Codeswitching (Multilingual Matters)

cover

title: author: publisher: isbn10 | asin: print isbn13: ebook isbn13: language: subject publication date: lcc: ddc: subject:

next page >

Codeswitching Multilingual Matters (Series) ; 89 Eastman, Carol M. Multilingual Matters 185359167X 9781853591679 9780585252001 English Code-switching (Linguistics) , Urban dialects, Languages in contact. 1992 P115.3.C59 1992eb 306.4/4 Code-switching (Linguistics) , Urban dialects, Languages in contact.

cover

next page >

< previous page

page_i

next page > Page i

MULTILINGUAL MATTERS 89 Series Editor: Derrick Sharp

Codeswitching Edited by Carol M. Eastman MULTILINGUAL MATTERS LTD Clevedon Philadelphia Adelaide

< previous page

page_i

next page >

< previous page

page_ii

next page > Page ii

Library of Congress Cataloging in Publication Data Codeswitching/Edited by Carol Eastman p. cm (Multilingual Matters: 89) Also available as a special issue of the Journal of Multilingual and Multicultural Development, Vol. 13, Nos 1 & 2, 1992 (T.p. verso) Includes bibliographical references 1. Codeswitching (Linguistics). 2. Urban dialects. 3. Languages in contact. I. Eastman, Carol M., 1941- II. Series: Multilingual Matters (Series): 89. P113.3.C59 1992 306.4'4-dc20 British Library Cataloguing in Publication Data A CIP catalogue record for this book is available from the British Library. ISBN 1-85359-167-X (hbk) Multilingual Matters Ltd UK: Frankfurt Lodge, Clevedon Hall, Victoria Road, Clevedon, Avon BS21 7SJ. USA: 1900 Frost Road, Suite 101, Bristol PA 19007, USA. Australia: P.O. Box 6025, 83 Gilles Street, Adelaide, SA 5000, Australia. Copyright © 1992 Carol M. Eastman and the authors of individual chapters. This book is also available as a special issue of the Journal of Multilingual and Multicultural Development Vol. 13, Nos 1 & 2, 1992. All rights reserved. No part of this work may be reproduced in any form or by any means without permission in writing from the Publisher. Printed and bound in Great Britain by Short Run Press, Exeter.

< previous page

page_ii

next page >

< previous page

next page >

page_iii

Page iii

Contents Codeswitching as an Urban Language-Contact Phenomenon Carol M. Eastman

1

Codeswitching Theory and Examples: Africa Comparing Codeswitching and Borrowing Carol Myers-Scotton

19

French in Urban Lubumbashi Swahili: Codeswitching, Borrowing, or Both Marjolein Gysels

41

Codeswitching and the Exclusivity of Social Identities: Some Data from Campus Kiswahili Jan Blommaert

57

Codeswitching in Bukavu Didier L. Goyvaerts & Tembue Zembele

71

Two Codes or One? The Insiders' View and the Description of Codeswitching in Dakar Leigh Swigart

83

Socialising Multilingualism: Determinants of Codeswitching in Kenyan Primary Classrooms Marilyn Merritt, Ailie Cleghorn, Jared O. Abagi & Grace Bunyi

103

Codeswitching Theory and Examples: Canada and Belgium The Politics of Codeswitching and Language Choice Monica Heller

123

French-Dutch Codeswitching in Brussels: Social Factors Explaining Its Disappearance Jeanine Treffers-Daller

143

Codeswitching in Contexts of Language Learning, Dialect Shift and Style Shift Codeswitching: Black English and Standard English in the AfricanAmerican Linguistic Repertoire Charles E. DeBose

157

Codeswitching, Convergence and Compliance: The Development of Micro-Community Speech Norms Susan Meredith Burt

169

Chinese Preschool Codeswitching: Mandarin Babytalk and the Voice of Authority 187 Catherine S. Farris Codeswitching for Humour and Ethnic Identity: Written DanishAmerican Occasional Songs Marianne Stolen

< previous page

page_iii

215

next page >

< previous page

page_1

next page > Page 1

Codeswitching as an Urban Language-Contact Phenomenon Carol M. Eastman Department of Anthropology, University of Washington, Seattle, WA 98195, USA Editor's Introduction The twelve papers in this volume share certain common themes. One is that efforts to distinguish codeswitching, codemixing and borrowing are doomed. In addition the study of loanwords per se out of context is a relic of the past. Attention primarily to urban settings where people from diverse linguistic and cultural backgrounds regularly interact makes it abundantly clear that in normal everyday conversation, material from many languages may be embedded in a matrix language regularly and unremarkably. At the same time, forms may be embedded noticeably. The marked and unmarked status of non-native material in the speech of urban multilinguals represents the central theme here. Where people use a mixed language regularly, codeswitching represents the norm (is an unmarked choice after Myers-Scotton, 1983). Where people invoke another language in an obvious way, position of relative social, political or economic strength is often being negotiated and codeswitching represents a marked choice. In addition, it will be seen that the absence of codeswitching in multilingual urban contexts may be indicative of the tacit acceptance of political, social, and/or economic power differenceswhere negotiated change has become impossible or is not yet possible. It is clear that once we can free ourselves of the need to categorise any instance of seemingly non-native material in language as a borrowing or a switch, we will be much further along in our effort to understand the way cognitive, social, and cultural processes work together in urban linguistic contexts. The 'Call for papers' for this volume met with an overwhelming response for which I am most grateful. Selected contributions may be seen to fall into three rather broad categories. A number of contributors examine codeswitching in Africa where unrelated languages come in contact. The result poses a challenge to certain generalisations about the nature of codeswitching when Indo-european or any other related languages are in contact with each

< previous page

page_1

next page >

< previous page

page_2

next page > Page 2

other. Two papers examine the social and political context of codeswitching in Europe (French and Dutch in Belgium) and North America (French and English in Canada) while a third set examines broader aspects of codeswitching. In Part Three, we see codeswitching as a conversational strategy by foreign language learners, as it involves different registers (here, baby talk and adult speech), at work in a non-conversational genre (as it occurs and has meaning in bilingual songs), and in use by bidialectal speakers of Black and Standard English in the US. Borrowing Versus Codeswitching Carol Myers-Scotton's paper serves as our lead-n article since it has as its basis a proposed model to explain codeswitching in general. In 'Comparing Codeswitching and Borrowing,' Myers-Scotton argues that borrowing and codeswitching need to be seen universally as related processes. According to her Matrix Language Frame (MLF) Model (Myers-Scotton, forthcoming), codeswitching occurs everywhere within a frame which is set by the matrix language. The term matrix refers to the language in which the majority of morphemes in a given conversation occur. The languages from which material enters a matrix language are referred to as embedded. Central to MLF theory is the idea that content (often nouns, verbs, etc.) and system (articles, inflections etc.) morphemes in the embedded language are accessed differently by the matrix language. The concern in this paper is with the basis for distinguishing which singly occurring embedded language lexemes are borrowings from those that are codeswitches, in light of the fact that both borrowed and codeswitched forms behave the same way morphosyntactically in the matrix language. Myers-Scotton's MLF model posits structural constraints which account for both borrowings and codeswitches. Particularly important here is the point that (contra Poplack, 1980, 1987 and others) codeswitching and borrowing should not be seen as distinct processes. Using data from Kenya (with Swahili as the matrix language [ML]) and Zimbabwe (with Shona as the matrix language [ML]), Myers-Scotton explores four hypotheses. Among them is the hypothesis that matrix language + embedded language [EL] constituents conform to the morpheme order of the matrix language and, in ML + EL constituents, articles and inflections (system morphemes with non-lexical information) come from the matrix language. Another hypothesis proposed, the Blocking Hypothesis, holds that EL content morphemes that are not congruent with stored lemmas (lexical information in a mental lexicon [after Levelt, 1989]) are blocked from appearing in ML + EL constituents. Embedded language forms which cannot be accounted for by the blocking hypothesis are subject to an embedded language trigger hypothesis. And, finally, an E1 Hierarchy Hypothesis predicts that peripheral and formulaic embedded language constituents may occur with relative freedom in a matrix language.

