investigated by Payne ho consis-tently rai se ¡<el in all and only the Philadelp lía environments are pre-cisely hose whose parents themselves came from Philadelphia. Againwe find that a complex phonological distiríction is simply not acquírabled ring accommodation. Speakers appear to have to learn certainphonologi al f at res from their parents. That is to say, there are c\earlimits en phonological "ccommodation, even in the case of children.

Conclusión

We have seen, then, that the quantitative linguistic analysis of theaccommodation rocess is a useful research tool. We have seen, too-;-t iat it IS at least sometimes possible to explain why some features ofsome accents are salient for their speakers andlor for speakers of other-accent . This salience appears to e ue to a number of factors, whichinclude contributi n to phonological contrast, relationship to ortho-graphy, degree of phonetic difference, and different incidence of sharedphonemes. We can, moreover, perhaps reduce these factors to two ,namely degree of phonetic difference and, more irnportantly, surfacephonemic contrast. Other factors presumably remain lO be detected,but in any case the salience of features can often be deterrnined by anexamination of the process of imitation. During accommodation, it isindeed salient features of the target varietythat are adjusted to exce tthat, in t e case of adu ts at east, a nu er o actors combine to delaythis modification to different extents. These factors do not necessarilyapply to the iinguistic behaviour of children. Nor do they ecess rílyapply at linguistic levcls other than the phonological. These factors

..

38 ACCOMMODATION BETWEE DIALECTS

. '.lude p onotaetie eo strain~s,. hom~nymie cIash, and extra-stronpsahenee (both of the latter agai involvíng, typieaIly, surfaee phonemiecontrasts). Other factors, on the other hand, may accelerate accornmo-dation to particular features. Thcse factor- incIude comprehension diffi-culties and p~onological natural.ness. The prescnce of these inhibitingand accelerating factors leads, long-terrn <commodatíon, to fi"prlroutes whereby all speakers accommodati g frorn one particular varietto another, whatev~r th~ir speed of accommodatio , acquire featurc~fra~ .t?e target van~ty 10 the samc order. The greater acquisitionalfíexibi ity of young children means t at the r are not subject to the effectof inhibiting factors to t e same degr e, and that they therefore dernon-str~te greater variety ~n the routes that they folIow during accommo-dation. Even.youn~ chI!dre~, ho ver, are subject to limits on degree ofaccomm~d:-·on,. \~Jt~ certam more complex phonological contrasts andallophonic Ond!tlOnmó patterns not being acquired correctly unlessspeakers have bcen exposed 10 them in the speech of their .arents.

2Dialect Contac

I•

40 DIALECf eONTACT DIALECf eo TACf 41

a ion s of individual words, may be imitated or copied from t levisionor radio (rather than accommodated to). is is today, for instance,1 :obahly the primo ry mechanism for the adoption of Ameri an Englishfeatures into British English. The phonology and grarnmar of moderoBritish English varieties remain a most totally unaffected by AmericanEnglish, and indeed it is probable that, in terms of phonetics a?dphonology, British and American varieties continue to diverge quiterapidly. On the other hand, British English speakers are constantlyacquiring originally American idioms and Jexis. Strang (1970) lists aconsiderable number of items that wer clearly 'Americanisms' in the1930s but which are an integral part of British English today. Theseinclude: bakery, grocery, bingo, cheese-c oth, raincoat, soft drinks,sweater, and toilet, The older British equivalen-e were: baker's shop,grocer' s shop, housey-housey , butter-musli. r, mackintosh, minerals , pull-over, and lavatory . More recent examples indude the following. Fromabout 1970 onwards, British English speakers have increasingly usedhopefully in the American manner, as a sentence adverbial, as inHopefully ir won't rain today. This usage was much attacked by self-appointed guardia n of the purity of British English in the earIy 1970s,but is now very common indeed in the speech of a majority of Britishspeakers. Most British speakers used the word wireless at least until1960, while today nearly everybody says radio. The early 1980s saw the(possibly t :nporary) British a option of the American expressi n awhole new ball game, even though ball game is never used (or evenunderstood proper y) in Britain in its literal sense. And there are alsosigns that the American usage of through, as in Monday through Friday ,is about to begin finding its way into British usage. Very many otherexamples could be giv n. It has to be assumed that radio and especially,television la' ama 'or role in t e diffusion of innovations of t -Ís t, e,t. o h of course written AmerIcan ~n~ ¡sh and face-to-face contaqwith Americans will also be of imp rtanee. However, precisely becaus~face-to-face contact with mericans is a re!ati'¡ely rare event for mostBntons, core phonology and s 'ntax remain uninfluenced.. It is m:por ant to notice, though, that there is one situation wherecore syntax a p ono,ogy can e In .uence by the media. This iswhere, fOI example, ther is considerable linguistic distance between anational standard and local dialec s (such as in Italy), and individualdialect speakers have made a conscious decision to acquire the standard.Then they may use the language of the media élS a model: again, imitationand copying is the mechanrsm Involved, and not accornmodation.

1.•.

-!n any. cas~, we can assume th~t face-to-face interaction ís necessarybeLore dlffuslOn takes place. preclsely because it is only during face-to-f!,ce interaction that accommodation occurs. In other words, the elec-~('nic media are not very instrumental in t le diffusion of iinguisticinriovatíons, in spite of widespread popula notions to the contrary. Thepoint about the TV set is that people, however much they watch andlisten to it, do not tal k to it (and even if they do, it cannot hear theml),with the result that no accommodation takes place. If there should beany doubt about the vital roJe of face-to-face contact in this process, onehas only to observe the geographical patterns associated with linguisticdiffusion. Were nationwide radio and television the major so urce of thisdiffusion, then the whole of Britain would be influeneed by a particularinnovation simultaneously. This of course is not what happens: London-based innovations reaeh Norwieh before they reaeh Sheffield, andSheffield before they reach Newcastle.

'[!lere are, f eourse, exeeptions to this. fertain highlv salient linsuís-tic features, sueh as new words and idioms or fashionable .

..

42 DIALECT CONT ACT DIALECT CONT ACT 43

Table 2.1 Norwich variable (o)

ReadingWord list passage Formal Casual

Class Sex tyle stye speech speech

Middle middle M 000 000 001 003F 000 000 000 000

Lower middle M 004 014 011 055F 000 002 001 008

Upper working M 011 019 044 060F 023 027 068 077

Middle working M 029 02;¡ 064 078F 025 045 071 066

Lower working M 014 050 080 069F 037 062 083 090

Source: Trudgill, 1983.

Cromere

-r::Great Yarrnouth

Lowcstoft

Map 2.1 East Anglian towns

sex differentiation, with working-class males hay! g more [o] thanworking-class fem les, and yet rniddle-class f males having more [o]than males of the same class. The conclusion to e draw frorn this isthat the newer rounde vowel is coming into Norwich English (see map2.1) frorn two different sourccs. First, it is entering as a prestige feature(and herefore a particularly female feature) frorn the RP accent. TheR~-type pronunciation i corningv in the first place, into rniddle-classspeech, precisely because it is rniddle-class Norwich people who havemost face-to-face contact with middle- and upper-class P speakersfrom Norwich and eIsewhere. Secondly, it i also coming into Norwichas a non-prestige feature (and therefore a particularly male feature)

from the working-class acce ts of surrounding areas. The working-classpronunciation is entering Norwich, in the first instance, by means ofworking-class speech, precisely bec use it is working-class Norwichpeople v ho have most face-to-face contact with working-class speakersfrom neighbouring towns.

Now, if it is the case that gcographicaI diffusion rcsults from accom-modation, we would expect the factors note in cnapt r 1 a beingoperative during accommodation to be found also at work in the case ofgeographic: 1 diffusion. In particular, we would expec salient features tobe diffused rather than non-salient f atures. And e would expect so mefeatures to be diffused more quickly than others, depending on thedegree of salience and the number and strength of inhibiting and/oraccelerating factors, as discussed in chapter 1, that are relevant in eachcase. (Geographical diffusion models can, of course, tell us to expectforms to diffuse out, ards fr m Iarge citics such as Phiiadelphia (Labov,1982), Liverpool (Newbrook, 1982), and London (see below). But they

44 DIALECf CONTP,Cf

cannot pre ict which features : ill be diffused, and which not.) Is it thenthe ea e that it actually is salient íeatures which pread most rapidly? Is10/, for example, a salient feature for speaker in East Anglia - and isthat why the pronunciation i~changing?

The case of 10/, in fact, is . viously a r ther com lex one, and we wil!attempt to tackle this kind of problem by exarnir .ig in some detail arange of similar but simpler diffusion phenomena from our research inEast Anglia. The evidence that we shall employ in this examination is asfollows. During the period 1975-7, tape r cordings were made of casualspecch in 21 towns in the English counties of 1 orfolk, Suffolk, andEssex (see map 2.1;, involving 348 individual spea ers. In addition, 60spe .. .ers were recorded in 1968 in the original Norwich study (Trudgill,1974); and 15 teenagers and young adults were recorded in a follow-upstudy in Norwich in 1983. Analysis of these recordings, concentrating onlinguis ic changes taking place in this East Anglian area, have beencarried out employing the apparent-time approach, which compares thesp ech oí younger and older informants.

This analysis reveals that in recent times the di fusión of pronuncia-tion features outwards from London into adjacent arcas of East Angliahas been quite dramatic. The general diffusion of linguistic featuresfrorn London is particularly noticeable in the case of the towns ofColchester, Claeton, and Walton. In these towns the older spea cerssound like East Anglians, s an overall impression, while many youngerspeakers, as is often noted by lay observers, sound like Londoners. (Inac ual faet, close analysis f the speech of these younger people showsthat they do, ho vever, in many cases preserve a number of East Anglianfeatures: see below.)

One well-known and well-studied phonological feature that has beendiffused outwar s from London into East Anglia within the past 150years or so is the loss of Ihl (see l. Milroy, 1983). As has been notedbcfore (se e Trudgill, 1983). Iz-!e sness is well-known not to occur in thetraditionaJ rural accents of East Anglia (see map 2.2). Our researchshows, however , that h-dropping is now a well-established, if variable,featur of working- lass urban speeeh in the entire East Anglian area.The feature is undergoing geographical diffusion outwards frornLondon, and is also spreac!ing into rural varieties even in the north ofthe region. At the moment, Ihl deletion is less frequent in J(ing's Lynn,Great Yarrnouth, and Lowestoft than it is in 1 [orwich , and jess frequentin Norwich than it is in the urban centres further south. (Note that 'lessfrequent' he re means that the feature is found in the speech of fewerindividuals, and that it occurs less often in the speech of dlose who dohaveit.)

,.•.

..

DIALECf CONTAcr 45

G (h) in hammer

Map 2.2 h-pronouncing areas in England (after Survey of English Dialects,Orto n et al., 1962-71)

This widespread diffusion of h-dropping i no surprise. Our discussionof accommodation, and the relationship of aceommodation to diffusion,leads us to regard h-dropping as a clear candidate for this type of rapiddiffusion. f it is indeed featur s which are salient that are accommo-dated to - and thus subsequently diffused - then Ih/ and its absence arecle rly highly salient. In Norwich English itself (Trudgill, 1974) (h) as alinguistic variable is very much a marker (see p. 10), and of course lackof h is a feature which is often commented on unfavourably and overtlyby teachers and others. This salience is obviously due to the phonemiccontrast factor noted in chap er 1, aIlied to the orthography of Englishand social class dialect (see p. 11). In addi ion, it is intere ting to

-46 DIALECT CONT ACT

su] pose that it is recisely those aspects of (h) which lead teachers tonotice and coudernn lack of h that actually speed the diffusion of thissame zero variant of the vr riable (cf. our discussion of American 10/,p. 17). The loss of Ih/ is also undoubtedly acceierated by the phonologi-

lo naturalness of a change that rernoves a glot al fricative frorn theinventory (see Lass, 1984), especially when that consonant has a veryrestricted privilege of oeeurrenee, i.e. syllable-initial only. The diffusionof h-drapping outwards from London, t at is, does nothing to disabuseus of the notion that diffusion results frorn aeeommodation.

We now, t .erefore, turn to an examination of other features under-going diffusion in East Anglia, as revealed in analysis of our tape-recorded data, to see if our hypotheses of diffusion througl accommo-dation and of salience are borne out. Four features stand out as being ofimportance: three are listed in the following paragraphs, and the fourthmerits a special seetion.

DIALECT CONTACT 47

(1) Cons rvativc rural East Anglian accents, at least in the north ofthe area, do not (01' did not) have 'dark l' as a. allophone of 11/; that is,hill, bel! were [hrl], [bel] rather than the more modern [hrl+ri], [bsl]. Onthe other hand, the working-cJass aceents of London and the HomeCoun ies (the counties adjacent to London) vocalize Il/ in the typicaldark 1 environments to give hil!, milk [IOi!], [mro.sk] (see Wells, 1982).Even middle-cJass speakers from these areas usually ha 'e very markedvelarization/pharyngealization and/or lip-rounding of [í].

This London-area treatrnent of [t] has also led to various interestingdevelopments in the vowel system (see Wells, 1982). notably the mergerof vowels before Il/. For many Londoners, pairs such as the followingmay no !onger be distinct:

(a) Distinct allophones of lu:1 and loul before /11 occur in all thetowns investigatc ' except Cromer, Dereham, King's Lynn, GreatYarmout, Lowestoft, and Norwich. In Hadleigh andSt wmarket this feature is eonfined to younger speakers ofapp oximate!y 30 and under.

(b) St ong velarization and labialization, but without completevoealization, occur in Clacton, Walton, Colchester, Wive hoe,Felixstowe, and Sudbury for al! speakers, and for youngerspeakers in Bury, Harwich , Ipswich, Woodbridge, and Hadleigh.

(e) The complete merger of lul and lu:/, and of 101and lo\>ן before IIl,as in pull:pool, doll:dole, has taken place in Clacton and Walton,as well as in the speech of people under 30 or so in Colcl e ter,Wivenhoe, and Fclixstowe, and is variably present in yo mgerSudbury peech.