< previous page

page_2

next page >

< previous page

page_3

next page > Page 3

Myers-Scotton's model based on these hypotheses provides a principled basis for considering both borrowing and codeswitching processes to be part of a single continuum. On the continuum, forms range from abrupt loans (e.g. the term school fees as used in Shona), to core borrowings (e.g. the term weekend in Swahili), to actual instances of codeswitching. According to MyersScotton, frequency is the best criterion to use to link borrowings to the mental lexicon in contrast to single-form codeswitches. Yet, there is very little reason to make a distinction between the processes. Neither morphosyntactic nor phonological integration criteria remain viable ways to decide whether embedded language material is the result of borrowing or codeswitching. The MLF model of codeswitching which can account for single form codeswitches also accounts for borrowed lexical items. Single or multi-word codeswitches are retained in the mental lexicon of the embedded language (are lemma entries there) while borrowings continue to remain matrix language lexical items. All codeswitches structurally represent material embedded into a matrix language while all borrowings are matrix language material par excellence. In accord with this model, the English preposition at, for example, is blocked from appearing in the Swahili sentence nilikuwa nataka kumpata stadium which translates as 'I wanted to find him stadium' rather than 'I wanted to find him at the stadium. At (or for that matter the as well) has no counterpart in Swahili, so stadium 'occurs as a bare form' (p. 26) making it look as much like a borrowing as a codeswitch. In contrast, other English prepositions do have congruent forms in Swahili and may occur themselves as single codeswitched forms (kati ya and between so match). Marjolein Gysels' paper 'French in Urban Lubumbashi Swahili: Code-switching, Borrowing, or Both?' serves as a complement to Myers-Scotton's paper in that her codeswitching model allows for a codeswitch-borrowing distinction but does not see such a distinction to be critical. Gysels takes neither a structural integration-based approach to distinguish borrowing from codeswitching (as Poplack did) nor a community and social function frequency-based approach (as Myers-Scotton does). Going beyond Myers-Scotton's stance that a distinction need not be made, Gysels argues that a categorical distinction CANNOT be made. Her argument is that neither structural features nor quantitative arguments can account for the fact that elements from an embedded language can at once be established borrowings (loans) and used as codeswitches. This is seen here as a matter of discourse organisation. The same form may be interpreted as either a borrowed item or a codeswitch. For example, Gysels discusses the Lubumbashi word authenticite which is from French. But, when a Swahili speaker uses it, it is motivated as a codeswitch, indicating, emphasising and highlighting the topic of conversation. Authenticite in Lubumbashi Swahili comes with ascribed emotional baggage associated with a set of beliefs sentimentally held as well as a set of fondly remembered customs (coutime) from the French colonial past.

< previous page

page_3

next page >

< previous page

page_4

next page > Page 4

This paper builds on the work of Fabian (1982) de-emphasising the importance of the instrumental hypothesis long used to explain borrowing as an effort to fill lexical gaps. Instead, borrowings are seen as creative ways to fulfil 'connotative and indexical functions.' (p. 42) The form of Swahili in Lubumbashi has a number of strands within it ranging from Shaba Swahili which arose at the time workers were brought together to work in the mines, through a form introduced in Katanga for political reasons, to the missionary grammaticalised version used to teach literacy and regarded as 'Swahili Bora' or 'superior' Swahili. Gysels' analysis of French in Lubumbashi Swahili is based on the idea that there are 'two mixed codes' (p. 48), 1) a colloquial variantwith French elements used sporadically and unconsciouslyi.e. as stable loans; 2) a register variantwith French loans and codeswitches used not as vocabulary but with grammatical (pragmatic) function as we see in the authenticite example. A person who speaks in the colloquial mixed code generally has no awareness of French language and culture per se. Such a person is reminiscent of the American who begins to study German and is surprised to find out she already knows some vocabulary (e.g. wienerschnitzel, dachsund, etc.). The French used in the colloquial variant indicates 'that French pervades daily life in Lubumbashi' much as German does in the UK with Oktoberfest, braunschweiger, bratwurst, etc. In the colloquial variant, there are statistically fewer French words. However, where French elements denote register, they may be used at anytime for discourse pragmatic effect. Once a speaker invokes authenticite a whole set of associated beliefs are made available in the conversation. Nonetheless, in both colloquial and register forms of French-influenced Lubumbashi Swahili, the French conforms to Swahili agreement patterns. To distinguish colloquial and register variants of French in Lubumbashi Swahili, Gysels (p. 50) provides this example: (20) kwa kugarantir eh social, kupata maisha mema (to guarantee eh social to get a good life.) Here the French phrase can only have meaning from the phrase that follows it. The French represents an 'intra-discourse convention' (p. 50) calling attention to the following phrase and imbuing it with a 'touch of class.' French elements in Lubumbashi Swahili have a pragmatic purpose, that isthey serve a topic marking, emphasis, and foregrounding function. French particles, too, may function pragmatically in Lubumbashi Swahili. For example, bon is always preceded by a pause and either marks a conversational turn or a change in subject. Jan Blommaert's paper 'Codeswitching and the exclusivity of social identities: Some data from Campus Kiswahili' examines language use on the part of the academic staff at the University of Dar es Salaam. The English in their Swahili is perceived to be 'grammatically and semantically correct.'

< previous page

page_4

next page >

< previous page

page_5

next page > Page 5

This functions in such a way that the staff is seen to have an exclusive sociolect which arose out of Tanzania's sociohistorical context. Blommaert's question has to do with the social meaning of this mixed code at the group level. As such this paper departs from many codeswitching studies which have focused more on the interpersonal level (including the work here of Myers-Scotton and Gysels just discussed). Blommaert argues that Swahili is now a first-language for more and more young people in Tanzania. From independence (1964) through the 1970s English was somewhat stigmatised (though professionals and academics necessarily used it) as not friendly to African socialism and the idea of nation buildingi.e. it was un-Ujamaa. When Ujamaa failed, English began to gain renewed acceptance in public life (evidenced by the 1975 refusal to implement Kiswahili as the language of higher education). At the same time the masses have been cut off from access to English in school. What English they come into contact with since independence is in, for example, pop-music, movies, and the like. As a result, the nature of English codeswitching in lower class speech is different from that among the elite academicians. The relative grammatical and lexical complexity in the English embedded in academicians' Swahili serves to distinguish their sociolect from that of the masses. This gives rise to a sociolinguistic situation in which the elderly associate English with colonialism and, in response, favour Swahili. The young associate English with modernity and favour it over Swahili. The elderly know Swahili-as-a-second language while the young have Swahili as a first-language. In this milieu, academic staff know both grammatical English and Swahili. Further, elite professionals have more exposure to English while factory workers are more familiar with Swahili. Blommaert argues that Campus Kiswahili has a discourse organisational function. The English marks what is important, relevant, or new information in an utterance, e.g. ni hali ambayo kwa kweli #it is still deteriorating#//lakini kwenye #situation# kama hiyo hata kama umepata nafasi ya kusoma nafikiri #it's just low# Blommaert agrees with the point Gysels makes that codeswitching and borrowing are not what is primarily at issue in the study of language contact phenomena. To understand language contact and language change attention must be paid to overall discourse structure. Blommaert views codeswitching as a type of social historiography. In contrast to Myers-Scotton, it is his view that codeswitching cannot start from assuming an egalitarian communicative competence. 'Differences in the repertoires of available choices may be precisely one of the main features of intercultural and inter-societal differentiation' (p. 63). Communicative competence is regarded here as a 'trivialised macro-concept. In many analyses, it stands variably for cognition, social norms and conventions, ethics, knowledge of politeness rules and so on.' (p. 64) His view, contra Myers-Scotton's idea that separate mental lexicons are maintained in matrix and embedded languages, is that people cannot have mental represen-

< previous page

page_5

next page >

< previous page

page_6

next page > Page 6

tations for what he sees to be unstable social phenomena. Instead, attention needs to be paid to the socioeconomic and political origin and development of speakers' potential for code choice (here Campus Kiswahili). Blommaert agrees with Myers-Scotton (1989) that codeswitching can construct identities by exclusion. In fact, the use of Campus Kiswahili by academic staff at Dar is a good example of her notion of elite closure. To Blommaert, '. . . it is the quality of the mixed languages that accounts for the sociolinguistic difference between Campus Kiswahili and other English-interfered variants of Dar es Salaam Swahili.' (p. 67) From this discussion, two phases of identity construction emerge: (1) boundary creationan asocial and unco-operative phase (2) the establishment of mutual role relationshipsa co-operative and social endeavour. Blommaert feels that phase 2, the level of group relations, has been neglected. As a result the social history which gave rise to variable access to linguistic resources is masked. Social identity needs to be seen as both situational and bounded by structures reflecting the history of macro-level social differentiation. Didier Goyvaerts' 'Codeswitching in Bukavu' looks at codeswitching as part of a comprehensive study of the overall linguistic and cultural situation of Bukavu town. This study will amount to a kind of sociolinguistic grammar of a mixed language in which codeswitching is the norm. Echoing the preceding papers in this volume, Goyvaerts sees codeswitching data as providing social as well as linguistic information. His goal is to use code-switching data to reveal social parameters. In Bukavu people mix French, Swahili, Shi and Lingala inter- and intra-sententially. Re-iterating Myers-Scotton's earlier markedness model of codeswitching (Myers-Scotton, 1983), Goyvaerts describes how codeswitching is used in Bukavu to encode power and solidarity. Myers-Scotton's model places power in the speaker who, in speaking, reveals multiple identities and may purposely engender an ambiguous response or interpretation on the part of the hearer/addressee. Conversations between a father and son (mixing French and Swahili), between a student and a headmaster (Fr., Sw.), two classmates (Fr., Sw.), two friends sharing a confidence (Fr., Sw.), among young people at home watching television (using Fr., Sw., and Lingala) and another in which a landlord speaks with Shi workers (in Fr., Sw., and Shi) are examined. Codeswitching as representing marked (unusual) or unmarked (usual) choices is tabulated. Switches are classified further as occurring inter- or intra-sententially, as intra-word switches, or as a matter of the relaxation of structural constraints across languages. Goyvaerts found that in the Bukavu data, marked codeswitching may only be inter-sentential. People do not switch languages midstream as a rule. A notable exception occurred in the conversations between the father and son, and between the student and headmaster. These conversations were

< previous page

page_6

next page >

< previous page

page_7

next page > Page 7

anomalous for a number of reasons. Goyvaerts found that young men in particular use intra-sentential switching as a speech pattern when talking specifically to their fathers. The seemingly marked codeswitches young men make are seen as attempts on their part to renegotiate the balance of power that exists. In terms of Myers-Scotton's markedness model, they are seeking to replace one set of unmarked roles and obligations with another. Boys in Bukavu are permitted to get away with this (hence the unmarked interpretation) while girls are not. In the conversation between the headmaster and student, the switch receives a marked rather than an unmarked interpretation due to circumstances of the discourse context. The student is attempting to establish solidarity with the headmaster and to curry favour. Goyvaerts' data show that conversational codeswitching as a marked choice is the speaker's way of negotiating power in terms of addressee's roles and obligations to speaker. The son appears to create social distance from the father. That this is expected allows for the codeswitching as an unmarked choice interpretation. On the other hand, the student switches to Swahili to try to lessen social distance from the headmaster. Where the landlord talks to Shi workers, the switch to Swahili is a marked choice to create more social distanceto be able to stay in charge. Codeswitching is seen here as a way to enhance speaker power in a discourse context. Language use in Bukavu is often a matter of codeswitching as an unmarked choice as is often the case in multilingual urban settings. This will be seen to be the case as well in Dakar (Swigart paper, this volume). People for whom codeswitching is the norm reveal complex identities not unrelated to the fact that the majority of people living in Bukavu are young as well as urban and seeking to maintain the town's ethnic balance. In such a community a set of unmarked choices are generated each consisting of a 'mapping of contexts and codes in terms of variables such as age group, ethnic affiliation, socio-economic class, sex, status, aspirations, family make up [etc.]' (p. 80). For a person to fit into a multilingual community, that person needs to have learned what is appropriate where. Such a person draws on these mapped contexts and codes by appealing to what Goyvaerts refers to as a markedness dictionary which is constantly undergoing change and revision as the social milieu itself changes. In 'Two codes or one? The insiders' view and the description of code-switching in Dakar,' Leigh Swigart questions the assumption found in many codeswitching studies that the use of two languages within a single speech interaction violates a norm and must thus be somehow explained. She describes the situation found in Dakar, the capital of Senegal, where the fluid and unmarked switching of Wolof and French has become the most common way of speaking among urbanised bilinguals. She suggests that this mode of speech is perhaps better described as a third and separate code than as the alternate use of two languages. Swigart argues that models found in the codeswitching literature have been developed to analyse 'purposive' or goaloriented codeswitching, and while many researchers have noted the

< previous page

page_7

next page >

< previous page

page_8

next page > Page 8

existence of 'urban codes,' their frameworks have not paid much attention to the more sociolinguistic aspects of this phenomenon. Although purposive codeswitching does exist in Dakar (as it does in Bukavu), it is limited to those domains of activity where the use of a specific language is highly sanctioned. In such cases, the symbolic association of one or the other language may be evoked, for example, to create a certain impression or negotiate a situation. But, for the most part, codeswitching between Wolof and French is not used to negotiate social roles or obligations (as accounted for by Myers-Scotton), nor is it a function of maintaining social or political boundaries (as Heller has examined). Urban Wolof is instead the accepted and, for some groups, the expected code to use. It is seen here as the documented but little described phenomenon referred to by Poplack as an overall discourse mode and by Myers-Scotton as 'overall switching as the unmarked choice.' As Blommaert has described Campus Kiswahili (this volume), Urban Wolof is an educated as well as urban phenomenon, and also like Campus Kiswahili, it is available only to those who have a good command of both languages involved. Unlike Campus Kiswahili, however, Urban Wolof is not thought of by its speakers as an elite way of speaking. It is usually regarded with a large measure of distaste and is often used with little conscious awareness. Swigart suggests that investigating the often contradictory attitudes surrounding urban codesthey are widely used but disparaged, covertly accorded a separate status but overtly called 'bad Wolof'is important in understanding the position of such codes in the communities in which they are playing an increasingly central role. The insiders' view in these matters is of vital importance since it provides a rich picture of the problematic status of urban codes and also indicates what their future role might be in the linguistic repertoire of urban speakers. From each of these papers, it is clear that codeswitching is as much a social and discourse phenomenon as a structural or purely linguistic one. Driving this point home even further, the paper by Merritt et al. 'Socialising Multilingualism: Determinants of Codeswitching in Kenyan Primary Classrooms' describes the very situation in which a markedness dictionary may be formed. In many ways the outcome of primary education in Kenya is a population for whom codeswitching represents an unmarked choice and for whom marked code choices represent shifts or attempted shifts in power. It is Kenyan educational policy that school certificate examinations are in English except for a compulsory language exam in Swahili. The language of instruction from standard 4 on is English. Standards 1-3 are taught in variously English, Swahili, or a regional first language. What really happens is that teachers teach to the KCPE certification tests. Authors looked at an English-medium school; a Swahili then English-medium school; and a Luo-medium school with English and Swahili as subjects in standards 1-3 with English-medium from standard 4 on plus Swahili as a subject. In all three

< previous page

page_8

next page >

< previous page

page_9

next page > Page 9

schools there was a preference to rely on English as much as possible. Merritt et al. are particularly concerned with teacher language choice which may '. . . be seen to be accountable to national and school policy, as mitigated by the immediate cognitive concerns of the classroom lessons' (p. 107). Examples are presented in which e.g. a teacher says the same thing three times in three different languages ('close your book' and 'you're not all participating'). Another pattern has to do with codeswitching as well as content shifting ('Do you hear that?' in Swahili, with 'What's causing that?' in English). It appears that Swahili is used to get student attention with English to carry the rest of the content. Where the same thing is repeated in the different languages, English is used first. A third pattern involves translation or word substitution especially where there is no equivalent word in the matrix language (e.g. the Swahili word matunga is used, barring any exact English equivalent term for this food made of intestines). Another pattern of language use observed in Kenyan schools is the use of markers from the different languages as a way to elicit behaviour such as 'try' or 'next person' or 'say loudly' in Swahili where the expected behaviour would be on the part of the student in English. Such markers have discourse pragmatic function as well when they accompany topic shifts as we saw to be the case with Lubumbashi speakers who use the French bon in otherwise Swahili contexts (Gysels, this volume). Intersentential shifts are often summaries of instructional materials (to be sure everyone gets at least the directions straight!). Teacher codeswitching occurs in order to focus or regain student's attention or to clarify material. It was also found that teacher linguistic insecurity (or failure to accommodate to a prescriptive ideal) arises when grammatical agreement is at issue in the embedded language (English) or when teachers lack control of the subject matter (again in English) due either to insufficient preparation or training. In Kenyan schools, then, teachers are teaching prevailing patterns of multilanguage usestudents learn when to use English, Swahili, or the vernacular. Teachers are teaching codeswitching as an unmarked choice and, as well, passing on to students a knowledge of the way to manipulate language by making marked codeswitches in the interest of successfully negotiating their way through society. An attitude seems to be being promulgated, perhaps unconsciously through the teachers' own command of and relatively limited access to English, that English is relatively unaccessible (socio-economically as well as cognitively). Codeswitching in a Political Discourse Context Monica Heller's 'The Politics of Codeswitching and Language Choice,' in keeping with the papers preceding it here, recognises that codeswitching may be conventional (representing a normal, routine way to use language) or anti-conventional (violating expectations). That is, codeswitching is a