(d) Complete vocalization of Il/ has occurred only in Clacton, andthere only for some speakers.

doll: dolepull : poolfill : [eel

If, then, we wish to aseribe iffu ion to accommodation, we wouldlike o be in a position to argue that vocalizati n of [!] is for EasAnglians a salient feature of London and Home Counties English. It isnot in faet a linguistic feature that i often commented on overtly byteachers or anybody el e. On the other hand, it is a feature which iswidely imitated when non-Londoners are copying London Eng!ish forhumorous or other purposes. It does not, of course, in its early stagesinvolve loss of surface phonemic contrasts, but in its later stages itcertainly does, leading, as we have seen, to a cornplex series of neutral-izations and the development of a whole new set of diphthongs. Wecannot, therefore , be absolutely convinced that [-vocalization is a fea-ture for which we would have predicted accommodation, but t ere is atleast some reason to suggest that the involv ment of surface phonemiccontrast does lead to a degree of salience. We can also argue for thephonological naturalness of this ehange, ince the vocaliza tion of dark !to an [u]-like vowel (and of c1ear l to [iD is very well attested in theworld's languages.

I.t>

E~cn in midd!e-class speech, moreover, and even if complete vocaliz-ation of [1] does not occur, vowels may have radically different allo-phones before Il/ as compared with elsewhere:

rudecode

rulecoal

[.lu:Ui][buui]

(2) The towns in the northern part of the East Anglian region- King's Lynn, Cromer, Dcreham, Norwich, Great Yarmouth, andLowestoft - have laul as in house as [ses]. All other towns have [EH) or[eu], In the northern towns, the phonological process that \Vells (1982)has labelled smoothing, whereby triphthongs consisting of diphthongsplus shwa become monophthongs, gives lau/ + I-;¡I > [q:], as in tower[tq.], ploughing [plq:n]. In middle-c1ass accent , this vowel is identieal

The interaction of the older East Anglian treatment of 11/with thisnewer London and Home Counties system makes for a complex patternof change as the Home Counties system spreads. The current situationappears to be as follows (see maps 2.3 and 2.4):

..

48 DIALECf CONTACfDIALECf CONT ACf 49

i

(1

\

ostrong velarizationand labializauon

strong velarizationand labialization

iMap 2.3

Map 2.4 /\/, younger speakers, East Anglia

'wrang' direetion poses some problems for an explanation based on adiffusion model that ineorporates distanee and population parameters.It is t se models that tell us to expeet the sta te of affairs that we mo toften find with respeet to geographieal diffus- n in East Angra, namelythat forms spread out from London, which is broadly sp aking to sayfrorn so th to north. It is not, in faet, yet clear why smoothing isspreading southwards, but it is likely that we will be able to seekan explanation in the faet th t smoothing already oceurs not only inNorwieh, for example, but also with certain vowels in London, in themidlands, and in the RP aeeent. Geographieal!y, smoothing may haveoriginated in a num er of different locations.

Whe her smoothing. again, can be assigne a high degree of saliencesuch that e would prediet that it would be accommodated to, andtherefore diffused geographically, is not c1ear. Again, it is not a featurethat appears to attraet much overt eomment. On the other hand, it isprobable that in Norwieh English at least it is a linguistie variable of themarker type. And once again it certainly does in olve surface phonemic

with thc 10:1 of arm, path, making tower and lar homophonous. Work-ing-class aeeents, on the other hand, have 10:1 as [a: J, and thereforeretain a distinetion between tar and tower even when smoothing hastaken place.

In faet, in these northern East Anglian tow is, smoothing is perhapsm re widespread than anywhere else in England, involving not on.y [a:]in tower and [a:] in fire, but also produeing player as [pla.], going as[go:n], seeing as [ss:n], lower as [lo:], and doing as [ds:n]. It is probablyalso part of a wider proeess that deletes post-vocalic l'dl, as in rhere [o E;)]> [ós.], sure [SH'd} > [sa.].

As map 2.5 shows, the smoothing of lou'dl to 10:1 and of othertriphthor ; is the only example the East Anglian study threw up of alinguistie change in progre ss that is spreading in a southerly rather thannortherly direetion: while older speakers in Ipswieh, Woodbridge,Stowrnarket, and Hadleig have tower [tEtlO ete., younger speakersvariably have the monophthonga! forms, especially in the lexieal set offire and of sure and there. This diffusion of a Iinguistie innovation in the

•.

50 DlALECf CONTACfDlALECf CONTACT 51

1 lorwich , Cromer, Dereham, Great Yarmouth, Lowestoft, and Stow-market a11 have an RP-like central [B1 for al! age groups. Th otherurban centres in the so th of the region, however, are undergoingchange in the realization of IN, in that the fronting of this vowel, whichis typical of London and Borne Counties speech, is on the increase.These towns, that is, have vowel qualities for iN ranging from [B] to [a!L]depending on the age of the speaker nd the proximity of the town toLondon.

This is a serious problem for the approach we have been adopting.We have no evidence that front realizations of IN are particularly salientfor East Anglian speakers, and there seems to be no particular reasonwhy they should be. The change is a gradual and phonetic one, with nophonemic oppositions involved. And he phonetic distance between thedifferent regional variants would no appe r to be sufficient to drawattention to this vowel. Yet, the evidence is ciear that diffusicn is takingplace. We are forced, therefore, to take the posi ion that in this case anon-salient feature is being di fused and therefore, we assume, beingaccommodated to, as mor northerly East A glian speal ers come intocontact with more sou hern speakers. lf this is the case, then it may bethat the explanation for the succes of this fronting of IN may lie in anaccelerating factor alone , narnely that of phonological naturalness - inthis case of the chain-sl.ift type, having to do with pressures in phono-logical space (Martinet, 1955) (see figure 2.2). That is, the same impetusthat led to tl e beginnings of this change in London itself is sufficientlystrong to encourage its spread geographically also.

Nevertheless, we must concede that any initial optimism about ourability to predict precisely which Iinguistic features wiIl be diffused fromone variety to another is a little dampened by the phenomenon of thefronting of IN in East Anglia. It is therefore comforting o note that theprincipies that we adopted in chapter 1 are of some considerable valuewhen we come to an exarnination of those London features hich couldhave been diffused out into East Anglia but which, so far at least,actually have not been .

Figure 2.1 IAI in East Anglia'..•.,.--,,~v----- ....-), l.l_"' .•....r-•.._ '\.... /_-- ..•._--_/

'J" I

I

t-I

Jí"''',,\ .> ----,-- '\..J '- ¿o

U\I\....._-- ...._-<.~:::::-::::::

•e

Map 2.5 Monophthongization of laudl etc .. East Anglia

contrasts, ince we ge , in northern East Anglia, the following equiva-Iences:

tower lou! + Idl > 10:1 as in tarfire loi! + Idl > 10:1 as in [ardo ir Itl:1 + Idl > 13:1 as in dirt

.1 pure Itldl > /3:1 as in purr.e,

going lu:1 + Idl > /:):1 as in lawn

\3) The boundary between East Anglia, which has IN and lu/ distinct,as 10 cud and could, and the midlands, which in vernacular speech hasonly lul, runs to the west of King's Lynn through the Fenland, close tothe Norfolk-Lincolnshire border (see Ch mbers and Trudgill, 1980).Nevertheless, in our East Anglian data there is a clear phonetic gradicntin the actual realization of the IN vowel (see figure 2.1; map 2.6).Wisbech has [r], and King's Lyr.n [1>],while 01 er rural Norfolk speakersin most of the county have the back vowel [Al (i.e. unrounded [oj).

..

-52 DIALECf CONTACf DIALECf CONTACf 53

(e) l'dl in water as ['d] rather than [¡:¡](d) l'dl in horses, wanted ete. rather than /JI (see also ehapter 4,

p. 135).

These features constitute, for the linguist, striking ifferenccs betweenLondon and East Anglian phonology. They are, on the other hand, veryslow indeed to dif use into East Anglia, if indeed th y diffuse at al!.Note, therefore, that feature (a) is a pure y phonetic difference that wewouId not exp et to be salient, and that it is the same feature that wesaw in chapter 1 to be not even aeeommod ted to within a single speeeheommunity. Featur (b) is similarly a purely p onetie ifferenee of verylow salienee. Features (e) and (d) both involve un ressed syllable butcould be argued to be of so me potential salience, sinee in feature (e)London [e] is identical to East An lian IA/, while in feature (d) alter-nation between two phonernes is involved - normally a sure sign th t adifference will be alient. It is therefore gratifying to note that in bothcases we cal point to the presenee of the same strong inhibiting factorw noted in the case of accommodation in chapter 1: phonotacticconstraints. If London [¡:¡] is indeed identified with East Anglian IN, itcan nevertheless not be tra rsrnitted as such into East Anglian Englishsince IN, one of the hecked vowels that occurs in closed syllabJes only,obviously cannot occur in word-final position. And unstressed III isunlikely, for many East Anglian speakers, to replace I';JI .n it ms such ashorses and wanted since their accents have a phonotactic rule whichallows l'dl as the only vowel which may occur pre-consonantally inunstressed syllables. Thus David la;:eiv';Jd/-/Cle:v';Jd/, village Mi';J]1 etc.Indeed, conservative East Anglian accents, at least in the north of theregion, have a rule which permits l'dl a the only unstressed vowel in anyunstressed syIlable, inc1uding word-final position:

money Im'An'dlvery Iv'Erd!window /w'rndo/Tuesday /t'u.zda/ etc.

Phonotactic constraints may change of course, as they have in oreinnovati g East Anglian accents, vhich now permit final -/i:1 in moneyetc. and -/u:1 in window etc. But in the meantime they may have a veryinhibiting effect on accommodation to other dialects and, as a conse-quence, on the proce s of diffusion.

[y]

~( -, .,.J

\··/· .••••.•• 1

/ _1Il__

(-(\

r " >~ \..O,..,.. ••••• --'\ ~~-¡ '-J ~t)

V\\ \._--- ..•...... ...;.~~;::,

Map 2.6 IAI in East Anglia.

,...

We noted above that in southern East Anglia younger urban speakers's und like Londoners, but that closer linguistic analysis s o ¡S thatthey do preserve a number of non-London, East _' nglian features.These features include:

(a) la:1 inJar ete. as [a:,] rather than [a:] (se e also chapter 4, p. 136)(b) li:! in meat as [ú] rather than [';JI]

Diffusion through accommodation: a ro em

Figure 2.2 Phonological pressures leading io IAI frontingTher is still one featu e subject to diffusion into East Anglia fromLondon that we have not yet discussed. This is the variable merger of 18/

..

_4 DlJ\LECf CONTACf DlALECf CONTACf 55

vith If/, and of 101with idl (word-initiall ,) and Ivl (elsewhere), w ticis a well-known íeatu e of London English (see WeJls, 1982), In theEast Anglian study, these mergers provide a striking and challengingexample of the geographical spr ad of linguistic innovations. In thespeech of informants in the region aged over 30, hese mergers are notfound at all anywhere, except for a small number of speakers in Clacto ,I the speech of informants aged under 25, on the other hand, and moststrikingly in the speech of teenage informants, we find these merg rs inthe accents of all the urban centres except Dereham, Cromer, andKing's Lynn. It is, however, for most speakers a very variable feature,and is much more common with lel than with iol, (It is also relativelyunusual in the speech of middle-class inforrnants.)

The extraordinari!' rapid eo ra hical diffusion of this articularlinguistic ealure i one tllat requires examination an ex lallation.Our ata show that the merger is tota y a sent from the speech of evenl l-year-olds in the 1968 Norwich survey, but that it is very commonindeed in the speech of working-class Ió-year-olds in the 1983 Norwichsurvey. That is, speakers born in 1957 do not nave it at al!, whilespeakers born in 1967 have it extensively (see map 2,7), The probl mth n is thi . If we are claiming that accommodation is crucial to thgeographical diffusion of linguistic .nnovations, and if we are a150claiming that face-to-face interaction i essential for accommodation totake place, then how do "ve explain the prevalence of this merger inN rwich adolescent speech? The London-based innovation is making itsway into Norwich and other East Anglian centres. but it is found for themost part in the speech of exactly those people who, probably, have~.ast face-to-facc contact with Londoners - .amely teenagers. \Ve haveno figures for face-to-face contacts, but it does seem lik ly that conver-sations with the working-class Londoners who have this merger are mostoften earried out by adu/t working-class Norwich people who travel tothe London area or meet Lon oners in the course of their work.

A number of explanations, all of them s eculative, can be advancedfor his phenomenon. For instance, we can argue for the importanc ofattitudina! factors, and c1aim that the desirability of Cockney for adoles-cent males, with its stereotyped image of street-sophisticated toughness,is more important here than accommodation in face-to-face contact.(Casual observers have in fact argued here for the impact of te evisionprogrammes such as the very popular 'Minder ' in which the maincharaeters spea Cockney. If this were the only influence, however. ewould expect to find lel being mergec with Ifl ali over Britain. This isdefinite!y not what we do find , ather we find a c1ear pattern ofgeogr:lphical spread, with towns nearer to London being influenced

\'.•.'--,

\\~.~ , .

r

"I"f\\f

"-,\.",.

~ \1',,-,,;

\.l' ....."")II

(-(\r,,~,\ >~ __-,

....._J J "",,-- ~

"\\I._-,--,_-<e~~~

o

e

olderspeakers

iMap 2,7 Merger of lel and IfJ, East Anglia

before those further away, and those even further away not beinginfluenced at all. Televisión may be part of a 'softening-up' processleading to the adoption of the merger, but it does not cause it.) W,; canalso argue, instead or as wel!, that face-to·face contacts do take place"but perhaps in orwich rather than in London, with tourists, in-migrants or even v¡sitin footbalI supporters bringing new lin uistic, orms in with t ", ne can even argue that the spread of the loss of theIfI-/el contrast might be due to an increasing failure by adults in Nor vichto correct If/ for lel as an infantilismo This in turn would be due toincreasing familiarity with - and therefore increased tolerance of - thisLondon feature n the pan of adults as a result of their face-to-facecontaets with Londoners.

Ho w ever, it eems unlikely that any of these factors on their own canseriously be advanced as the major explanation. In particular, tourists in

lo

DIALECT CONT ACT 5756 DIALECT e 'TACT

Norwich tend to be of the middle-c ass, cathedral-visiting, 18/-pronounc-ing type. and in any case the threshold hypothesis that we developedabove suggests that occasional e ntacts with temporary visitors areunlikely to have any strong influence. It i possible, however, that someor all of these factors in cornbination may be of so me relevance.

There are also, however, two other potentially important ways iwhich East Anglian .cenagers might have extend d face-to-face contactswith speaker of London English withou themselves actually leavingtheir own arca. 1 both cases, the bearers of the London-type forrns arein a very srnall minority, and so we must assume considerable inftuencfrorn attitudin IIactors; but at least we can point to genuine íace-io-facecontact, and thus accommodation. First, we can recall frorn chapter 1that it has now become clcar that there may be speakers who hay livedal! their lives in a particular area who have failed, at some points, toacquire the local accent correctly. We saw that Norwich speakers whoseparents a not natives of the arca fail to acquire the normal moan-mown distinction hat the majority of local people ave. This of courseopens up the possibility that Norwich English will eventual!y lose thisdistinction, not only as a result of accommodation by speakers ospeakers of RP, London, and other external forms of English, but alsothrough accommodation to these 'fifth columnists' who appear to speakthe local dialect, but who in Iact do not exactly do so. We have noevi .....nce that this is what has happened in th e se of [m-thin , but °t i atleast a possibility.