< previous page

page_9

next page >

< previous page

page_10

next page > Page 10

way people handle resources. When power is stable, conventions prevail. Codeswitching, Heller argues, needs to be seen as part of a range of linguistic practices people use to establish social goals. She uses French-English codeswitching in Quebec and Ontario to suggest a framework to examine universal characteristics of development and change in codeswitching practices. Bourdieu's (1977) notions of symbolic capital and marketplace and Gumperz' (1982) speech economics and verbal repertoires are called into play. People need to be able to invoke appropriate codes in situations where access to resources is at issue. To the extent they can, they are able to achieve dominance and sustain power. The question is, however, how does change occur and where is symbolic capital obtained? In this paper, Heller uses codeswitching as an entry point to explore the way dominant groups use language choice to maintain power, while subordinate groups variously acquiesce or resist, redefine the meaning of choice, or alter the value of linguistic choices as resources in the marketplace. The history of French and English in Quebec and Ontario up to 1960 resulted in 'conventions of language use and language value [which] were indexical of an ideology of ethnic relations in which ''national" character suited each group to its place in the economic and social order' (p. 128). Since then, francophone ethnic (national) mobilisation destabilised this situation. Codeswitching reflected the renegotiation of power relations such that the symbolic value of French and English was altered. Individuals develop verbal repertoires 'by virtue of their social position' (p. 130) and use the linguistic resources they have to fill social goals involving strengthening or erasing group boundaries. Some people do not codeswitch because they lack the resources to do so (e.g. anglophones in Ontario who have not had contact with French or francophones in Quebec and Ontario who have not had contact with English). Some do not codeswitch because they want to maintain ethnic boundaries (mobilising francophones in Quebec in the 1970s, mobilising francophones in Ontario and English-rights activists today). Some do codeswitch to cross or level ethnic boundaries (anglophone parents who study French and enrol their kids in French immersion programmes because they see economic gain in being able to use French, and French managers who need anglophone expertise). Some codeswitch to ease inter-ethnic tensions (schoolchildren may do French-associated tasks without contradicting Englishassociated peer relations in a French language elementary school). Codeswitching is seen to be less a matter of social transformation than of realigned power relations among ethnic groups. In the situations Heller describes, it is more a matter of francophone control of resources having replaced anglophone control than of an alternative marketplace having been created. English is still highly valued but due more to its role in a global economy than to its stature in Canada. To be elite in Canada today, the ability to use French is much more important than it had been. Both English and French elites are learning each others' language. Providing a nice counterpoint to the Canadian situation described by

< previous page

page_10

next page >

< previous page

page_11

next page > Page 11

Heller, Jeanine Treffers-Daller's 'French-Dutch Codeswitching in Brussels: Social Factors Explaining its Disappearance' examines a situation in which the political and economic discourse context of codeswitching results in minimising rather than expanding mixed language use. Based on the work of Poplack (1980, 1987) and Nortier (1989), Treffers-Daller hypothesises that proficiency in French and Dutch would be a predictor of codeswitching behaviour. It was expected that knowledge of standard Dutch and attendance at Dutch (rather than French) schools would correlate negatively with codeswitching. In addition, positive attitudes toward codeswitching were expected to correlate positively with its occurrence. That is, people in Brussels who think it is appropriate to mix French and Dutch regularly (i.e. to use codeswitching as an unmarked choice) were thought to be likely to do so. Echoing Myers-Scotton, Treffers-Daller assumed that codeswitching as an unmarked choice does NOT occur when the two codes symbolise groups in conflict. In fact, her conjecture is that this may be why French-Dutch code-switching has disappeared. Today only old people in Brussels associate themselves with both languages while the young choose to associate with one language or the other (either French or Dutch). In addition, Brussels French and Brussels Dutch are being replaced by their respective standards. This is in marked contrast to the situation detailed here in Part One where the opposite is happening in urban areas in relatively newly independent Africa. In the African situations examined in these pages, standard language varieties are invoked primarily by elites or educated people seeking to maintain their privileged position (e.g. Campus Kiswahili as described by Blommaert). Treffers-Daller examined the conversations of 34 people between the ages of 21 and 80 from Anderlecht (where Standard Dutch is spoken more widely than in Brussels) and from central Brussels (where codeswitching is widespread). All the Anderlecht speakers were under 50 years of age. After calculating mean sentence length in her corpus and calculating via number of French and Belgian words per sentence how many French and Dutch sentences she had, she arrived at a total number of possible intersentential switches. The data reveal no correlation between age and codeswitching initially but when the usage of people over 60 was specifically examined, there appeared to be a trend toward more intrasentential switching. Treffers-Daller's focus is on why younger people seem to make little use of intrasentential switches. She found that knowledge of standard Dutch was an important predictor of switching within a sentence. If a person had good command of standard Dutch, that person could get away with it, without recourse to French. Multiple regression analysis revealed schooling to be the most important predictor of intersentential switching along with speaker's background. In Dutch schools, the teaching of 'correct Dutch' is strongly emphasised (more than French in French schools)so, intersentential switching is less popular among Brusselers 'who speak standard Dutch relatively often.' Such people who do use French in their conversations disapprove of their own usage.

< previous page

page_11

next page >

< previous page

page_12

next page > Page 12

Treffers-Daller argues (p. 154) that it is less easy for Brusselers and Ottawa-Hullians to codeswitch than, e.g. for Puerto Ricans in New York (as studied by Poplack) to do so. Identity conflicts in Belgium and Canada are more politicised. The two papers in this section, when considered with those earlier in the volume, clearly point out the way political factors influence the actual realisation of codeswitching as a marked or unmarked code without regard to degree of language contact among speakers. Situational Uses of Codeswitching Community wide, by language learners and across dialects, registers, and genres With Taylor (1975), DeBose argues for looking at language use in the black community as a whole rather than focusing (as many scholars have over the years) on distinguishing black and standard English. In his paper 'Codeswitching: Black English and Standard English in the African-American Linguistic Repertoire,' he argues for markedness of choice as an aspect of communicative competence that allows appropriate language use to be seen in a community-wide context. Using Nick Faraclas' methodology, BE (Black English) speakers were taped in an ingroup situation, i.e. where BE is the appropriate code. The tapes revealed that everyone codeswitched, no one stuck to BE. As conversation progressed, switches to BE increased while at the outset the conversations could have been construed to be in Standard English (SE). Much of the BE involved words attributed to others (as if the speaker were taking another's part and doing it 'in dialect'). This data is presented as counterevidence to claims that BE is spoken mainly by the poor and uneducated. P, the person scrutinised most in these conversations, is a middle-aged, middle-class adult male, raised outside the south. DeBose's working hypothesis that SE is the unmarked choice of bidialectal African-Americans for communicating with outsiders is confirmed. African-Americans who use SE are in a position quite the reverse of Brussels' Dutch speakers. It pays bidialectal AfricanAmericans with social and economic aspirations to use SE in political contexts, saving BE for use at home. Within the black community in the United States, people of all socio-economic circumstances retain and use BE as the unmarked choice with each other. However, in quoting and self-characterising, SE-BE switches are unmarked. Debose finds that community-wide rights and obligations need to be considered in order for language choice patterns to be understood, be the pattern a matter of BE or SE or BE/SE mixed. When an African-American codeswitches it is in the absence of social constraints. When the balance of rights and obligations changes, the African-American uses SE only. Codeswitching represents an unmarked choice; SE, on the other hand for the African-American, is a marked choice.

< previous page

page_12

next page >

< previous page

page_13

next page > Page 13

Susan Burt, in contrast to DeBose, is concerned with micro- rather than macro-community speech norms and argues for scholars to consider combining the tools of speech accommodation theory (where convergence and divergence are important notions) and the insights of pragmatics (especially with regard to politeness theory) in conversation analysis. In 'Code-switching, Convergence and Compliance: The Development of Micro-Community Speech Norms,' Butt looks at speakers of different first languages who have learned each other's first language, i.e. pairs for whom there is no unmarked code. Specifically, she is concerned with English-speaking American students who have learned German in Germany. When these are matched for sex, class, socio-economic factors etc. what happens? '. . . in the absence of shared norms for code choice and for the interpretation of choices (the absence of "parochial pragmatics," as Green [1990] puts it), speakers are thrown back on universal pragmatic principles for the interpretation of their exploratory moves' (p. 171). Codeswitching is pragmatically ambiguous; if students, speakers, interlocutors continue to be acquaintedif they establish a speech micro-communitythey use codeswitching to develop shared conversational norms. Burt uses the term compliance to refer to codeswitching sequences in which speakers accommodate more to interlocutors than they do in situations of convergence. In conversations where compliance occurs, speakers successfully have established shared norms for future conversations. Situations will result with each speaker in a two-party conversation using the first-language of the other party as a second language in a conversation. That is, a speaker will switch to the language of an interlocutor (which for the speaker is a second language). Burt explains that the pragmatic dilemma faced by such speakers may be seen in terms of face wants (Brown & Levinson, 1987). To be unimpeded in speech, negative politeness would indicate enabling a speaker to use a first language. In contrast, positive politeness would involve using a second language to enable another to use a first language. Here, where each is equally able to choose either strategy a dilemma occurs since the speaker does not know initially which face want is uppermost in the mind of an interlocutor and an interlocutor does not know whether to converge or diverge in response. Myers-Scotton's (1983) set of maxims, drawn on in many papers discussed earlier as a guide to language choice, are seen here to underline the speaker's dilemma. If speakers are aware of both the deference maxim (defer to those from whom you want something) and the virtuosity maxim (make a marked choice to avoid being infelicitous), 'two possible interpretations are available for any individual code choice' (p. 173). Speakers of German as a first language who converge to English might lead native speakers of English to think they are being deferential OR that they think the, e.g. Americans don't know German well enough to handle the conversation successfully. Codeswitching provides a way for such speakers to 'hedge their implicative bets' (p. 173). Further if speakers want to invoke