Secondly, we can look at another group of individuals who may havean influence out of all proponion to their percentage in the p pulation.The American lingu ist and dia!ectologist Gary Underwood reports (per-sonal communication) from his childhood in the rural American souththat childri ..n who moved with their families to urban areas such asMemp iis ai.d then returned, say, two years later, having acqui ed theurban dialect , were very inftuential in spreading urban speech forms totheir rural friends. These indivi uals were known and considered still tobe locals, insiders, Their langua e was therefore not ignored or rejectedas being foreign and alíen as it would have been had they been genuineoutsiders. They were therefore accommodated to, particularly sincethev were felt to e more sophisticated than the stay-at-homes. Thesarne point is ma e by the pioneering Norw gian social dialectologistAnders Stcinsholt. In his study of the dialect of Hedrum, southernNorway, and the inftuence on it of the dialect of the neighbouring town

f Larvik, Steinsholt (1962) develops he notion of the sprákmisjoncer or'language missionary'. He writ s (my tran lation):

The urban dialect spreads into Hedrum partly as a result of theinfluence of particular individuals living in different parts of the

area. S . .' indivi uals - we can call them 'lang age missio aries'-may be village people who have been particularly heavilyinfluenced by the urban dialect. The most important languagemissionaries are first the young girl wh e me horne after livi gfor a while in the town, and secondly the whalers.

Factors such as these cannat e incorporu.ed readily into explanatorydiffusion models. They do nevertheless stress the importance of linguis-tic accommodation in the dif usion process. If the attitudinal factors areright, and particularly if individuals are p rceived as being insiders by acertain group of speakers even though they are linguistically distinct,then they can have a considerable linguistic influence through face-to-face contact in spite of being heavily out umbered. This is to say that,while a number of different factors have probably been at wo k inbringing about the drarnatic introduction of the /f/-/81 me ger toNorwich (and other centres), a very important feature may well havebeen the in-migration of a relatively small nurnber of families andir ividuals into the city from the London afea, and the return toNorwich o families temporarily resident in this sarne area. Cert inly,in-rnigration fr m the Horne Counties to Norfolk has been heav . in tirepast 25 years,

The Ift-/el merger are also, of course , not at al! surprising frorn t leperspectiv of saiience and accommodation. The mergers, obviously,involve a loss of contrast between phonemic units (with orthographyperhaps having some influence ~, and as such must be highly salient.There is, it is true, ome possibility of delay due to the inhibitinginftuence of homonymic clash, but the functional load in English of 191and 101 i rather low ( ee Gim on, 1980), and minimal pairs such asthinfin, lather.lava are rather hard to come by. And set against thatthere is the considerable acce erating infiuence of the high degree ofnaturalness of the loss of lei and 10/. Both are, of course , unusual in theworld' languagcs, acquired late by children. and subject to loss orchange in many varieties of English. T ey are ph nologica!ly mar ed.and good candidates for variable merger and eventual loss.

Partí al accommodati on in contact sltuations

We hay n ar uing, then, that accommodation. with its constraintsand herefore its regularities, 1S an essentla part o t e geograp-hica.ldiffusi n of at least hon o ical forms. For a complete understanding,however, of what happens in contact between dialects, it is necessary tonotice an important complication. This is that the linguistic form whichis, as it were, transmitted frorn the ori inatin diaiect, is not '1ecessarily

..

58 DlALECf CONTACf DlALECf CONTACf 59

pit /¡/ lul putpet IEI ItJ butpar 1l1!1 101 pot

Southern accents distinguish could and cud, put and putt. Northernaccents do ot, having lul throughout.

Dialectological research by the Survey of English Dialects based atLeeds University (Orton et al., 19 2-71), and by others (see Chambersand Trudgill, 1980), shows that whiJe there are large areas of orthernand southern England where the fivc- and six-vowel systems respec-tiveIy are found, there is also a tra isition zone of some considera l~ s~lebetween the two wh re intermedia te varieties occu . These are vanetieswhich have the contrast between lul and lA!, but only to a certain extent.The south rn six-vowel system is gradually spreading northwards, andin this transition zone (depending also on phonological environment,frequency of occurrence, formality of style, and so on) so me speakershave transferred or are transferring particular words from the lul pro-nunciation to the lA! pronunciation (see table 2.2). Dialects which are ofthis sort we can caJl mixed dial cts. Clearly, the speakers of thesedialects are not accommodating to the southern vowel system as such,but changing their pronunciations of individual lexical items.

Table 2.2 Transition in mixed dialects

pul bui! pusii but up cup butter love come

Northern u u u u u u u u u

{ ~u u u u u ulA A A

Mixedu u U/A A A A Au

Southern u u u A A A A A A

Sources: Orton el al., 1962-71, and Chambers and Trudgill, 1980.

pit 111 lulpet lE! 1:')1

par la!

put, bpor

Notice that we would expect this change to be spreading northwardrather lowly since, as we saw in chapter 1, the lA! vowel of southernaccents is not especially salient for northern speakers because, f r t ern,it is not involved in any phonological contrast. (The whole change, ofcourse, .onsists of the acquisition of the relevan! contrast.) \Ve wouldtherefore expect relatively little accommodation to occur, and hencerelatively slow diffusion. On the other hand, the large phonetic distancebetween high back rounded lul and low central unrounded lA! will, wewould expect, make for a certain degre of salience, and explain whythe isogloss continues to move northward to the extent it does.

The ..rne sort of process, but in reverse, can be een at work (seeTrudgill, 1983) i 1 the difft-sion of the loss of the moan:mown cont ast(see chapter 1) out from the Lond n area into East Anglia. The contrastwhile southern varieties have the six-vowel system:

..

..Table 2.3 Transition by word transfer

Stage 1 tage 2 Stage 3 Stage 4

road lu:1 lu:1 lu:1 IAuImoan lu:1 lu:1 IAUI h,uIboat lu:1 IAUI lsu! íl\ullow Il\uI /hul Ihul Ihulknow 1t.U1 Il\ul hui Il\ulold IAUI IAuI II\UI Il\uI

Sources: Orton el al., 1962-71, Chambers and Trudgill, 1980.

60 DIALECT CONTACT DlALECT CONTACT 61

Table 2.4 Transition in fudged dialects

put bull pusñ bu! IIp cup butter love come

Northern u u u u u u u u u{~1{ 1{ 1{ 1{ 1{ y 1{ 1{

Fudged u u u 'i y Y 1{ 1\

Soiuhern u u u /1. 1\ 1\ /1. /1. /1.

So urce: Chambers and Trudgill, 1980.

between pair such as nose.knows , sole:soul, road:rowed is disappear-ing, and the way in which it is disappearing in some areas is a mirrorimage of the process illustrated in table 2.2. Working-class speakers inthe southern part of East Anglia, ..5 a result of contact with and¡incomplete) aceomrnodation to speakers of dialect which have thernerger, re effceting the merger i their o n speech by transjerringwords, individually, from the lu:1 set to the ,'Au! set. Table 2.3 sumrnar-izes t e type of diachronic proeess involv d. Stages 2 and 3 repr sentmixed di al cts. F. further exarnples of the same phenomenon, seeMilroy (1978). forms. The lu:1 vowel and the IAul vowel are both gradually modified

phonetically until they eet, as in tabIe 2.5. (In the first instance, asstage 2 shows, the forrns produced may be intermediate between tí oseof the original and target dialects: original [u:] > interme iate [ou] >target [eu], parallel to the [u] > ['d (> [A]) cas above. U tirnately,however, sine this is a merger and not a split as in the lsl-hs! case, theend result, if the process goes to comple ion, may also be a vow 1int rmediate between the origi al dialect's formerly distinct vowels.)Stages 2 and 3 are typical of fudged dialects. Note that fudged diale tsforce a redefinition of lexical diffusion whieh, in that it focuses on thespread of changes through the lexicon, is usually characterized (seeWang, 1969) as being 'phoneticalIy sudden but lexically gradual'.Clearly, fudging is both phonetically and lexically gradual.

--3> Interrnediate forrns

l\ ixed dialects are varieties where accommodation is takin lace butW lere 1 las not gone to comp etlOn. e note now, however, that th reare oth '-r ways in which accommo ation can also be partial. Mixeddialects are lexically partially accommodated. In other v~rieties hichfollowin ham ers see am ers an ru gll , we can ca lfudgeá dialects, the accommo atlOn IS !Deom Jete b be!Da artlalp onetica y. a' IS involve 15 the development in dialect contact oforms that are phoneticaf/y intermediate between those of the riginal

and target dia ects. Table 2.4, for example, shows the sort of situationthat OCCur5 in fudged dialects in the lu/-IA! transition zone betweennorthern and southern England, in which con act between varieties withthc vowel lA! and vari eties with only the vowel lul have given rise to anintermediatc vowel quality [Y] .

Similarly, in the case of the East Anglian moan:mown rnerger, somespeakers, partieularly those in the north of the rea who come frommiddle-class backgrounds and have face-to-fac contaets with RPspeakers, are completing the merger by a process of approximation,which again involves the development of phonetically interrnediate

Table 2.5 Transition by approximation

Stage 1 Stage 2 Stage 3 Stage 4

road lu:! [u:] [ou] leul

moan I :/ [Ou] [eu] leulboat íu:/ [ou] [eu] leul

low Irm! [g ] [eti] leul

know Irm/ [ou] [eu] leul

old /¡m! [eu] [gu] laul

Source: Chambers and Trudgill, 1980.

.'

In the East Anglian case, it is clear why the t 'o different strategíes oftransfer and approximation are emp oyed, and why the two differenttypes of dialect - mixed and fud ed - result. Middle-class East Anglianspeakers are accommodating to oth-r rniddle- or upper-class speakers,including those present in their own cornmunity, ho already have Por near-RP accer t in which the vowel o boat, low ete. is in fact [eu] orsomething close to it. The working-class speakers in the south, on theother hand, are accomrnodating to other working-class speakers resi-dent in geographieally adjacent areas who, as is typical of the Londonregion, have a vowel of thc type [tm- u] in boat, low etc. In bot : cases

62 DIALECf CONTACf DIALECf CONTACf 63

the end result is phono ogical: a mcrger of two formerly istinct vowelsand, at least in the next generatio, of speaker, the reduction of theinventory of vowels by one. The impetus for the el ange is al so phono-logical: accornmodation takes place because this feature has to do withphonemic contra t and is therefor salien. 'ut in both cases, theimmcdiate motivation is phoi eti - the acquis.tion of a pronunciation ofa particular word (and, subsequently, an increasingly large group ofwords) phonetically similar to that of the target accent. This motivationwill, of course, also be operati e even in cases where no phonologicale l nge resu!ts: there are East Anglia accents, for instance, where lu:1has changed to /eul in the set of boat, road etc. under the infl ience of

Por other varieties, but where fAul in low, know etc. is still distinct.Accornrnodation, therefore, may be incomplete in three different

ways. Speakers may reduce pronunciation di similarities with therspeakers (a) by alternating their own variant of a form with that of theothcr soeakers; (l ) by using the other speakers' variant in some wordsbu! not others (transfer/rnixed dialects); and (e) by using pronunciationsintermediate between those of the vo accents in contact(approxima ionlfudged dialects). Of course, all three rnav occur inconjunction with each other. -

be regarded as being phoneticaily interrnedicte between 10:1 and loml,ut this vowcl did not develop as a result of (social) dialcct contact. It

was already in existence, in words such as Ig0yl 'fun'. What happenedwas simply that words were reallocated to this vowel that formerly had10:1 or Ia:.tl/, and the selection of this vowel took place beca use it wasphonetically intermediate.

The label 'intermediate' can also be a lied to interdiaJect woreforms, SUC1 as tose studied by Rekdal (1971; cited in enás, 1982).Rekdal invcstigated long-term accomrnodation by speakers frorn Sunn-dal, orway, to Os10 Norwegian, after residence in Oslo offrom one tofive years. She noted the occurrence of a number of 'hybrid' forms in thespeech of her informants that are found in neither Os! nar Sunndal'Norwegian. Ex mples incIude:

'to work''the matches'

SunndaIjubl

Ify~tibJ1J

Osio/joba/

Ify~(bngl

interdialect/jubo/

Ir. Hikanl

Inter ialect

Developments of this sort have, of course, long been noted by dialectgeo ;"'phers as occurrincr in gf'ograpbical djalect cautact arcas andresulting in permanent interdialect forms in transition zones. At thelexica evel, for instance, t ere IS the \ 'e - n n dialectexarnple where an area in which 'patato' is Grundbirne 'ground pear' iseparated from an area where it is Erdapfel 'earth apple' by an interven-ing area i which the fa mis Erdbirne. A modern British example of thesame phenomenon is the usage of take away in central and southernEngland to refer to Chinese and other evtablishments from which hotfood can be bought for consumption off the premises. This southernarea of Britain is divided from a northern arca (rnostly Scotland andNorther: Ireland), where the term carry out is used. by an intermediatearea (part of northern England) in which the intermediate form take outis employed.

It is impor ant to note, howe er, that interdialect forms, defined ::lSforms arising out of dialect contact whicn do not occur in the origina.!....chalects tñat are or were 10 contact, do not necessanly have fa b~mtermeamte 10 an sirn le or straightforward way. In the complex serieso interactions that may arise in la ect contact situations, interdia!ectforrns may arise out of accornrnodation 1hat is 'imperfect' in ways othe~,~ñan by slmply being 'ncomp!cte,

A good grammatical example of this ype of accornmodation is pro-vided by Cheshire (1982) on the speech of working-class adolescen s inReading, England. She observes a confusing situation in her tape-recorded data with respect to present-tense forms of the verb do: onefinds in her data not only 1 do and he does, a in standard English, but

.•

64 DIALECT CONTACT DIALECT CONTACT 65

third-person singular and other per ons. Now the final stage of theprocess involves the im orration frorn standard English of this persondistinction: standard English forrns are signalle in the table by thesign :j:. Note that auxiliary dos, whieh oceurs in neither the originalReading dialeet nor standard English, seo res O per cent. Note also thatfirst-, second-, and third-person plural auxiliary do, which occurred inboth dia!ects, scores 99 per cent, while l ird-p rson singular main verbdoes, when combined with the similar forms dos in the same context,scores 86 per eent. The other standard forms - first-, second-, aad [hird-persoi plural main verb do (36 per cent) and third-person singularauxiliary does (32 per cent) - are doing quite well, but non-standardialect forms are doing even better: fi st-, second-, and third-personplural main verb does at 57 per cent, and third-person singular auxiliarydo at 68 per cent, aithough the former, as a result of standard influence,has almost replaced the original form dos (7 per cent). Finally, the 1 percent figure under first-, second-, and third-person plural auxiliary does isprobably so low as to be irnpossible to discuss with any confidence.What, howev r, of the figure of 14 per cent under third-person singularmain verb do? This is a real puzzle because, while it does not occur ineither of the two original dialects, it is nevertheless used 14 times out ofevery 100 by young Reading speakers, thus:

also 1 does and he do, as well as 1 dos and he dos (/du:z1). It does notappear possible to correlate these forms satisfac orily with any socialtactors. Cheshire notes, howe er , that it is sensible to recognize that dois in faet two verbs in English, the main verb and the auxiliary. The sameis true , of course, of have. In Reading English, the non-standard formhas is used with al! persons of the ve b, a. d indeed, as in many othersouth-western dialects, the local dialcct has -s as the marker of thepresent tense throughout the paradigm for al! verbs: l· Izas, we goes, theylikes etc. Note, however, the percentage of non-standard has employedby the three groups of t e agers Cheshire investig ted when tokens ofha ve are divide into auxiliary and main verb:

percentage non-standard 'has'main verb auxiliary

group A 43group B 100 Ogroup C 52 O

The form has, that is, is only used for the ful! verb have. Where have isthc auxiliary, forms without -S occur: ~ e has a good time vs. We've donoir. The sarne hing turns out to be true, although in a rather morecomplicated vay, of do. If we distinguish between main verb andauxiliary categories, and als look separately at scores for third-personsingular, which behaves irregularly in standard English, then Cheshire'sdata gives liS the percentage of do, does, and dos forms given in table2.6.

standard English1do it, do I?He does it, does he?

original Reading1 dos/does it, do I?He dos/does it, o he?

younger Reading1 does/ o it, do I?He do/dosidoes it,do/does he?

Table 2.6 Forms of do in Reading English (per cent) It can be argued, 1 believe, that the form he do . has developed anoccurs as an interdialect formo It is a forro that occurs in neither theoriginal Reading dialect nor in standard English, but arises out ofinteraction between the . It is not really, of course, a fudged or anintermediate form, but it is a form that has arisen out of dialect contact.The mechanism is presumablv hypercorrection or some other form ofhyp radaptation (see below), but straightforward confusion in a rathercomplex situation - involving three forms, only one of which does notoccur in the standard, and a switch-over from an auxiliary/main verbdistinction to a person distinction - cannot altogether be ruled out. Inany case, the main lesson we can draw from this - and it is an importantone, since we shall be dealing in later chapters with dialect mixtureswhere more than two contact varieties are involved and whe genuinelyintermediate forms are therefore less likely - is that dialect ontact víaaccommodation, with or without diffusion, is a complex pracess. Wemust be alert to interaction among dialects, rather than straightfor wardinfiuence, as being instrumental in the development of interdialect.

,1

Main verb Auxiliarydo dos does do dos does

1, 2, 3 plural 36t *7 t57 *99+ O 13 singular 14 *43 t43t "68 O 32tSource: Cheshire. 1982.

We interpret the figures in table 2.6 as follows. The original Readingdialect (and indeed this is confirmed by observations of the speech ofelderly Reading speakers) distinguished between do for all persons asthe auxiliary and dos for all persons as the main verb: he forms labelled* are the original dialect forms. The next stage, represented in the tab eby the sign t. involved the replacement of the dialcct forrn dos by thestandard English form does. Note, however, that we assume at this stagemerely the importation of s andard forms, not function: the distinctionremained one between auxiliary and rnain verb, and not one between

..

66 DlALECf CO TAcrDlALECf CONTACT 67

Hyperdialectísms

Gi 'en that interdialect fo s can rise out of interaction, as well ascompromise, between dialects, 'we may now note further example ofir, eraction of different types, and at diffe ent linguistic levels. Theexample from the grammar of Reading Engli h that we have just . cendiscussing involved contact between social ialects, and t e social diffu-sion of linguistic fonns through accommodation. Equally interesting aresimilar interdialect forms that have arisen out of the geographical diffu-sio. of linguisti features of the sort we discussed earlier in this chapter.

lf we think about this type of diffusion in mi!itary terms, as it is oftentempting to do, then it is perhaps not too fanciful to say that many urbancentres in the south of England are, as it were, under attack linguisti-cally frorn London, Our recent research in Norwich (se e above) hasdemonstrated quite c1carly that London-based forros such as the mergerof IV and 181are making their way into the English spoken there. It alsoshows, however, that in this state of siege a number of speakers ofNorwich English appear to be actively engaged in fighting back. Theyare mostly younger working-class men, and the form their action againstHome Counties and London incursions takes is a interesting one forhistorical linguistics and the study of linguistic cnange generally.

\Ve can perhaps best describe the forro that this linguis ic rearguardaction is taking by the labe! hyperdialectism, Hyperdialectism is a formof hyperadaptation, the best-kno vn form of \'x;hich ¡s, of course, h er-correction. 1-ypercorrecnons consist o at empts to adopt a more pr sti-gíous variety of speech which, throughovergeneralization, eads to t <>

roduction o orms w le o not occur in the target resti e varie . Awell-known ntish example of this is provi e by north of Englandspeakers' attempts to acquire a south oí Englan pronuncia ion:

'correction' Ibutl > Ibll.tI buthypercorreetion Ibubl > /bt·l;gl butcher

In a.n importa~t p~p'.'r, Knowles (1978) has pointed out that hypercor-rection (and this will In fact be true of any form of hyperadaptation) is oft:",o different typ~s. In the first type, .speakers perpetrate hvpercorre.c.::..tlO11Sbecause as It were the do 110tknow . b . their anal ses oft e arget variety are faulty. In the second, speakers do have a correctana ysis of the target variety, but they make mistakes 'in the heat of themament' as performance errors which they may notice and may correctoIn the flow of connected speech, the a 1 a conversion rule in an~correet envirQ!l.ment, Knowles points out that this is particularly likelyto bappen where two tok ns of a segment that is a candidate for changeoccur in close proxirnity, but where only one of them sbould be

changed. In the case of north of England to south of Engla: d adap-tations, examples might include:

gas-f,wskcup-hook

northIg<esm<esklIkuphukl

south/gesmc.sk//kxphuk/

hypercorrect/go:smo:sklIki\phi\kI

and of CO"13e forms such as Igo:sm<eskl and /kuphxk/ may also occur.The hyperdiale .tisms that we are dealing with here ail appear, impor-

tantly, to be of th first, mísanalys.s, type. The for 1 that the I yperdia-!ectism takes in Norwich is as follows. Parallel to the contras! bet .eenEast nglian lu:1 moan and fAul mown (see above), older ari ties ofEast Anglian English also preserve the original Middle English á and aimonophth ng/diphthong contrast as in, for examp e:

daze Ide:zJ = [devz]days Id<eizJ = [deiz]

TI at is, words uch asface, gate, plate, mane, made etc. have le:/, whilewords such as play, way, plain, main, maid ete. have l<eil. The loss ofthis distinction in East Anglia predates the 105s of the hs.l-Is»! distinc-tion considerably, and in Norwich in 1968 (see Trudgill, 1974) it was adistinction that was retained on y vestigial!y, and especiaíly by olderspeakers, although most natives of the city were familiar with thepronunciation. In eed, Kókeritz (1932) pointed out that, of the ruralSuffolk localities he investigated, the dialect 'as spoken by elderlypeople, clearly dis inguishes between words such as name (pronouncedvith .,::] and nail [p onounced with [<el] or [El] which in standard

English are pronounced alik ' (p. 55), but he also poi ted out that thisdistinction, even then, was dying out under the infiuence of RP andCocknej , with younger pea pIe genera!izing [<el-El-el] to both groupsof items. Similarly, in the records made by the American dialectologistGuy Lowman in the 1930s (se e Trudgill, 1974), a vowel of the type [<et]is found throughout. orfolk and Suffolk in eight, pail, they , way, while avowel of the type [e-a+eo+ee] occurs in paper, lane, apron, make etc.However, the word chamber has [as] rather than [e-e] i most of thelocalities, and in the Suffolk village of Martlesham the words bracelet,relations , make, apron all have [<et-a] alternating with [e-o], which islabelled as 'older'. The 1950s Survey of English Dialects Norfolkrecor s, made by W. Nelson Francis (ms.), show many cases of the l<ei/-le:! distinction preservcd, but Francis writes in his notes under thevillage of Ludham that ME {¡ has 'several different variants, perhapsindicative of change - [E-e] no longer than half-long with lax high off-glide - forms with [a»] may show phanemic shift with refiex of ME ai,ei', The extent to which the le:1 vowel had become a relic form in

..

68 DlALEcr CONTACTDlALECT CONTACT 69

Norwich in 1968 is indicated by the fact that it vas used by only 11 out of60 informa ts and that all of them were aged 45 or over.

In spite of this relatively low evel of usage, however, 1 argued in myreport on the 1968 survey (Trudgill, 1974) that native speakers ofNorwich Engli h nevcrtheless had distinct . nderlying vowels for the setsof name and nail, and/or that they had access to so me form of commu-nity diasystem, hich preserved this distinciion. The evidence was, inpart, that speakers who norrnally never made the di tinction were ableto do SO, without error, if e ey wished to do so for humorous or otherpurposes. Indeed, during t e 1968 survey, a number of younger infor-mants who did not have the di tine ion were able to produce it, con .is-tently and correctly, when asked to read aloud a passage as they thoughtolder speakcrs would read it. This distinguished them from outsiderswho, in imitating the local dialect, often introduced the distinctive /e:/vowel into words where it did not belong. As far as local Norwichspeakers were concerned, however, even if II speakers did not makethe surface contrast, they did all have access in. ome sense to a cornmonset of distinct underlying forms.

1 am now persuaded (see Trudgill, 1983) that thi . 'community diasys-tem' view is in any case wrong. But it also appears that the situation inNorwich is now no longer what it was ir. 1968. It is now no longernecessarily the case that members of the local speech comrnunity can bedistinguished from outsiders in their ability to differentiate between thtwo lexical sets. The fact is that a nurnber of 1 orwich speakers -especially, as we saw above, younger working-cJass males - are wusing the vowel le:/ in the wrong lexical set, and employing pronunci-ations such as day Ide:1 etc.

We can suppose a development as follo s. Contact between dialectsís leading to the dying out of original East Anglian forms in the face ofinvading London and standard forros. In this dialect death situationyounger speakers no longer acquire the correct, original, phonologicalvowel distinction. They nevertheless retain a know!edge of phoneticdifferences between the older local dialects. Favourable attitudestowards the old variety and/or unfavourable attitudes towards the newinvading variety lead to the maintenance of the lder phonetic forrna d, crucialIy, its extension into words where it is n01 historicallyjustified. Hyperdialectisms of the type days Ide:zI thus arise out ofdía ect interaction:

arise out of their interaction one with the other. They are thereforeinterdialect forrns.

A very early observation of interdialcct form of this hyperdialectismtype comes frorn the work of the Norwcgian dialectologi t. Amund B.Larsen, who must have been one of the first linguists in the worlcl tohave carried out researc into urban dialects. His publications incIudeKristiana Bymál (the urban dialect of Christiana-Oslo) (1907) and, withother authors, Bergens Bymál (1~11-12) and Stavanger Bymál (1925).1 Larsen (1917) he develops the notion of nabo-opposisjon, lit rally'neighbour opposition', to refer o a type of hyperdialectal phenome-non. He notes the following dialect fo ms in the speech of the inn rSogn area of Norway (see map 2.8):

standardSogn NorwegianIbj rkl /bj0rkl 'birch'19':)tl /r;rtJt/ 'meat'Ism':)!'1 Ism0r/ 'butter'