< previous page

page_13

next page >

< previous page

page_14

next page > Page 14

the roles and obligations of either native or non-native speakers, language choice will be governed by that want. Codeswitching, then, may be used as 'a vehicle for speaker conflict' (p. 173). In such a double-bind situation, as Burt refers to it, codeswitching does occur as an unmarked choiceeach switch carries much meaning, i.e. is pragmatically ambiguous. Depending on approach to the data, a given conversation may be seen as either convergence or complementary schismogenesis. In this latter case, speakers may be seen as accommodating to each other so that they end up speaking each other's first language. Alternatively, speakers may model their behaviour to what they would like to see interlocutors display. The American learning German will try German to get the German native speaker to speak 'naturally' and thereby be helpful to the learner. However, the German may well see the situation as one in which to practise English. As Burt notes, '. . . since the two individuals differ on the values assigned to the behaviours in question, each continues to behave in precisely the opposite fashion than that which her interlocutor wishes' (p. 175). A contrasting pattern occurs wherein a speaker complies with an interlocutor's code choice (speaking the language the interlocutor begins in or switches to). Here we have compliance rather than complementary schismogenesis. Burt argues that the type of codeswitches that occur under compliance differs from the type where convergence (complementary schismogenesis) takes place. In the case of compliance, there is an extended period of codeswitching to a language other than that which the interlocutor last spoke. With convergence, there is an extended period of use of a language other than that which the speaker herself last spoke (p. 174). In 'Chinese Preschool Codeswitching: Mandarin Babytalk and the Voice of Authority,' Catherine Farris discusses her study in a Taiwan preschool where, using Mandarin Chinese, teachers shift between a babytalk register and an adult register. Farris' developmental perspective reveals that a baby-talk register in Mandarin has a socialising function in the preschool context. The use of babytalk in codeswitching amounts to metaphorical switching (Blom & Gumperz, 1972) evoking the parent-child relation in, e.g. teacher-child or lover-lover interactions. Research has been done on babytalk's role in child socialisation but it has not yet been looked at from a codeswitching perspective. Parent and teacher speech modifications in the form of babytalk have been shown not to be involved in the way a child learns a language. Instead, what is discussed in babytalk marks material in daily life that has relevance to children specifically as opposed to adults. Interestingly, Elinor Ochs (1982) found a relative absence of a caregiver register in Western Samoa. Could this mean that the child and adult share their worlds more? In Taiwan, as reported here, Farris found that there is a relative prevalence of babytalk on the part of Chinese preschool teachers to children between the ages of 2½ and 6. Teachers switch to babytalk to soften their message

< previous page

page_14

next page >

< previous page

page_15

next page > Page 15

to the younger children. The amount of codeswitching decreases in inverse relationship to the age of the child being addressed. Babytalk is used to young children by adults as a voice of authority alternating with adult speech. This paper shows how registers alternate as the teacher talks to the younger and then progressively older children. Farris distinguishes a light and heavy babytalk register. As children grow up, they acquire the ability to use the babytalk register to plead, or whine, or cajole (much as it had been used to them) and also to use it to talk to younger children. A switch to babytalk by adults (as we have seen throughout to be the case with codeswitching in general) is a way to mark or highlight a language situation to change the balance of power in an interaction, moving the relationship from one of distance to closeness, for manipulative purposes. Babytalk by adults to children (or others) functions in the interest of social manipulation rather than language learning. When teachers aren't getting anywhere they switch to babytalk as the voice of authority. In Mandarin, the term sajiao refers to the petulant speech style that obtains when children are trying to get their wayi.e. to the use of babytalk by children in order to manipulate adults. A woman may use sajiao to a lover, and by so doing, flouts the normal conversational unmarked choice of the adult register in order to alter the regular set of rules and obligations that hold between a man and a woman to that of intimates. By using babytalk, a woman is able to manipulate her lover and simultaneously alter the social power differential that usually holds across genders. The language of love is equal, ageless, manipulative, and intimate and it involves codeswitching between adult speech and babytalk, perhaps universally. The final paper in this collection 'Codeswitching for Humour and Ethnic Identity: Written Danish-American Occasional Songs' by Marianne Stflen details a bilingual songwriter's use of codeswitching strategies for purposes of humour. Her study took place in a Danish-American social club, Harmonien, where people get together primarily to have fun but also to reinforce ethnic ties. This paper is somewhat of a departure from the usual work on codeswitching since it focuses neither on speech nor conversation but on songs embedded in writing. In another paper based on this research, Stflen (1990) showed that the secretary of this Danish organisation produced minutes in which written language alternation reflected attempts to keep English intrusion to a minimum when the future of the organisation may have been in doubt. This showed up in the minutes of meetings where it was clear that English words had been replaced by Danish. When things were relaxed, the minutes would show loan-blends and bilingual punning (transliteration), e.g. Danglish. But, Stflen found little evidence of actual codeswitching in the data. This finding runs counter to what the literature on codeswitching in general leads us to suspect. In general, where there is perceived to be no danger of a change in the balance of power, codeswitching as the unmarked choice may come to be the norm in linguistic interaction. That this was not

< previous page

page_15

next page >

< previous page

page_16

next page > Page 16

the case in the minutes is analysed by Stflen as due to the formal and constrained style demanded by minutes as a genre. Intrigued by this apparent contradiction, Stflen decided to look at the songs composed for Harmonien celebrations which had been written down by the same person who kept the minutes. The purpose in this study is to see how Danish and English, in a less formal written genre, might interact. The prediction was that codeswitching would occur frequently in the songs, i.e. that the written songs might evince codeswitching as an unmarked choice much as everyday conversation would among DanishAmericans. The 20 songs examined represent the Danish genre lejlighedssange 'event songs' or 'occasional songs' which have historical roots in street ballads sung on special occasions. These songs, written down by the secretary and which go back to the 1920s, reveal a definite style associated with members of the Harmonien. This style, in part, has to do with the appearance of more markers of bilingual conversation appearing in the songs than in the minutes where the secretary purposely tried to keep English out. Interestingly, in the songs, codeswitching could be used to complete a rhyme. The composer could take advantage of the two codes for increased virtuosity. In writing minutes, on the other hand, the secretary of the group was constrained by the genre 'minutes' to avoid language play and bilingual humour. In the event songs, codeswitching emerged as a strategy used to mark 'common heritage and ethnic distinctiveness.' (p. 228) Conclusion A rather comprehensive survey of codeswitching theory and description cross-culturally is presented in this special issue. Throughout these pages, codeswitching is generally seen to be the use of at least two languages (or dialects or registers) within a particular genre (song, conversation) during a speech event often in a multilingual (primarily urban) setting. This is an elaboration of Heller's (1988) definition of codeswitching as 'the use of more than one language in the course of a single communicative episode.' Codeswitching as a process may involve single lexical items, borrowing of whole phrases, and the alternation of codes intrasententially to such an extent that little is gained from structurally isolating codeswitching types. It is not so much a matter of structural typology as it is an aspect of the kinds of multilingual contact situations in which it takes place. These papers bring out the fact that codeswitching behaviour needs to be seen in its sociohistorical as well as ethnographic context. Attitudes toward codeswitching, and the actual degree to which it is practised is not solely a matter of language contact but may also be a reflection of ethnic and race relations (as seen in the papers here about Africa and the US), political and economic realities (see especially Heller and Treffers-Daller, this volume), generational differences (see especially Treffers-Daller again, and also Farris and Stflen,

< previous page

page_16

next page >

< previous page

page_17

next page > Page 17

this volume), and factors of 'covert prestige' value within a community (see Swigart, this volume). Codeswitching may also be a community's normative way of speaking as in the case of Urban Wolof in Dakar or the variety of Swahili used in Bukavu. A number of papers have also underscored the fact that codeswitching itself, in particular settings or situations, may take on a normative role and become a distinct mode of speech in itself (as in Blommaert, Goyvaerts, Swigart's contributions). References Biota, J. P. and Gumperz, John J. (1972) Social meaning in linguistic structure: codeswitching in Norway. In J. Gumperz and D. Hymes (eds) Directions in Sociolinguistics. NY: Holt, Rinehart, Winston. 407-34. Bourdieu, P. (1977) L'économie des échanges linguistiques. Langue française 34, 17-34. Brown, P. and Levinson, S. C. (1987) Politeness. Cambridge: Cambridge University Press. Fabian, Johannes (1982) Scratching the surface: Observations on the poetics of lexical borrowing in Shaba Swahili. Anthropological Linguistics 24, 14-50. Gumperz, John J. (1982) Discourse Strategies. Cambridge: Cambridge University Press. Heller, Monica (1988) Codeswitching. Berlin: Mouton de Gruyter. Levelt, Willem J. (1989) Speaking: From Intention to Articulation. MIT Press. Myers-Scotton, C. [as Carol Myers Scotton] (1983) The negotiation of identities in conversation: a theory of markedness and code choice. International Journal of the Sociology of Language 44, 115-36. (1989) Elite closure as a powerful language strategy: the African case. Paper presented at spring seminar Language in Power, University of Washington Program in the Comparative Study of Ethnicity and Nationality (CSEN). May. (forthcoming) Accounting for Structure in Codeswitching: The Matrix Language Frame-model. Nortier, J. M. (1989) Dutch and Moroccan Arabic in contact: Code switching among Moroccans in the Netherlands. Unpublished PhD thesis. University of Amsterdam. Ochs, Elinor (1982) Talking to children in Western Samoa. Language in Society 11, 77-104. Poplack, Shana (1980) 'Sometimes I'll start a sentence in Spanish y termino en espanol': Toward a typology of codeswitching. Linguistics 18, 581-618. (1987) Contrasting patterns of code-switching in two communities. In E. Wande et al. (eds) Aspects of Multilingualism. Proceedings from the Fourth Nordic Symposium on Bilingualism. 1984, 51-77. Stflen, Marianne (1990) Harmonien: an ethnohistorical sociolinguistic analysis of a Danish-American organization. PhD thesis. University of Washington. Taylor, Orlando L. (1975) Black language and what to do about it: Some Black community perspectives. In Robert L. Williams (ed.) Ebonies: The True Language of Black folks. pp. 29-39. St. Louis: Institute of Black Studies.