The Sogn forrns are unusual, unexpected, and impossible to explainhisto ically. Larsen explains their occurrence by pointing out that thereare a larg number of other wo ds wher Sogn dialect (and standard

~~~.~.•··~Hóya~ge:J s/~~~L.•.¿~~~<C'.'ltO~,..§ognefford~ in~r.~~ Sogn.~fj

Berg'~: L ..j1,S4~'- ordaland~rHr~~

I,i

daysdaze

LondonIdrelzlId<r.lzI

NorwicnIdreizl :? Ide:zI/de:zI

The new forms occ r in neither of the two dialects in contact, and yet Map 2.8 Sogn a.id al ingdal. Norway

..

-70 DIALECT eo TACT

Norwegian) have h/ as a normal historical development, and where theneighbouring Halling dialect h s the vowel /0/ (a development whichdoes have an historical explanation). \Ve thu have the development:

Hallin Sognbircn /bjerk/ /bjerk/ ~ /bjork/top /t0p/ /t';)p/

The regularity of the /0/-/,;)/ correspondences between the Sogn and1 = lIing dialects was so salient for Sogn spcakers that they were led toi: troduce the vowel typical of Sogn, as opposed to Hailing, even intolexical ite • s where they did not belong. Interaction between two dia-lects led to the development of forms that did not originally occur ineither of them.

Larsen's paper may be the first report of the phenornenon of hyper-dialectism. Once one has become alerted to this phenomenon, however,it becomes apparent that it is probably a not uncommon consequence ofcertain sorts of dialect contact. 1 cite three more examples from recentwork on dialects of English English.

(1) In south-western English and southern Welsh traditional dialects,there is an interesting aspectual distinction unknown in most othervarieties of English. It is of the f llowing type:

punetual habitual1 went ther last night. 1 did go there every day.1 go to Bristol tomorrow 1 do go there every week

In the habitual forms, the did and do are unstressed, and in fact the do ismost often pronou iced /ddl. (Indeed, it is highly probable that this is theso urce of the /de= da' habitua!Jprogressive marker that is found in theEnglish-based Atlantic Creoles.)

Iha ainen (1976) has shown that in the south-west of Enzland theb ,

habitual/punctual aspect distinction is best preserved in the speech ofolder dialect sp akers. That is o say, very many middle-ag d andyounger speakers no longer make the distinction. We can, once again,assume dialeet contaet in which traditional south-western dialect formsare being replaced by forms from the south-east and/or from the stand-ard. It is therefore interesting to observe that the recent research of BertWeltens has shown (see Edwards et al., 1984) that non-standard pasr-tense forrns of the type 1 did see ir every de; are still widely used by somegroups of younger working-class speakers in the Somerset-Wiltshire area.Weltens (ms.) al so found, however, that the same speakers are alsousing constructions such as 1 did see ir last night. They are, that is, usingthe no -standard habitual forms with punctual meaning. The non-standard south-western grammatical forrn is retained in the face of

DIALECT eONTACT 71

competition from other diale L, v t as a result of contact with thesedialects the e rrect sernantic distinction is no longer retained. The non-standard habitual form hr s been generalized, as a hyperdialec ism, intocontext where the original dialect would have had punctu 1forms suchas 1 seen it last night, It therefore seerns Iikely that if the typical south-western forrns with unstressed do ai d did survive, they may actuallyincrease in frequency but at the expense of the loss of the tradi .ionaldiaiect dis inction. Dialect contact ill have led not to the loss of aparticular dialect form, but to the Ioss of a grammatical distinction.

. (2) Similarly, it is well known that many dia!ects of English haverestored the singular-plural distinction in second-person pronouns. Thisdistinction was lost when originally plural you was extended in oliteusage to the singular and subsequently, except in a number of ruraldialects in Britain, replaced thou altogether. Well-attested exarnples(see Francis, 1967) of plural second-person pronouns (contrasting w. .. \singular you) include you-all, y'all (southern USA); you'uns, youseyins(Scotland and elsewhere); you ... together (East Anglia, e.g. Comeyou on together! = Come on!). Irish English in many of its varieties hasa singular-plural distinction you-youse which is categorical for veryman speakers. (Lesley Mi roy, 1984 reports that she caused confusionby greeting a group of women in Belfast with How are you?) This you-youse distinction is not known in that form anywhere in Britain, exceptwhe e it has been introduced through large-scale immigration fromIreland, such as in G sgow and Liverpool. From the inner-city areas ofLiverpool, however, it has now spread out into the surrounding areas of

rseyside , as have many other features of Liverpoo! English. In thidialect contact situation, however, it is apparent that hyperdialectalusage ha become established. ewbrook (1982) reports that the non-standard, originally plural form youse is now widely used by speakers inthe Mers yside are a as a singular pronoun, as in Hello John, how areyouse? A similar development appears to have taken place in parts ofthe USA (Keith Walters, personal communication) where y'all hasbecome singuiar (as well as plural) for some speaker; (although this hasbeen disputed; see also Spencer, 1975). In both cases, he non-standardform is not only retained but ext nded into grammatical contexts whereit does not be long as a result of dialect contact.

(3) In English accents around the world, a number of interestingphenomena occur concerning non-prevocalic Ir/ - the Irl in the lexical serof cart, ear ete. Some of these phenomena are r lated to dialect contact ,and some noto It is useful to distinguish between these diff re t phcn -mena in as accurate a manner as possib e .

..

72 DlALECf eo TACfDIALECf eONTAcr 73

As is well known, English accer ts fa 1 into two main types withrespect to this feature: the non-rhotic or 'r-less' varieties, which do nothave non-prevocalic Ir/; and the rhotic or 'r-ful' accents, which do (seeWells, 1982; and chapter I}.

The non-rhotic varietics demonstrate the folJowing features:

(a) Linking Ir This is not found in so me varieties oí South Africanand Black American English, but is normal in other r-l ssaccents. Words such as car are pronounced without an Irl exceptwhen followed by another vord or morpheme beginning with avowel. The Irl which occurs in this en ironment is known aslinking Ir!.

(b) Intrusive Ir! Words such as bra are pronounced without an Irlexcept when followed by another word or morpheme beginning\ -ith a vowel. The Irl which occurs in this environrnent is knownas intrusive Irl - 'intrusive' b cause it is not 'historically justified'or present in the orthography. Most accents which have linking Irlalso have intrusive Irl, at least in some environments, but it isregarded as undesirable by s me purists.

The evelopment and occurrence of intrusive Irl is norrnallyexplained in the following ay, Non-rhotic accents are r-lessbecause of a sound change, which appears to have begun in thesouth-east of England, in which Irl was lost before a consonant, asin cart, or before a pause, as in caro In words where a vowelfollowed, such as carry and rat, the Ir! remained. As a con se-quence, words such as car, where the originallrl was word-fin 1,actually acquired wo pronunciations, one without an Irl, as innew car, car port, and one with an Irl, as in cal' insurance. Thesound change

intrusiv Irl and linking Irl insertion rule (they are of course thcsame rule, the ter .inological dis inction being a purely diachro-nic and/or prescriptive one) is e nfine o environments followingtho e vowels. Indeed, south-eastern English English accents cannow be said to have four distinct vowel subsystems:

(i) Those which produce a following Irl when word- r mor-pherne-final and when another vowel follows:

II';JI as in beerIE'd1 bear13:1 [us10:1 cari'J:1 forl'dl letter

Only Ir':11as in idea; 10:1 as in bra, chacha-ing; i'J:I as in law,drawing; and l'dl as in America, banana-isñ can be saíd to produceintrusive Irl, since IE'd1and 13:1deri e only from historical V + ltl .

(ii) 'owels which produce a following Iwl when word- or mor-''', eme-fínal and another vowel follows:

lu:1 as Inlou/lauJ

youknowhow

{cIrl> 0/- ~F

thus led to alternating forms such as Ika:rl and Iko:/, dependingon the environment. This eventually became reinterpretedsynchronically, by analogy, not as a ruJe deleting Irl before aconsonant, but rather as its mirror image - a rule inserting Irlbefore a vowel:

(iii) Vowels which produce a following Ij/:

li:1 as in melei! playloi! líeIJil boy

(iv) Vowe s which cannot occur word-finally - the 'checked'vowels:

0> Irl / _ V

(where preceded by an appropriate vowel - see below). Wordssuch as bra thereby also acquired two pronunciations - one with afinal/r/, as in bra advert, and one without, as in new bra - andthus rhymed with car in al! envir nments.

By the time this change took place, only a certain number ofEnglish vowels occurred before Ir/, and thus the operation of the

/¡I as in pitIsI petlrel patlul putIAI puttIDI pot

As we shall see below, this historical explanation for the develop-ment of intrusive Irl, while surely correct, may not be he wholestory .

..

74 DIALECf CONTACf DIALECf CONTACf 75

(e) Hyperadaptive Irl In hose parts of th USA where the ma jorityof the popula ion are r-Iess but where rhotie accents are held tobe prestigious, such as, increasingly, parts of eastern NewEngland as well as in American Black English, non-prevocalic Irlmay occur in words where it does not 'belong', through hyper-eorrection. Similarly, English pop singers (see Trudgill, 1983)and actors imitating American aceents (and indeed rhotic Britishand Irish aceents) can a1so be hcard to employ 'hyper-AmericanIrl' in these sume lexical sets. The environment in which thioccurs ~~e after the vowels listed in (b )(í) above (or theirAmerican equivalents) for linking and intrusive Irl, but alsobefore a eonsonant, as in dawn, bought, palm, or before a pause,as in law, America etc. That is, Bostonians who say Chinar andJapan are employing an intrusive Irl hich is part of their nativeaccent; while if they say Japan and Chinar, they are indulging inhypercorrection. Similarly, British a tors imitatíng Americans bysaying dawn /do.rn/ are perpetrating hyper-American Ir/.Clearly, hyperadaptive Irl is a dialeet eontaet phenomenon.

The rhotic varieties of English, in their turn, dernonstrate the followingfeatures:

(e) Hyperdialec al Ir! We now return to the subject of hyperdialect-isms. The urvey of English Dia' ...ets (SED) materials (Ortcn etal., 1962-71) show a number oí interesting occurrences of what isobviou Iy hyperdialectal Irl in rhotic areas. This is particularlyclearly illustrated in a number of the maps in the Linguistic Atlasof England (L,'\E) (Orton et al., 1978), one óf which - the mapf r last - i reproduecd here as map 2.9. This shows clearly thatthere is a small area of Shrop hi e where the pronuncia tia n of the

.ord last in a number of localities is not the usual [lest], [last], or[la.st], but [la.Ist]. Map 2.10, the LAE map for arm, shows thatthis sarne area of Shropshire, at the level of traditional ruraldialeet, is right at the boundary between rhotie and no -rhoticareas.

Again, we can assume that the meehanism that is at work hereis hyperadaptation. In the border dialeet contact situation, localspeakers observe that their lo:rl in items such as arm correspondsto neighbouring non-rhotic 10:/. The r-ful pronunciation thereforebecomes a local dialeet symbol, and he use of that pronunciationa way of indicating dialect and local loyalty.

It i also important to observe that hyperdia e tal Irl is notconfined to Shropshire. The SED materials give transcriptionssuch as

i.¡;.

(a) Analogical Ir! In the rhotic aeeents of, for example, the south-west of England, individuallexieal items oecur from time to (mewi h non-prev calic Irl where no Irl wo Id be expected. Thisoccurs with neologisms and proper names as a rcsult of faultyanalyses of eorrespondences b tween rhotic and non-rhoticvarieties. For instanee, the word Dalek from the BBC TVprogramme 'Dr Who' \ as frequently pronounced /da.rlsk/ bychildren fr m the south-west of England who were familiar withthe faet tha RP and south-eastern 10:1 often corresponds tosouth- vestern la:r/. (Similarly, khaki can be heard as Ikarkil inboth Canada and Northern Ireland.) This is again, cJearly, adialect contact phenomenon.

(b) Phonotactic Irl In a number of south-western English cities,incIuding Southampton and Portsmouth, words such as banana;vanilla, America are pronounced with finallrl. This appears to bea ph nomenon different from analogicallrl, since it is widespreadand normal as an integral feature of literate adult speeeh andoccurs in well-established lexical items. Moreover, it occurs onlyin word-final position. Neither is it to be confused with intrusiveIrl, since phonotactic Irl oceurs pre-pausally and pre-consonantally. Note that it is regionally restricted even within therhotic area. We discuss the origin of this feature below.

walkealfstrawdaughter

[wo.Ik][ko.If'+ka.Jf][Sú;):J)[da.Ital-vdo.Ital]

in other r otie/non-rhotic bar er areas of Herefordshire, Mon-mouthshire, Worcestershire, and to a lesser extent Oxfordshire.War ickshire, Berks ire, and Buckingharnshire. It is significantthat there are no such hyperdialectisms in the rhotic heartlandssueh as Devon and Somerset.

Presumably the psychological mechanism involved here is hesame as that dealt with by Labov in his work on MarthasVineyard (1963). As is well known, Labov showed that thoseVineyarders who identified strongly with the island and wished toremain there had more centralized realizations of the first ele-ments of íail and lou/, which -ere typical of the loca! dialeet. thanspeakers who did not so identify. The latter, on the eontrary, hadmore open first elements, typicaI of the mainland. It seerneprobable that loyal Vineyarders not only were not partieipating insound changes of the type [~!l)> [al], but also were aetuallyreversing them, as [ 1] > [~I] .

•.

76 DIALECT eONrACT DIALECT eo fTACT 77

a:

.-'Ir~

\ a

Map 2.9 last (from Linguistic Atlas of England, Orton et al., 1978) Map 2.10 arm (írom Linguistic Atlas of Engl.: d, Orton et al., 1978)

We can regard hyper ialectal Irl on the rhotic side of the rhoticlnon-rho ic border arcas as a way of reacting to and resisting new,non-r otic pronunciations, since it is obvious that throughoutEngland rhotic pri.nunciations are receding quite rapidly in theface of non-rhotic. We can also regard them - since multiplccausation is always likely in linguistic change - as a r sult ofdialect contact lending to a dialect deatn p ocess, with a conse-

I~iqucnt loss of knowledge by local people of how exacrly the local1\ dialect is spoken.

Similar developments are reported to have occurred (KeithWalters, personal comrnunication) in rhotic/non-rhotic borderareas in the United States, such as parts of North Carolina andT xas. In these areas, items such as walk and daughter may be

pronounced vith Ir/. In this case, however, we must note that inmost are as of the USA rhotie pronunciations are more statusfulthan non-rhotic and are spreading at their expense. We cannot,therefore, adopt the 'reaetion' explanation for the occurrence ofthis feature in the USA. It may, in fact, be an example either ofhypercorrect Irl, or of hyperdialeetal Irl, or of both. If it ishyperdialectal Ir!, then it can be due only to the dialect deathfactor.

In any case, it is interesting 10 note that in England it is not justactors, pop singers, and other outsiders who misanalyse theoccurrence of Ir! in rhotic aecents. Local dialect speakers thern-selves, particularly if under attack from outside, may also overdothings in fighting back, and/or may lose track, in a dialect death

•.

78 DIALECT eo TACf

situation, of the rules of their diale t and pro uce hyperdialec-tisms.

DIALECf eONT ACf 79

locality of Weston, Bath, Somerset (now Avon), near Bristol, asoccasionally having [t] after final I-';}/.

It is instructive to attem t to provi e an ex Ianation for the develop-ment of this feature. It is after al! unusual and not repeated, to the bestof my knowlcdge, anywhcre else in the English-speaking worId. (Thereare, it is truc, similar features: speakers of the oId Isle of Wight dialecthave drawling for dr wing, and some USA diaIects have 1 sawl it atherthan 1 saw ir (Erik Fudge, Walter Pitts, personal communication); butthese are linking, sandhi phenomena, whereas t e BristoI 1 is not. TheBristoll, a!though confined to word-final position, does n t depe d onwhether a consonant, vowel, or pause follows.) Altho gh Lis and 1vocalization are very well known indeed in the history of the world'languages, 1 addition is not common, to say the least.

A very plausible explanation for the historical addition of 11/ lies inhypercorrection. WeIls (1982) writes:

Interdialect, then, may take the form of intermediate vowels, hyper-dialectisms, or other form that did not exi t in any of the originalcontact dialects. Note, however, that the notion of interdialect, as thisarises in dialect contact, takes s beyond the notion of accommodatioas such. It is probable, as we have suggested with our 'fighting back'analogy, that it is actually divergence rather than convergence that is therelevant mechanism i the case of hyperdialectisms. As Giles hasargued, speakers who wish to show disapproval of others will make theirspeech more unlike that of their interlocutors. In the case of at leastsome o" the hyperdialectisms cited above, speakers may do this to theextent of introducing elements of the insiders' dialect into environmentsor lexical s ts here they ormerly did not occur. Notice also, however,that divergence , just as much as convergence, affects forms that aresalient, Both of the phonological hyperdi lectisms we have cited involvesurface phonological oppositions: in the Norwich days.daze case, thepresence vs. absence of a contrast; and in the hyperdialectallrl case, thepresence vs. absence of a shared phonological unit.

Long-term hypera ap ation

lntrusive 111is not -andhi phenomenon: it can apply equally to aword which is sentence final or in isolation, and it varies allophoni-cal!y between clear and dark according as the following segment isor is not a vowel. ... Its origin must pres mably líe in hypercor-rection after the !oss of final 11/ after I';}/, a hypothetical ['repg] forapple. When the 11/ was restored under pressure from standardaccents, it was added analogically to al! words ending in [;}J.

As we saw above, hyperdialectisms are but oue manifestation of th~contact phenomenon o yperadaptation, the best-known manifestation~f whIch IS hypercorr ctlOn. The hypercorrections that most oftenattract attention are those of the butcher Ibl\bl type that we mentionedabove, a. d that seem to be either temporary or to affect only indi-viduals. Occasionally, however. jt js c!ear that hypercorrection gives riseto large-scale linguistic change and 'results in interdiaiect forms becorn-In&-an iñfegrai part of a particular dialect. It is possible, for example,that he midwestern USA pronunciation of wash etc. with írsl arase inthis way.

O:1e such origina 'y interdialect phenomcnon in Britain is the 'Bristol1', an accent feature which is wel! known to students of English Englishaccents (see Wells, 1982) and to many English people ge erally. Theterm 'Bristoll' refers to the fact that in the working-class speech of themajor city of Bristol, and in certain irnmediately neighbouring ruraldialects, words such as América, banana, idea are pronounced with afinalll!. That is, ideal and ii'ea, evil and Eva, normal and Norma, aerialand area are hornophonous. This pronunciation feature is referred to ina number of po ular pub!ications (e.g. Robinson, 1971). And theSurvey of EngJish Dialects materials (Orton et al., 1962-71) show the

In other words, 'C have a scenario as follows:

evilEva

llossli .vol/ > li:v';}l

li.v»!

correction> li:v:J1/> li:vdll

This explains this somewhat peculiar development in terms of dialectcontact and, perhaps, imperfect accommodation leading to an interdia-lect formo

There i , however, another fac or whic we ought to acknowledge.Observe, first, that it is possible to point to a number of difficulties withthe hypercorrection cxplanation. One is, of course, that while a numberof Eng ish varieties demonstrate 11/10ss, only Bristol has the Bristol/l/.Why is this? nother i that Bristol English does not have law */b:l/,paw */p::>:lI afte the pattern of wall, pall. It is perfectly possibJe, ofcourse, for a variety to lose 11/ only in final unstressed syllabJes, but mostvarieties that have 111 Joss or vocalization do so in all syllable-finalpositions. Similarly, Bristol English does not have intrusive II! in itemssuch as medicine, cavity, finery etc. after the pattern of meddlesome,[aculty , cavalry, hostelry etc., where again one would expec - although

..

80 OIALECf CONTACf

this do es not inevitably follow ·-lloss to occur if it were occurring word-finally in items such as medal, tackle, ravel, hostel. The facts are,however, that medilcine, cavilty , andfinelry do not occur.

An additional factor that might account for this fact - that hypercor-rection affect d only word-final Ig! - is a phonotactic ene. Nearly all thewords which have the Bristo! Il/ in the Bristol accent are words endingorthographically in -a, and are relatively recent arrivals in English.Many of them are extremely recent, such as Tanzania, Zambia, CocaCola. Others are relatively recent, such as Canada, Arizona. And eventhose which have been established in English for a few hundred years,sueh as idea, India, China, are medieval or post-medieval borrowingsinto English and not part of the indigenous Old English, Scandinavian,or Frenen vocabulary. Now, as these words were being introduced intoEnglish ~nd spr ading from learned into general usage, it is probablethat there as an ar a of south-eastern England where they were notp ionotacti ally odd, since from the seventeenth century or so onwardsvarieties there had already lost final/rl in words like finer /faino/, so thatnew words like Chii,a /óaino/ were no problem. In other parts of eEnglish-sp aking worId, however, where non-prevocalic Irl had notbeen lost, such as western England, Scotland, and North America,words such as China, Canada, America must have been phonotacticallyodd, because there ere no words in the indigenous vocabulary withfinal -/g/#. Different rhotic varieties therefore adopted different meth-ods of adapting these n w words to their phonotactic structure, since, aswe sa v in chapter 1, phonotactic constraints may be powerful anddifficult to overcome. Some of these methods are as follows:

OIALECT CONTACf 81

Nevady = Nevada. Similar pronunciations are also reported fromIreland.

(3) Sccttish va.ieties of English, or at least some of them, a e able toavoid thi problem by employing the vowel of pat word-finally inthese words. This is the result of the fact that all vowels in ScotsEnglish, with the cxception of hl, lel, and IN, are able to occur inopen syllables. Tn-re is, for example, no contrast between thelexical ets Ú1 pul' and pool, with the result that the luI of hoodcan al so occur in who. Similarly, there is no contrast between thesets of COl and caught, so that t I~I of lot can also occur in law.And, finally, there is no contrast between the vowels of the sets ofPam and palm, with the la! of pat occurring also in pa. Thus,words like China may end in -/a/, and words like algebra e n beginand end with the same vowel (Milray, 1981 on Belfast).

(4) Bristol Engli h, in its turn, has accommodated the phonotacti-cally uncomfortable !oan words into it phonotactic system by theaddition of final -/l/. Our argun ent is, in other words, that whilethe initia! ímpetus for the development of the Bristol l wahypercorrection induced by dialect contact, this was reinforced -again noting the value of multiple causation as accounting for vhya particular change, out of al! possible changes, actually tookplace - by the addition over the years to the vocabulary of Englishof words that would, unmodified, have been phonotacticallyacceptable only in non-rhotic accents.

,•

(1) As vee saw above, Wells has reported that so me south-westernEng'ish Eng ish dialects have converted these new words into anacceptable pattern by the addition of phonotactic Irl. A word likeChina is no longer phonotactically difficult because it is pro-nounced Icain'Jr/. Similarly, there are many American varieties(in addition to those where hypercorrect Ir! occurs) where wordssuch as idea are consist ntly pronounc d with Ir/ in a11environ-ments.

(2) There are also many varieties of English where wo d-final -a isrealized as li:1 or /¡I as in very, money. For example, soda iscommonly pranounced Isoudi:1 in rural American dialects, andmany other such words either still preserve -ii:1 in rural non-standard speech, or else formerly had such pronunciations, someof which are still preserved in songs and/or folk memory: Virginny= Virginia, Ameriky = America, and so on. Butters (1980) cites,in Appalachian dialects, extry = extra, soJy = sofa, chiny = china,

There are, of course, some difficulties with these explanations. Wehave , for example, no reasonable way of accounting for the fact that it isonly Bristol English that has solved this prablem in this particular way.And there are difficulties with widespread reports that Bristol Englishhas final -I'dll al so in words such as tango, window, 1 have myself noevidence of thi , and if these forms do occur they may be hyperdialect-isms. t y data, taken frorn tapes supplied by Bristol Broadsides andemployed by them in studies of local folk history, has older Bristolspeakers employing word-final 111in area, Eva, Australia, extra, idea,Victoria, cholera, gala, swastika etc. There is, however, not a singleoccurrence of 111with items such as window, barrow, calico, narrow,borrow, piano, widow, fellow, radio, tallow, beano, potato. It is, how-ever , certainly the case that the name of the town itself used to beBristow, frorn n earlier Brycgstow 'site ofthe bridge' (Ekwall, 1960). Inspite of these difficulties, however, it is c1ear that any explanation forthe development o: the 'BristoII' that did not look some degree to therole of dialect contact would ignore hat is obviously a major causalfactor .

..

82 D1ALECT CONT ACT

Concluslon

3Dial ct Mixture andthe Growth of New Dialects

We have just seen that dialect contact may lead to the development ofinterdialect forms, including intermediate form . \Ve have discussed this

evelopment in atomÍsiic terms,ñoting how ~e process of partia!accommodation ma lead, in phonology, to alternatIOn between vanantpronunciations of the same vowel or consonant,; to eXlcal diffusion;and/or to the rowth of vowels r co sonants that are phonetlcall~mtermediate betwe t e vanants in con act.

We now turn to a more holistic approach to dialect contact pheno-mena, i which we note that dialect mixture ma ive rise to whole new_interdialectal varieties (or interdialects), including new interm diate dia-ec s. ~, emerges t at 1 IS partleularly rewarding to investigate this type

Ofe"velopment in divergent dialect commu ities (see below) and insituations involving dialect transplantation, since in these cases thedegr e oí dialect difference between the varie ies involved tends to begreater than in straightf rwar geographical dif usion and contact inwell-establis ed areas, as diseussed in chapter 2. This is because in thelatter, as a result of perhaps centuries of diffusion, the dialects that arein contact tend to be very similar anyway, with li tle room therefore forv hoJe new intermediate varieties to develop. \Ve aceor ingly now b ginto tackle the problem of new-dialect [ormation by eoncentrating nsituations where transplantation of som form has occurred.

Language transplantation: F onteírico

One situation that makes the point about transplantation and new-dialeet formation very clearly is that which is found in the Brazil-Uruguay border area. On the Iberian peninsula, as is well known, thereis a geographical dialeet eo tinuum (see Matias, 1984; Kurath, 1972)where dialeets of Catalan, Spanish, and Portuguese merge graduallyinto one another, and where the number of 'Ianguages' recog. ized asbeing spoken depends on the number of au onomous, standard varieties

References

Aitken, A. J. (1984) Scottish accents and dial!::s In P T d '11 ( d) L .. h . . . . ru g¡ e. anguagem r e Britisñ lsles. London: Cambridge L''':\ernty Press.