< previous page

page_17

next page >

< previous page

page_18

next page > Page 18

< previous page

page_18

next page >

< previous page

page_19

next page > Page 19

Comparing Codeswitching and Borrowing Carol Myers-Scotton Linguistics Program, University of South Carolina, Columbia SC 29208, USA Abstract An issue regarding codeswitching discourse is the extent to which material from a donor language (the Embedded Language or EL) appearing in a recipient language (the Matrix Language or ML) shows internal differentiation. Three questions are relevant: (1) Are all singly-occurring EL lexemes in such discourse borrowed forms, or are some codeswitched forms (CS forms)? (2) If some are CS forms, how are they differentiated from borrowings? (3) What is the relationship of either established borrowings or singly-occurring CS forms to multiword stretches of codeswitching? Working within a model of the structural constraints on codeswitching (the Matrix Language Frame Model), this paper argues that while some important differences do exist among the various forms of EL material appearing in codeswitching discourse, in general the forms arise from related processes. Therefore, any model of the structural aspects of codeswitching must provide a unified account for all EL material in codeswitching utterances. Introduction This paper 1 attempts to clarify the status of lexical material from an Embedded Language (EL) which appears in codeswitching discourse in a frame provided by the Matrix Language (ML). Such a discussion is an important preliminary in assessing competing models of structural constraints on codeswitching, since a first step is to identify those linguistic items which properly must be accounted for as part of codeswitching utterances. Codeswitching involves at least two languages used in the same conversation. Of these languages, one is the ML, the language which sets the morphosyntactic frame for codeswitching utterances. The ML can be identified on the basis of relative frequency of morphemes. In turn, psycho- and sociolinguistic factors influence the choice of the ML. One or more languages may serve as embedded languages, providing both singlyoccurring lexemes in

< previous page

page_19

next page >

< previous page

page_20

next page > Page 20

constituents otherwise in the ML, and also EL islands, constituents entirely in the EL. The major interest here is to discuss the similarities and differences between those singly-occurring lexemes from the EL which are borrowed forms (hereafter B forms) vs. those which are codeswitching forms (hereafter CS forms). The paper begins by providing examples of the type of data to be considered. Examples (1) and (2) illustrate sentences from codeswitching discourse. They include EL-origin lexemes in constituents otherwise in the ML. The ML is Swahili in (1) and Shona in (2). (1) Setting: A man from the Luo ethnic group and another from the Kisii ethnic group are talking about the relative merits of some driving schools. They speak mainly Swahili, but with some switches to English. Kisii: Kitu kizuri ni kwa wewe ku-visit karibu kila shule ili uone difference. thing good is for you to-visit nearly each school sothat you-see difference (The best thing for you is to visit almost every school so that you can see (the) difference.) (Myers-Scotton Nairobi conversation, No. 15) Note that shule 'school' is a long established borrowing into Swahili (originally from German); however the verb stem -visit and the noun difference are considered codeswitching forms, largely because they show no frequency of occurrence; they occur only in this one conversation in a set of 40. (2) Setting: A welder, who has schooling only through grade seven, is being interviewed in Chitungwiza, a suburb of Harare, Zimbabwe. The interview is in Shona. Interviewer: Ko kana pa-tii time munenge muchitaura nezvei? (When you have tea time, what do you talk about?) Welder: Tinotaura nyaya dzekumba nedze weekend, especially Monday unotaura zvaunenge uchiita pe-weekend. (We talk about our homes and about the weekends, especially Monday you talk about what you did on the weekend.) (Bernsten & Myers-Scotton Zimbabwe interview No. 21) Note that weekend is a codeswitched form in this data corpus (it appears in two or fewer interviews). Monday is a borrowing, appearing in three conversations, as is especially, appearing in 10 out of 129 interviews. Frequency as a criterion to establish borrowing will be discussed below. The Argument of This Paper Singly-occurring EL lexemes (both B forms and CS forms) resemble each other more than they differ, but they are not identical. Their major similarity is that they undergo largely the same morphosyntactic procedures from the ML during language production. Also, some B forms arise as CS forms.

< previous page

page_20

next page >

< previous page

page_21

next page > Page 21

However, the constraints on their occurrence are different. This is a reflection of the fact that B forms have become part of the ML mental lexicon; whereas CS forms remain as EL material which only occurs in ML morphosyntactic frames during codeswitching discourse. Accordingly, the constraints on the occurrence of CS forms are specially related to those governing multiword codeswitching material (i.e. EL islands). Therefore, a continuum of relationships exists between borrowing and all forms of CS material so that codeswitching and borrowing are not distinct phenomena, as some have suggested (e.g. Poplack, 1980; Sankoff, Poplack & Vanniarajan 1990: 97). It follows that an adequate model of the morphosyntactic constraints on codeswitching will account for both types of singlyoccurring lexemes from the EL, B or CS forms. The Data Base The major data sets 2 considered are two. 1. A set of naturally-occurring conversations (N = 40) based on about 20 hours of audio recording in Nairobi, Kenya provides some examples of both B forms and codeswitching utterances. In these conversations, Swahili is the ML and the English is the EL. 2. Quantitative data on B forms is considered from a set of 129 interviews, for a total of about 22 hours of speech, audiorecorded in two different locales in Zimbabwe, one urban and one rural. All interviewees and interviewers were native speakers of Shona, and Shona was the main medium of the interview, but interviewees used many English B forms and also engaged in some codeswitching. In both Kenya and Zimbabwe, data were collected by local residents serving as research assistants. In Kenya, most speakers were not aware they were being recorded; this step was taken to ensure that the conversations were indeed natural. However, speakers' permission to use the recordings anonymously was sought after the recording. In Zimbabwe, interviewees were not aware that the real purpose of their being interviewed was to study their use of English loan words; the interview dealt with matters in their everyday lives. The Matrix Language Frame Model (MLF) Before a discussion of the characteristics of CS vs. B forms, a model to account for all codeswitching utterances will be presented in brief. (See Myers-Scotton forthcoming for a full discussion3). The model is called the Matrix Language Frame Model (MLF model) because its centrepiece argument is that codeswitching takes place within a frame set by the Matrix Language (ML). It places codeswitching restrictions at a relatively abstract level which is 'pre-syntactic' within any overall model of language pro-

< previous page

page_21

next page >

< previous page

page_22

next page > Page 22

duction. Among other things, this means that the provisions of current syntactic models are not considered adequate to account for codeswitching structures. While the premise of the model is that structural configurations in code-switching utterances can be accounted for by a limited set of structurally-based hypotheses, codeswitching is a social phenomenon. That is, if code-switching occurs at all in a community, when it occurs, and to what extentall of these matters have a psycholinguistic/sociolinguistic basis (cf. MyersScotton, forthcoming). The model proposes that there are two hierarchies which shape code-switching utterances and are especially salient when constituents consisting of morphemes from both languages (ML + EL constituents) are produced: (1) The ML is more activated than the Embedded Language (EL). (2) There is differential accessing of content vs. system morphemes. Content morphemes are similar to 'open-class items' and system morphemes are similar to 'closed-class items'; noun and verb stems are prototypical content morphemes, and inflections and articles are prototypical system morphemes. The Concept of the Matrix Language Since the ML plays such a major role in codeswitching utterances, identifying the ML vs. the EL objectively is essential. For the sake of brevity, let it suffice to say that a frequency metric will identify the ML: The ML in any CS utterance is the language of more morphemes in the type of discourse where the conversation in question occurs, if cultural borrowings for new objects or concepts are excluded from the morpheme count. That is, the ML can be identified empirically, provided a discourse sample (i.e. more than just the sentence in question is considered). 4 Typically, the ML is the more unmarked choice for the interaction, in the sense of 'markedness' as discussed in Scotton (1988a) in connection with social motivations for CS. But, of course, there are interactions where codeswitching itself is the unmarked choice; this is why the frequency metric is proposed. Types of Constituents in Codeswitching The MLF model identifies three types of codeswitching constituents, governed by related constraints (ML + EL constituents, ML islands, and EL islands).