André~en, B. S: (1968.) Pr~-glottalisarion ir. ~glish Standard Pronunciation,Oslo. Norwegian University Press.

Bahr. D. (1974) Standard English und setr.: 5':,..sohiscnen Varianten Munich:Fink. . . '.

Bailey, C. J. (1973) Variation and Linguistic Jr-eory. Washington: CALBaker, S. (1966) The Australian Language 2lxl ed S d C .B . . y ney: urrawong

auer, L: (1979) .~e second Great Vowe ~? Journal of the Intern'ationalPhonetic Associannn 9, 57-66.

Beal, J. (forthcoming) The Grammar of Tyr~ and Northumbrian EnglisbLondon: ESRC. .

Bell, A. (1984) Language style as audienes cesizn. Language in Society 13145-204 - ,

Be6;~;~: J. (1969) On the uniformity of spo~ Australian English. Orbis 18,

Bhatia, .T. (!9~2) Transplanted South Asian ~g~: an overview. Studies inthe Linguistic Sciences 11, 129-34.

Bicke.rton~ D. (1975) Dynamics of a ~ System. London: CambridgeUniversíry Press. .

Bi~~~~o~, D. (~orthcoming) Explaining ~ge transmission'. Paper given. e vmposium on Language Transmissior:t. CARBS 1984.

BI::, ,r (1~75)dOl the origins of Australian proounciati;n. Working Papers of. peec an anguage Research Centre; 1ha¡uarie University 17-27

Bliss, A.. ~1984)English in the south of Ireland. In P. Trudgill (ed.) Lan ~ e inthe Britisñ lsles. London: Cambridge U~ Pr g g

Br dI D ( '. - ess.a ey, . 19~1) ~eglOnal. differences in Anstralian English phonology.Melbourne University Working Papers in l..i.r.g-J.iscics6, 7'>-93.

Brloselow, E. (1984) An investigation of transfec in second language phonologyMLll .

Butters, R. (1980) Unstressed vowels in Aooalachian English A .Speech, 105-10. ..' menean

'1L

REFERE CES 163

Caonella , J. (1981) Mutual infíuence in expressive behaviour. Psvchological5~llctin 89. 101-32.

Charnbers, J, (1973) Canadian raising. Canadian Journal of Linguistics 18,113-35.

Charnbers, J. (1979) Canadian English. In J. Chambers (ed.) The Languages ofCanada. Montreal: Didier.

Charnbers, J. (1980) Linguistic variation and Chornskv's 'hornogeneous sneechcornrnunity'. Papers from the Fourtli Annual Meeting oj the Atlaniic ProvincesLin~uisrics Associaiion. •

Chambers, J. and Hardwick, M. (1985) Dialect homogeneity and incipientvariation. Sheffield Working Papers in Language and Linguisrics 2.

Chambers, J. and Trudgill, P. (1980) Dialectology, London: Cambridge Univer-sity Press.

Chapman, K. (1962) Icelandic-Norwegian Linguistic Relationships. Oslo:Universitetsforlaget.

Cheshire, J. (1982) Variation in an English Dialect. London: Cambridge Univer-sity Press.

CoJiins. J. (1975) The sources of Australian pronunciation. V.·orking Papers o]the Speech and Language Research Centre, Macquarie University , 116.

Cooper, V. (1979) Aspects of St. Kitts-Nevis Creole phonology. Journal oftheCollege of the Virgin Islands 4, 5-22.

Cooper, V. (1980) Basilectal Creole, decreolisation and autonomous languagechange in St. Kitts-Nevis. Princeton University: Ph.D. thesis.

Coupland, N. (1984) Accommodation at work. International Journal of theSociology of Language 4--6,49-70.

Dittman, A. (1962) The relationship between body movements and moods ininterviews. Iournal of Consulting Psychology 26, 480.

Dittman, A. (1972) Developmental factors in conversational behaviour. Journalof Communication 22, 404-23.

Dixon, J. (1822) Narrative of a Voyage to New South Wales and Van Diemen'sLand. Edinburgh: Anderson.

Domingue, N. (1980) Syntactic innovations in Mauritian Bhojpuri. Unpublishedmanuscript.

Domingue, N. (1981) Internal change in a transplanted language. Studies in theLinguistic Sciences 4, 151-9.

Dorian, N. (1973) Grammatical change in a dying dialect. Language 49, 41'>-38.Edwards, V. K., Trudgill, P. and Weltens, B. (1984) The Grammar of English

Dialect. London: ESRC.Ekwall, E. (1960) English Place Names, 4th ed. Oxford: Oxford University

Press.Elizaincín, A. (1973) Algunos aspectos de la sociolingüistica del dialecto fronter-

izo. Montevideo: Instituto Interamerican del Niño.Feldstein, S. (1972) Temporal patterns of dialogue. In A. Siegman and D. Pope

(eds) Studies in Dyadic Communication. New York: Pergamon.Ferguson, C. (1959) Diglossia. Word 15, 325-40.Francis, W. N. (1967) The English Language. London: English Universities

Press.

I.l

164 REFERENCESREFEREI'CES 165

Francis. W. N. (ms.) Field notes for the Survey of English Dialects. Universityof Leeds. .

Gatewood, J. and Rosenwein. R. (1981) lnteractional synchrony: genuine orspurious? A critique of recent research. JournaJ o[ Nonverbal Behaviour 6,12-29.

Gerstaecker, F. (1853) Narrative oi a Journey round the World. London: Hurstand Blackett.

Giles, H. (1973) Accent mobiliry: a model and some data. AnthropologicalLinguistics 15, 87-105.

Giles, H., Taylor, D. and Bourhis, R. (1973) Towards a theory of interpersonalaccommodation through speech: some Canadian data. Language in Society 2,177-92.

Giles, H. and Srnith, P. (1979) Accommodation theory: optimal levels ofconvergence. In H. Giles and R. St Clair (eds) Language and Social Psycho-logy. Oxford: Basil Blackwell.

Gimson, A. (1980) An Introduaion to the Pronunciation of English, 3rd ed.London: Edward Arnold.

Gordon, E. (1983) New Zealand Enzlish pronunciation: an investization intosome early written records. Te Reo-26, 29-+2. -

Gordon, E. and Deverson, T. (1985) New Zealand English. Auckland: Heine-mann.

Gregg, R. (1972) The Scotch-Irish dialect boundaries in Ulster. In M. Wakelin(ed.) Patterns in the Folk Speech of the Britisñ lsles. London: Athlone.

Gregg, R. (1973) The diphthongs oi and oi in Scottish, Scotch-Irish and Cana-dian English. Canadian Ioumal of Linguistics 18, 136--45.

Hammarstrom, C. (1980) Australian English: lts Origin and Status. Hamburg:Buske.

Harris, 1. (1984a) English in the north of Ireland. In P. Trudgill (ed.) Languagein the British lsles. London: Cambridge University Press.

Harris, 1. (1984b) Syntactic variation and dialect divergence. Journal of Linguis-ties 20, 303-28. .

Haugen, E. (1966) Semicommunication: the language gap in Scandinavia.Sociological lnquiry 36, 28G-97. .

Haugen, E. (1976) The Scandinavian Languages. London: Faber.Henry, P. (1957) The Anglo-Irish Dialeet of Nonh Roseommon. Dublin:

University College.Hensey, F. (1972) The Sociolinguistics ofthe Brazilian-Uruguayan Border. The

Hague: Mouton.Hensey, F. (1982) Spanish, Portuguese and Fronteirico: languages in contact in

northern Uruguay. Intemational Iournal of the Sociology of Language 34,7-24.

Hickey, R. (1984) Coronal segments in Irish English. Iournal of Linguistics 20,233-50.

Holm, 1. (1986) Creole Languages. London: Cambridge University Press.Hughes, A. and Trudgill, P. (1979) English Aeeents and Dialects. London:

Edward Amold.Ihalainen, O. (1976) Periphrastic do in affirmative sentences in the dialect of

East Somerset. Neuphilologische Mitteilungen 77, 609-22.

Jaffe. J. and Feldstein. S. (J970) Rhvthms of Dialogue. New York: Academic.Jahr. E. H. (197 ) Min egen svntaks i interviuer med kvmner og menn fra Oslo.

In K. Gregersen (ed.) Papers [rom tlie Fourth Scandinavion COllferellce ofLinguistics. Odense: University Press.

Jahr, E. H. (1979) Er det sánn jeg snakker? In J. Kleiven (ed.) Sprák ogSamfunn. Oslo: Pax.

Jenstad, T. E. (1983) Eit Nytt Sentrumsmal Veks Fram. Trondheim: Tapir.Johnston, P. (198~~ 'ariation in th stand aré Scottish English oí íorningside.

English World-Wide 4, 133-86.loos, M. (1942) A phonological dilernrna in Canadian English. Language 18,

141-4.Kazazis, K. (1970) The relative importance of parents and peers in first-

language acquisition: the case of some Constantinopolitan families in Athens.General Linguistics 10, 111-20.

Kendon, A. (1970) Movement coordination in social interaction. Acta Psyeholo-gica 32, 1OG-25.

Kerswill, P. (1985) A Sociolinguistic Study of Rural Immigrarus in Bergen,Norwav, Carnbridze Universitv. unnublished Ph.D. thesis.

Knowles: G. (1978) The nature of phonological variables in Scouse. In P.Trudgill (ed.) Sociolinguistie Patterns in Britisn English. London: EdwardAmold.

Knops, U. (1981) Assirnilatie verschijnselen van Vlamingen in NederIand.Gramma.

Kókeritz, H. (1932) The Phonology of the Suffolk Dialect. Uppsala: UppsalaUniversity.

Kurath, H. (1972) Studies in Area Linguisties. Bloomington: Indiana UniversityPress.

Labov, W. (1963) The social motivation of a sound change. Word 19, 273-309.Labov, W. (1966) The Social Stratifieation of English in New York City, Wash-

ington: CAL.Labov, W. (1972) Soeiolinguistie Patterns. Philadelphia: University of Pennsyl-

vania Press.Labov, W. (1982) Building on empirical foundations. In W. Lehmann and.Y.

Malkiel (eds) Perspectives on Historical Linguistics. Amsterdam: Phila-delphia.

Labov, W. (1983) The three dialects of English. In P. Eckert (ed.) QuantitativeAnalyses of Sound Change in Progress. New York: Acadernic.

Labov, W. (ms.) On the adequacy of naturallanguage.Labov, W., Yaeger, M., and Steiner, R. (1972) A Quantuative Study of Sound

Change in Progress. Philadelphia: U.S. Regional Survey. . .Lanham, L. and Macdonald, C. (1979) The Standard in South African English

and its Social History, Heidelberg: Groos.Larsen, A. (1907) Kristiania Bymél. Kristiania (Oslo).Larsen, A. (1917) Naboopposition - knot. Maal og Minne.Larsen, A. and Stoltz, G. (1911-12) Bergens Bymál. Kristiania (Oslo).Lass, R. (1984) Phonology. London: Cambridge University Press.LePage, R. and Tabouret-Keller, A. (1985) Acts of Identity, London: Cam-

bridge University Press.

166 REFERENCES

Maclaran. R. (1976) The variable !.\): a relic form with social correlates. BelfastWorking Papers in Language and Linguistics 1,45-68. .

Martinet, A. (1955) Economie des Changements Phonétiques . Berne: Francke.Matias, M. (1984) Bilinguismo e Níveis Sociolinguísticos numa Regiáo Luso-

Espanhola. Coimbra: Universiry of Coimbra.Miller, S. (1978) Camp transport before the machine age. The Falklands lsland

Joumal.Milroy, J. (1978) Lexical alternation and diffusion in vernacular speech. Belfast

Working Papers in Language and Linguistics 3, 100-14.Milroy, J. (1981) Regional Accents of English: Belfast. Belfast: Blackstaff.Milroy, J. (1982) Probing under the tip of the iceberg: phonological 'norrnalisa-

tion' and the shape of speech eommunities. In S. Romaine (ed.) Sociolinguis-tic Variation in Speech Communities, London: Edward Arnold.

Milroy, J. (1983) On the sociolinguistic history of lhI-dropping in English. lo M.Davenport et al. (eds) Proceedings of the Second lnternational Conference onHistorical Linguistics, Odense: Odense University.

Milroy, J. and Milroy, L. (1978) Belfast: change and variation in an urbanvernacular. In P. Trudgill Ied.) Sociolinguistic Patterns in British English,Loodon: Edward Amold.

~lroy, J '. and Milroy, L. (1985) Linguistic change, social network and speakermnovanon. Journal of Linguistics 21.

Milroy, L. (1980) Language and Social Networks. Oxford: Basil Blackwell.Milroy, L. (1984) Comprehension and context: successful eommunication and

communication breakdown. In P. Trudgill (ed.) Applied Sociolinguistics.London: Academic.

Miranda·, R. (ms.) On the origins of Fiji Hindi.Moag, R. (1977) Fiji Hindi. Canberra: Australian National University Press.M~ha~: P. (1978) Trinidad Bhojpuri. University of Michigan: Ph.D. thesis.Muhl~ausler, P. (1977) Pidginisation and Simplification of Language. Canberra:

Pacific Linguistics. .Newbrook, M. (1982) Seot or Seouser?: an anomalous informant in outer

Merseyside. English World-Wide 3, 77-86.' .Norde~stam, K. (1979) Svenskan i Norge. Gotheoberg: University Press.0lrnhelm, P. (1983) Sa Sogningen til Fjordingen. Bergen: Sogn Mállag, Firda

MAllag.Omdal, H. (1976) Heyanger har skiftet talemál. Bergens Tidende, 13 April.gmd~,. H. (1977) H0yangermal~t - en ny dial~kt. Spráklig Samling 1.O:wnthe, D. (1977) The English Langua~e 1I1 I~eland. Dublin: Mercier.

on, H. et al. (1962-71) Survey of Englisk Diaieas, introduction and fourvolumes. Leeds: E. J. Amold.

Orton et al. (1978) The Linguistic Atlas of England. London: Croom Helm.Patte~son, M. (1973) Compensation of nonverbal immediary behaviours: a

review, Sociometry 237-52.Patterson, M. (1983) Nonverbal Behaviour. New York: Springer.Pa~e, A. (1976) The Acquisition of the Phonological System of a Second

Dlalect. Unpublished thesis. University of Pennsylvania.

REFERE CES 167

Pavne , A. (1980, F.:-::¡l'~, :,':-.:rollin~ the acquisiuon of thc Pniladelphia dialectby out-oi-stare cnucre-: i:: \ . Lanov (ed.) Locating Languagc in Time andSpace, ew York: A ..:.iermc.

Pringle, I. (1981) The Gaelic substratum in the English of Glengarry County.Canadian Iournal of Linguistics 26.113-17.

Quirk , R., Greenbaurn. S .. Leech, G. and Svartvik , J. (1972) A Grammar ofContemnorarv Enelis+ 1ondon Longrnans

Rekdal, O. (1971) .\JOál.-_\t'r: Dialeki. Unpublished thesis. Oslo University.Robinson, D. (1971) SOII of Bristle. Bristol: Abson.Rogers, I. (1981) The infíuence of Australian English intonation on the speech

of two British children. \\ 'orking Papers of the Speech and Language ResearchCentre, Macquarie Universitv 3, 25-42.

Rona, J. (1963) La frontera lingüistica entre el portugués y el español en el nortedel Uruguay. Veritas 2. 201-20.

f. Rona, J. (1965) El Dialecto 'Fronterizo' de Norte del Uruguay. Montevideo:í Lénaldi.1 ¡r Selinker, L. (1972) Interlanguage. IRAL 10, 209-31.

Shockey , L. (ms.) A pih'lh'.c'gicai change in Eastern Transatlantic English.Shopen, T. (ms.) Research on the variable (ng) in Canberra, Australia.Shukla, S. (1981) Bhojpuri Grammar. Washington: Georgetown University.Spencer, N. (1975) Singular y'all. American Speech.Steinsholt, A. (1962) Málbryting iHedrum. Oslo: Universitetsforlaget.Story, G., Kirwin. W, and Widdowson, J. (1982) Diaionary of Newfoundland

English. Toronto: Amnesty Press.Strang, B. (1970) A Historv of English. London: Methuen.Strange, 1. (1973) Introduction of stock to the Falkland Islands. The Falkland

lslands Iournal.Thelander, M. (1979) Sprákliga Variationsmodeller Tilliimpade pa Nutida Bur-

triisktal. Uppsala: Uppsala University.Timberlake, A. (1977) Reanalysis and actualisation in syntactic change. In C. Li

(ed.) Mechanisms of Syntactic Change. Austin: University of Texas.Trudgill, P. (1974) The Social Differentiation of English in Norwich. London:

Cambridge University Press.Trudgill, P. (1982) Linguistic aceommodation: sociolinguistic observations on a