< previous page

page_22

next page >

< previous page

page_23

next page > Page 23

ML + EL Constituents These consist of any number of ML morphemes and (generally) single-lexeme EL forms. It is the comparison of these EL forms with B forms (also from the EL) which is the main subject of this paper, of course. In examples (3) and (4) from Swahili/English codeswitching, u-na-m-time and plates tatu illustrate these constituents. While the claim here is that these EL forms are codeswitching material, note that their status is controversial. For example, Poplack and her associates refer to (most of) them as 'nonce borrowings' (e.g. Sankoff, Poplack & Vanniarajan, 1990); but others, such as Nortier (1990: 209), see no motivation for categorising such forms as 'nonce borrowings', at least in their data. (3) Setting: Several young men in Nairobi are discussing their attendance at a dance. Luyia: . . .wewe ulikuwa umejikunja kwa corner u-na-m-time tu. 2nds-PROG-her-time just (. . .you had folded yourself in a corner (and) you were just 'timing' her.) (Nairobi conversation No. 16) (4) Setting: Later in the same conversation: Nandi: Wewe, Ben, siku ile ulisosika 5 plates tatu safari moja! plate(s) three journey one (You, Ben, you really ate a lot that daythree plates at a go!) (Nairobi conversation No. 16) ML Islands ML islands in codeswitching utterances consist only of ML morphemes. They are well-formed according to the ML grammar; they must show internal structural dependency relations (i.e. show more than linear juxtaposition). In (5), inaanza usiku 'it begins in the night' exemplifies an ML Island. (5) Setting: Two Luyia men discuss the recent rains. One man: Na nyumbani imezidi hapa. Inaanza usiku na kuendelea throughout the day. (And at home it exceeds (the rains of) here. It begins at night and continues throughout the day.) (Nairobi conversation No. 34) EL Islands EL islands parallel ML islands. They are composed only of EL morphemes, following EL grammatical constraints, and show internal structural dependency relations. In (5), throughout the day is an example of an EL island.

< previous page

page_23

next page >

< previous page

page_24

next page > Page 24

The Central Hypotheses of the MLF Model The MLF Model is structured by three central hypotheses, with one supplemental hypothesis. The ML Hypothesis The ML hypothesis: The ML frames the morphosyntax of ML + EL constituents. Two empirically verifiable principles realise this hypothesis: The Morpheme Order Principlestates that morpheme order in these constituents must follow that of the ML. Swahili can be identified as the ML in the examples cited. Thus, in (4) plates tatu 'plates three' follows the head-first order of Swahili. Or, in (3) u-na-m-time (you-PROG-her-time) 'You were timing her', Swahili order prevails. The System Morpheme Principlestates that 'active' system morphemes come only from the ML in these constituents. For example, in the VP in (3), u-na-m-time, the inflections are all from Swahili. ('Active' morphemes are those which participate in relationships within the sentence which are external to the head of the morpheme itself.) System Vs. Content Morphemes What is a system morpheme? Space does not permit a complete discussion of the system vs. content morpheme distinction. 6 In brief, three features can distinguish the two types of morphemes.7 (1) Categories of morphemes with a plus setting for the feature [Quantification] are system morphemes. Such categories pick out particular individuals or events. For example, this means that [+ Quantification] is a property of quantifiers, determiners, and possessive adjectives (since they pick out individuals). It also is a property of such morphemes as tense and aspect, as well as certain adverbs (since they pick out events). (2) Positive settings for two other features indicate content morphemes. These features are [Thematic-role Assigner] and [Thematic-role Receiver]. Most verbs and prepositions are thematic-role assigners. Nouns, free-form pronouns, and descriptive adjectives are typically thematic-role receivers. However, those morphemes with a minus setting for these features are system morphemes (e.g. the copula, which does not assign thematic role). Data to Falsify the ML Hypothesis In general, all available data offer strong empirical support for the ML Hypothesis. That is, neither the data sets studied for this paper nor examples in any available published studies on CS show more than one or two counter-examples to the ML hypothesis. Evidence falsifying the ML Hypothesis would show either or both of two features: (1) In ML + EL constituents, the morpheme order would violate that

< previous page

page_24

next page >

< previous page

page_25

next page > Page 25

specified by the ML. However, only one such example occurs in the Nairobi corpus, while there are more than 100 examples showing ML morpheme order in mixed constituents, in support of the Morpheme Order Principle. (2) In ML + EL constituents, 'active' system morphemes (i.e. those with governing relations external to the lexical head) from the EL would be present. Again, only one such example occurs in the Nairobi corpus in contrast with over 150 examples showing only ML system morphemes. The Blocking Hypothesis A Blocking Filter prevents surface realisations from ML lemmas (the building blocks of ML + EL constituents) by noncongruent EL content morphemes. Levelt (1989: 6) uses the term lemma information (lemma for short) to characterise all of the non-phonological part of an item's lexical information. Lemmas are in the mental lexicon, which Levelt refers to as 'the store of information about the words in one's language'. Congruency between ML lemmas and EL content morphemes depends on matchings in regard to system vs. content morpheme status and also related subcategorisation restrictions. 8 That is, even if the EL realises a given grammatical category as a content morpheme, if it is realised as a system morpheme in the ML, the ML blocks the occurrence of the EL content morpheme in ML + EL constituents. The filter also blocks an EL content morpheme if it is not realised in the ML with a congruent content morpheme; non-congruence results when there is not a match for an EL morpheme in the ML regarding subcategorisation on its head in the maximal projection of which it is a complement. Content Morpheme Insertion This hypothesis takes on importance at the stage of content morpheme insertion, or after the frame is in place (consisting of ML morpheme order and ML system morphemes, in accordance with the ML Hypothesis). While any ML content morphemes may be chosen, the Blocking Hypothesis limits the set of EL content morphemes which are permitted. There is congruence between most nouns in most codeswitching examples, a reason why so many singly-occurring CS forms, and ultimately B forms, are nouns. Examples of Congruency: Noncongruency Noncongruencies in data studied to date are most apparent with prepositions and pronouns. For example, some English prepositions which are content morphemes (i.e. they are thematic role assigners) are not congruent with any Swahili counterparts (i.e. the Swahili verb ordinarily assigns thematic roles). Therefore, these English prepositions cannot occur in ML +

< previous page

page_25

next page >

< previous page

page_26

next page > Page 26

EL constituents in codeswitching. For example, at in I wanted to find him at the stadium, has no congruent counterpart in Swahili. Therefore, at cannot occur in the codeswitching equivalent of this sentence (6); its complement, stadium, occurs as a 'bare form'. (6) Setting: Two Luyias and a Kikuyu are talking about a fourth person and his behaviour. Kikuyu: Hee, yule cousin wangu, hata ni mjinga. Siku moja nilikuwa nataka ku-m-pata stadium to-him-find (Yes, that's my cousin, he's even a fool. One day I wanted to find him (at the) stadium.) (Nairoi conversation No. 28) But in the case of other prepositions, where there is congruence between Swahili and English (i.e. there are equivalent prepositions in both languages which are content morphemes), the Blocking Filter allows the English prepositions to appear as single CS forms. In (7) English between is accessed, because it is congruent with Swahili kati ya. (7) Setting: Three women are talking about an accident. Second person: Ilikuwa between saa mbili na saa tatu asubuhi. it was between hour two (= 8am) (It was between 8 and 9 in the morning.) (Nairobi conversation No. 40) Further, while pronouns are content morphemes in English, their usual counterparts in Swahili (agreement critics) are system morphemes. Thus, the prediction is that in Swahili/English codeswitching (with Swahili as the ML) no pronouns from English will appear in ML + EL constituents. Only one exception (i.e. only one English pronoun as a singly-occurring CS form) occurs in the 40 conversations in the Nairobi corpus. Reporting on a Tamil/English corpus, Sankoff, Poplack & Vanniarajan (1990: 80) state that there are no singly-occurring English pronouns appearing as direct objects in what the MLF model would call ML + EL constituents; yet there are many Tamil pronouns as direct objects (40% of the Tamil direct objects in the corpus as a whole are pronouns). This finding also is predicted by the Blocking Hypothesis. The EL Trigger Hypothesis The EL Trigger Hypothesis is this: Accessing any EL morpheme not licensed under the ML or Blocking Hypotheses triggers the 'grammatical formulator' (i.e. 'control central' in post-conceptual stages) to inhibit all ML morphosyntactic procedures and complete the current constituent as an EL Island (i.e. by activating EL morphosyntactic procedures).