sociopsychological theory. In T. Fretheim and L. Hellan (eds) Sixtb Scandina-vian Conference of Linguistics, Trondheim: Tapir.

Trudgill, P. (1983) On Dialect. Oxford: Basil Blackwell.Trudgill, P. (forthcoming) Language in Isolation. •Trudgill, P. and Hannah, J. (1982) Intemational English: a Guide to Varieties o/

Standard English. London: Edward Amold.Turner, G. (1966) The English Language in Australia and New Zealand. Lon-

don: Longmans.Venás, K. (1982) Mal 08 Mili«. Oslo: Novus.Viereck, W. (1980) The dialectal structure of British English: Lowman's

evidence. English World·Wide 1, 25-44.Vigeland, B. (1981) Dialekter i Norge. Oslo: Universitetsforlaget.

168 REFERENCES

Wang. W. (1969) ~::::;,e:ing changes as a cause of residue. Language 45.9-25.Weinreich, U. (19:-_'1 Languages in Contact New Yo k' L' " C' 1. r. mgUlstlC Ire e.Wells, J. (1982) Aeee1l!!' 01English three volurnes London: C brid U'it P ". am n ge mver-SIy ress.Welt~n~, B. (ms.) :'\onstandard periphrastie do in the dialeets of south-west

Britain. Index

accommodation, 1-38: hierarehy in,20-1,38; irregularity in, 28-31;limits to, 31-7; long-terrn, 11-21,161:panial. 57-60. 62-5.Q4:proeess, 23-8, 126; quantificationof, 4-10, 37; short-term, 3-4

accommodation theory (Giles), 2-3,39

Adelaide,145Aitken, A. J., 157Aitken's Law, see Scottish Vowel

Length Ruleallophonie variants, 125, 126American Black Vernacular English,

15,145American English, 127:

accommodation of EnglishEnglish speakers to, 12-20;contact with English English,1-2,41; idioms and lexis, 41;rhotic pronunciations, 14, 15-16,20,76-7; simplification in, 147-8;of Whites in South, 15

Andrésen, B. S., 132Appalachian dialects, 80-1, 145approximation process, 60-1, 94areal variants, 121, 126attitudinal factors in accommodation,

39,54-7Australian English: conservative

nature of, 130-1, 133;interdialect in, 142; lrish Englishin, 139-41; levelling in, 143-6;mixing, 129-42; reallocation in,

161; Scottish English in, 141-2;theories about, 130-42; twinsfrom Reading accommodate to,28-31

Bahamas, the, 127, 160 ,Ballymacarett, Belfast, 121-2Belfast English: allophonic variation

in, 159; homonymic cJash in, 17;reallocation in, 119-25

Belize,86Bergen, 24-8, 95Bermuda, 127, 160Bernard, J., 136, 145Bhojpuri, 100: in Fiji, 100-2; in

Mauritius, 108-10Bickerton, D., 87, 90-1bidialectalism, 1bilingualism,lBlair, D., 142, 144Bokmál, 96--8,103-5Botany Bay, 136Bradley, D., 145-6Brazil: Portuguese in, 84-5Bristol Broadsides, 81Bristoll, 78-80, 81Broselow, E., 16Burtrask.Rl-aButters, R., 80-1

California, 21Canadian English, 127: cornpared

with Scottish English, 157-8;ethnic origins cut across social

-

170 INDE:\

class. :':':- ...~\ elling in, ]46; amixed dia.ect , 159; see also ,Canadian Raising

Canadian Marrames dialect, 145, 155Canadian Raisag. 32-3,153--61Cardiff Eng!i,::' rravel agency study.

4-5Caribbean: Eagíish in, 127, 160Catalan dialects. 83-4Central América: English in, 127Chambers. J.: on Canadian Raising,

158; on Ievelíing in CanadianEnglish. 1~: on TorontoEnglish.3:-3

Cheshire, l ..~5children: limns to total

accomz.ccacon in, 34-7, 38;long-term linguisticacco mmodarion, 28-31;occasionaíly bidialectal, 32; usedialect oí peers, not adults, 31

Clonard, Belfas; 121-2Collins, l.: 00 Anstralian English,

130, 13/-S. i44Colonial Engíisa: koinéization in,

127-<il: realJocation in, 152-3,161; sUnpttñcarion in, 147-8,161

'community~m' view, 68comprehe~~.21-3,38Cooper, V.: on St Kitts and Nevis,

87-90Cornwall: Engtisb in, 127; westem,

145Coupland, N.: travel agency study in

Cardiff,4-5Creole: in Belize, 86; Mauritian

French, 106, 108; post-Creolecontimmm, g¡, 90-1; inTrinidad. 106

creolization pidginization, 107, 145

decreolizarion. g¡dialect boundaries: in Belfast Irish,

. 119; 5- and ~vowel systems, 58-9dialect contaa..l. 39-82; or language

. contact, 106, 147; long-term, 161dialect death sirnarions, 68, 76-8

dialect mixture. 83-126; see alsomixíng, dialect

dialects, fudged, 60-2, 94; mixed, seemixed dialects; new, 83-126

dialect-switching, 91-3diffusion: dernographic factors. 39.

126,161; examples of, 42-53;geographical factors, 39, 58-9;habituation process in, 40;outward from large cities, 43;through accommodation, 53-7;varieties, 85-91

divergent dialect communities, 83,91-4

Dixon, J., 142Domingue, N., 108-10

Early.Modern English, 111: effect onIrish English, 150-1

East Africa: English in, 127; Gujeratiin, 100

East Anglian English: in Australia,134, 136; features undergoingdiffusion, 46-57; h-dropping in,44-<i; hyperdialectisms in, 67-8;realization of intervocalic /tI, 19 .

Elizaincín, A., 85English English: accommodation by

Americans living in England to,21-3; in USA; 15-16

face-to-face interaction, vii, 39-40, 82,126, 161

Falkland Islands: Canadian Raisingin, 160; English in, 127, 128, 144;loan words from Spanish in, 128

Ferguson, c., 103Fiji: Hindi in, 100-2, 106; levelling,

99-102Fjordane dialect, 95-9focused varieties, 8s-{)focusing, 96-7,101,107,126; along

social dialect continua, 91-4Francis, W. Nelson, 67-8, 112Fronteirico, 83-5Fudge, Erik, 79

tudged dialects. 6(~2. 9~

GVS, see Great Vowel ShiftGalician,84Gerstaecker, F .. 142Giles. Howard: accent convergence

and divergence, 2, 78;accommodation theory, 2, 39; ondangers of circularity in research,5; on human beings, viii

Gordon, E., 138-9Great Vowel Shift (GVS), 134, 15s-{)Greece, 32Gregg, R., 157Gujerari: in East Africa, 100Guyana: Hindi in, 100; post-Creole

continuum,87

Hallingdal dialect, 70Hammarstrórn, c.: Australian English

as unmixed nineteenth-centuryLondon Englisti, 130-7, 144; onsociolects in Australian English,153

Hammer, the, Belfast, 121-2Hardwick, M., 146Harris, J., 149Hedrum, Norway, 56Hensey, F., 85Hindi: in Fiji, 101; new dialects

resulting from transplantation,106; outside India, 100; standardin Fiji, 100-2; in Trinidad, 106-7;see also Bhojpuri

Hindustani, 100Home Counties English, 46-7;

compared with AustralianEnglish, 131, 134-5

homonymic clash, 17, 21, 38, 57Hordaland,95Heyanger, 95-9: dialect

transplantation in, 95-7; levellingin, 98; simplification in, 102-<i

hyperadaptation, 66, 75, 123: in IrishEnglish, 151; long-term, 78-81

hypercorrections, 66, 82: misanalysed,66,67; performance errors, 66

INDEX 171

hyperóiaiectisms. 66-78. 82

lcelandic, 130Ihalainen, O., 70imitation, not accommodation, 12-14,

37.40-1in-rnigration: from London to

Norwich, 55, 57; from rural areasto Norwich, 110-19; see alsosprákmisjontzr

indicators, 10-11: dialect, 93;standard, 93

interdialect, 62-5, 68-9, 78, 82-3: inColonial English, 142, 151, 161;as a result of borrowing, 101; as aresult ofhyperadaptation, 123; asa result of systemic pressures, 101

interlanguage (Se1inker), 62intermediate dialect variety, 93intermediate forms, 60-2, 82: in

Australian English, 142Irish: role in formation of Irish

English, 149-52lrish English, 81,127: in Australia,

139-41; in Newfoundland, 129;role of language contact in,149-52; second person pronounssingular/plural distinction, 71,140; see also Belfast English

Isle of Man, 127Italy, 41

Jahr, E. H., 7, 8

Kazazis, K., 32Knowles, G., 66-7koinéization, 106, 107-10, 126: in

Colonial English, U7-<i1; seealso levelling; simplification

Kokeritz, H., 67,112

LAE, see Linguistic Atlas 01EnglandLabov, W.: on accommodation in

children, 32, 34; on markers andindicators, 10; on Martha'sVineyard, 75; on mergers, 105,119; New York City work, 5, 91;

1 2 INDEX

on 'observer 's paradox', 7, 17: onPhiladelphia, 158;on tensevowels rise, 133

language contact: or dialect contact,106,147; 'interference' in, 1;problem of, 148-52

ianguage death, 106-7Languages in Contact (Weinreich), 1Lanham, L. W., 139Larsen, Amund B., 62, 69-70Larvik, Norway, 56Le Page, R.: focused and diffuse

varieties, 85-6levelling, 98-102, 107,126: of

regionally marked forms, 98,101; of socially marked forms,101: in southern hemisphereEnglish, 143--6,161

lexical differences, 24-5lexical diffusion: phonetic motivation

in,58-Q1liberia, 127Linguistic Atlas of England (LAE) , 75linguistic change, 11London English: comparison with

Australian English, 130--7;infiuence on East Anglia, 46-57;intervocalic ItI, 19

Lowman, Guy, 67,112Ludham,67

Mac\aran, R., 121, 122markers,lO--11: salience of, 11,45Martinet, A., 51Martlesham, 67Mauritius: Bhojpuri in, 108-10;

French Creole, 106,108; Hindiin, 100, 106; stylistic variation,119; variability in mixed dialects,108-10

media: role in diffusion, 40--1Melboume, 145-6Melchionne, Tom, 128mergers: and American English, 148;

in East Anglia of 161and 1ft, 53--7;in contact situation, 105, 119

Middle English, 34-5, 67, 112, 155-6

rnid-Ulster English. llY. 121-3Milroy, J., 17.119. 120. 123--5Milroy, Lesley, 71,120,121-2Miranda, R., 101mixed dialects, 59-60: growth oí, 94;

variability in, 108-10mixing. dialect: in Australian Engiish,

129-42;process,127-8Moag, R., 101, 106modification, 10--11,37morphological differences, 24-5Mühlhausler, P., 103

nabo-opposisjon ('neighbouropposition'), 69

naturalness, phonological, 22, 38, 46,51. 161

Nevis, 87-90New York City, 5, 91: children from in

Philadelphia, 34, 36-7New Zealand: compared with

Australian English, 143-4;English in, 127, 161: h-Iessness,138-9; mixing in, 129, 131;Scottish English in, 141-2

Newbrook, M., 32, 71Newfoundland: English in, 128-9,

155; lrish English in, 129; south-. westem English in, 129

non-rhotic varieties of English, 72~Nordenstam, K., 24-8Norwegian: accommodation of

Swedish in Bergen to, 24-8;dialects, 69-70, 95-9; interdialectin, 62-3; Oslo Norwegian, 62-3,95

Norwich English: case of 101(1983),42~; h-dropping in, 110--11;reallocation in, 110--19, 125-6;studies (1974, etc), 6-11, 34-7,110--12

Nynorsk, 96-8, 103-5

'observer's paradox' , 5, 7, 17Orndal, H., 95Ontario, 146orthography, 37, 45

Pan iabi dialec in Britam. JOOPayne , A.: on ·ev. York children in

Philadelphia. 34, 36-7Philadelphia, 21: New York children

in. 34. 36-7phonemic contrast , 14,20--1,37,38pnonetic distan te. L. 1~. 37phonological accornrnodation, 24-5phonological contrast, 11, 14,37,38phonotactic constraints, 16,21,38,53,

80--1pidginizationlcreolization, 107,145Pitts, Walter, 79pop singers, British imitate American,

12-14Port Stanley, 128, 144Portuguese: Brazilian. 84-5: dialects,

83-5Pulham, 112

Irl non-prevocalic in English English,71-8: analogical, 74;hyperadaprive, 74;hyperdialectical, 75; intrusive,72-3; linking, 72; phonotactic,74,80

RP (received pronunciation), 21, 35,42, 6G-2, 111

Reading: English use of 'do' in presenttense forms, 63-5; twins fromaccommodate to AustralianEnglish,28-31

reallocation, 110--26:in ColonialEnglish, 152-3, 161

received pronunciation, see RPregional standard (Thelander), 93-4regional variants, 110Rekdal, 0.,63rhotic varieties: of American, 76-7;

different methods of adaptingnew words to phonotacticstructure, 80--1;of English, 74-6

Rogers, Inge: study of Reading twins'accommodation to AustralianEnglish,28-31

SED, see Survey oí English Dialects

INDEX 173

.Saba: Canadian Raising in. 160SI Helena. 127. 160St John's, ewfoundland.129St Kitts. 87-90salience. linguistic: attaches to

markers. 11. 45: extra-strong,1&-19.2 . 3b: iactors in. 11. 37.43; inhibits accommodation, 125,161

Scots: compared with CanadianEnglish, 157-8

Scottish English, 81: Aitken's Law,157,158; in Australia, 141-2; ofHighlands, 127, 145

Scottish Vowel Length Rule, 157, 158self-analysis, linguistic, 7, 15Selinker , L. 62sex: infiuence on Iinguistic variables,

7,8,29Shockey, L, 21-3, 58Shopen, T., 8simplification, 102-7, 119, 126: in

Colonial English 147-8, 161;increase in morphological andlexical transparency, 103;increase in morphophonemicregularity, 103

smoothing, East Anglian, 47-50social c\ass variants,l18-19, 121, 126social dialect: diffuse continuum,

86-91; focusing along continua,91~

social psychologists, 2~sociolinguists: accommodation to

interviewees' speech, 5;quantification analysis ofaccommodation, 4

Sogn dialect, 69-70, 95-9South Africa: English in, 127,129,

135,139; infiuence of Afrikaans,144; reallocation in English of,161

South America: English in, 127;language transplantation in, 83-5

South Carolina, 154south-eastem English, 73south-western English: adds

174 INDEX

phonotactic Irl, 80; inNewfoundland, 129: non-standard past tense forms, 7Cf-1

Spanish: in Belize, 86; dialects, 83-5;loan words in Falklands, 128;Uruguayan,84-5

sprákmisionar (language missionarv),56-7

Steinsholt, Anders, 56-7stereotypes, 10, 18stigmatization, overt, 11Story, G., 129Strang,B.,41stylistic variants, 109-10, 121, 125, 126Suffolk,67Sunndal,63Surinam: Hindi in, 100; Sranan, 106Survey of English Dialects (SED):

Bristoll, 78-9; on East AnglianEnglish, 132, 134; hyperdialectalIr/, 75; Norfolk records (1950s),67-8; Pulham (1950s), 112; ontransition zone between 5- and6-vowel system areas, 59

Swedish: in Burtrask, 91-4; in Bergen,24-8

Sydney,145syllable structure transfer hypothesis

(Broseiow),16syntactic differences, 24

Tabouret-Keller, A., 85Talmálsundersekelsen i OsIo (TA US),

7Tamil,l00Thelander, Mats, 91-4Toronto, 32-3, 146towns, new, 95-7transfer, word: process, 59-{j(),94transition: by approximation, 60-1,

94; in fudged dialects, 6(}-1;byword transfer, 60; zone whereintermediate varieties occur, 59

transplantation: dialect. 95-7.100:languagc.Bó-S

Trinidad: English Creole in, 106;English in, 106: Hindi in, 100,106-7; simplification in, 106-7

Tristan da Cunha.127, 160

Ulster Scots, 119, 121-3Vnderwood, Gary, 56United States of America, see

American Englishurbanization, linguistic, 110; see also

towns, newUrdu, 100Uruguay: Spanish in, 84-5

Vancouver, 146variability in mixed dialects. 108-10.

121, 123-{í;see also reallocationvariant-switching (variantviixling)

(Thelander), 91-2variety-switching (Thelander), 91-2Virginia, 15, 154

Wales: English in, 19, 127 - North,145, South, 7(}-1

Walters, Keith, 71, 76Weinreich, Vriel, 1Wells, J.: Australian and London

diphthong shift, 134; on ColonialEnglish, 144; on glottal stop, 132;on hypercorrection with intrusive/1/,79; on Irish compared withAustralian English, 141; nodivergent traditional dialects,145; on phonotactic Ir/, 72, 80; onregional variants in Britain, 152;on smoothing East Anglian laul,47-50; on social class variants inAustralian English, 153

Weltens, Bert, 7(}-1

Zimbabwe: English in, 127