< previous page

page_26

next page >

< previous page

page_27

next page > Page 27

Relating CS Forms and EL Islands Note that the EL Trigger Hypothesis is not an isolated (i.e. ad hoc) prediction, but is a corollary to both the Morpheme Order and the System Morpheme Principles. That is, if a CS form appears in an ML + EL constituent, it must conform to these principles. The Trigger Hypothesis states that CS forms which do not meet the principles must be accessed only as parts of EL islands. Thus, single CS forms and EL islands are governed by related constraints, a reason to argue they are related phenomena, as this paper does. Exemplifying the EL Trigger Hypothesis The Trigger Hypothesis predicts for the Nairobi corpus that if an English adjective is produced prior to its head, the result must be an EL Island (since the ML Hypothesis specifies ML morpheme order, which is head-first in NPs). Also, for example, if a quantifier from English (e.g. all) is accessed, an EL Island must follow (e.g. all the clothing), since quantifiers are system morphemes and only ML system morphemes are possible in ML + EL constituents. These predictions are supported. Thus, the hypothesis predicts that an example such as (8a) is possible, but not (8b): (8a) Nitafanya kazi hii next weekend. (I will do this work next weekend.) (8b) *Nitafanya kazi hii next mwisho wa wiki. (I will do this work next weekend.) The EL Hierarchy Hypothesis While the EL Trigger Hypothesis predicts obligatory EL Islands, the EL Hierarchy Hypothesis predicts optional ones. It is best stated as two sub-hypotheses: (1) The more peripheral a constituent is to the theta grid of the sentence (to its main arguments), the freer it is to appear as an EL Island. (2) The more formulaic in structure a constituent is, the more likely it is to appear as an EL Island. Examples (9a) and (9b) illustrate optional EL Islands (every morning and the best): (9a) Setting: a schoolboy is talking about his daily schedule. Schoolboy: Mother hunipeleka every morning. (Mother takes me every morning.) (Nairobi corpus No. 20) (9b) Setting: Six young men are talking about local football players.

< previous page

page_27

next page >

< previous page

page_28

next page > Page 28

Kikuyu 1: . . .Na alikuwa the best. (. . .And he was the best.) (Nairobi corpus No. 19) ML Islands The MLF Model has little to say about ML islands. They are the least considered aspect of codeswitching as a phenomenon, requiring more study. There appear to be no structural criteria requiring ML islands to be formed; therefore, ML islands differ from EL islands in this way. They also differ from ML + EL constituents in their composition. While the ML is active as the language setting the frame for both ML islands and ML + EL constituents, the two types of constituents differ at the content morpheme insertion stage. Only ML morphemes appear in ML islands while either ML or EL morphemes are present in ML + EL constituents. Summary of the MLF Model The MLF model is based on a differential between the ML and the EL, with the ML taking the more prominent role in codeswitching utterances. The frame of ML + EL constituents comes from the ML: its morpheme order must not be violated and also 'active' system morphemes in these constituents must come from the ML. In addition, any ML content morphemes may be inserted into the frame. The role of the EL is restricted: in ML + EL constituents only EL content morphemes appear, and even there, the ML Blocking Hypothesis prohibits some EL content morphemes. Further, the only way an 'active' EL system morpheme may occur in a codeswitching utterance is in an EL island. Lexical Borrowing: Two Different Processes Few of the codeswitching studies of the 1970s and 1980s consider relationships between borrowing and codeswitching; they simply treat all single EL lexemes as B forms (e.g. Reyes, 1976). Pfaff (1979) and Poplack (1980) at least raise the problem of distinguishing the two phenomena. But almost all of these early studies analyse only full constituents (or clauses or sentences) as 'true' codeswitching. However, increasingly, researchers are considering some (or all) singly-occurring EL-origin lexemes as material which must be accounted for in any codeswitching model (e.g. Eliasson 1989; 1991). The problem is that those same codeswitching researchers who assume a sharp line between borrowing and codeswitching seem to presuppose that all lexical B forms are cultural B forms. Cultural loans, of course, are the usual textbook example of B forms. They stand for objects or concepts new to the ML culture. Evidence that many early researchers were thinking only

< previous page

page_28

next page >

< previous page

page_29

next page > Page 29

of cultural B forms is that they refer to B forms as 'filling lexical gaps' (cf. Sridhar & Sridhar, 1980: 409; Bentahila & Davies, 1983: 302). Cultural B forms are undoubtedly most common in most cases of B, a claim supported by the distribution of cultural vs. core B forms in the Zimbabwe corpus. 9 Examples include lunchhour and examination. But the point is that such B forms are not the only singly-occurring lexemes from the EL and not even the only B forms; there are also CS forms and core B forms. Cultural Borrowings and the ML In terms of their relationship to the ML, cultural B forms are best characterised as at the categorical end of any continuum of B. That is, once the lexeme encoding a new object/concept is used in the ML; it is predicted those speakers who first used it will use it again when the need to signify the same referent comes up. Therefore, the claim is that cultural B forms enter their ML lexicon abruptly.10 In the Zimbabwe corpus, those interviewees referring to school fees categorically use the English compound to do so; the only variation is in its degree of phonological integration (as [skurufiz] or [skulfiz]). Core Borrowings and the ML Because cultural loans do enter the ML abruptly, they are rather 'instant' B forms and unrelated to codeswitching as a phenomenon. The relation of core B form to codeswitching, however, is another matter. Core B forms are borrowed because certain types of contact situations promote desire to identify with the EL culture or at least aspects of it. As Haugen writes (1953: 373), 'Borrowing always goes beyond the actual ''needs" of language'. Examples of core B forms include problem and weekend from the Nairobi corpus and twenty and because from the Zimbabwe corpus. Of course some core B forms may enter the ML in the same way as some cultural B forms: a prestigious person uses them and others then follow suit. But one of the major arguments of this paper is that the unmarked mechanisms for introducing core B forms are the same production constraints on EL material when codeswitching takes place (contained in the MLF model). Core Borrowings and ML + EL Constituents The stage in CS production relevant to this discussion is the content morpheme insertion stage in ML + EL constituents. Recall that once the frame is set, content morphemes may be accessed from either the ML or the EL, as long as the EL morphemes pass the Blocking Filter. Thus, the lines are open at this stage to position EL morphemes in slots in basically

< previous page

page_29

next page >

< previous page

page_30

next page > Page 30

ML constituents. Thus, the fact that most B forms are content morphemes is entirely predictable under the MLF model, if they start out as CS forms in ML + EL constituents. That is, core B forms are first CS forms: some CS forms are ephemeral, but others are repeated until gradually they become B forms. This is the basis for speaking of B and CS forms as falling along a continuum. A different explanation holds for cultural B forms. While they may well occur alongside core B forms in ML + EL constituents, they would enter the ML with or without CS, 11 since they indeed do fill lexical gaps. Many of them are nouns because they stand for new objects/concepts. Other Views of Borrowing Vs. Codeswitching There may be many language contact researchers who have always assumed such a connection between borrowing as a process and CS, even if they did not differentiate cultural and core B forms (e.g. Gardner-Chloros, 1987: 102). But within some codeswitching research circles, the idea of such a continuum is not so acceptable, nor is the relating of borrowing to longer stretches of CS material. While Poplack and her associates now do not distinguish between singly-occurring CS forms (largely isomorphic with their category, 'nonce borrowings') and B forms (what they call established loans), they are silent on the derivational history of loans. However, they do explicitly state (Sankoff, Poplack & Vanniarajan, 1990: 98), that 'nonce borrowing' is not 'an intermediate process situated somewhere between code-switching and borrowing in the more traditional sense.' Comparing Borrowing and Codeswitching If single B and CS forms fall along a continuum, is it possible to differentiate the two? Arguments typically advanced will be reviewed, although many of them apply implicitly to distinguishing single B forms from longer stretches of codeswitching. I will suggest that absolute frequency/relative frequency is the single criterion best linking B forms more closely with the ML mental lexicon than single CS forms. This difference is related to how the forms are handled by the 'formulator' (i.e. directing morphosyntactic procedures during language production) and how they are entered in the mental lexicon, I argue. But I will also conclude that there is little reason to differentiate the two forms as processes, that is, either in terms of their derivations or the ML morphosyntactic processes they undergo for surface realisations. Certainly, there are more similarities than differences. That is, I agree with Treffers-Daller (1991) that 'both code-switching and borrowing may be considered in the first place as the interaction of lexicons.'12 I would put it a little differently in terms of the MLF model: while codeswitching and borrowing

< previous page

page_30

next page >

< previous page

page_31

next page > Page 31

are always possible between any two languages, the specific outcomes depend on the existence of 'lexicon-driven congruencies'. Comparing Phonological Integration An early position (e.g. Poplack, 1980; Bentahila & Davies, 1983) was that B and CS forms could be distinguished on the basis that B forms typically undergo phonological adaptation to the ML. But it soon became apparent that while most established B forms may well be phonologically integrated into the ML, by no means do all B forms show such integration. The same psychosociolinguistic factors favouring the borrowing of core lexemes also often fayour the non-integration of any type of B form from the donor language. Such a situation arises, I suggest, when the EL is the language of a group with more socio-economic prestige than the ML. Also, I have argued elsewhere (Myers-Scotton, 1990) that educated bilinguals may practice 'elite closure' by making their speech and use patterns different from that of the masses. Surely one form of elite closure is for bilingual elites to pronounce loans as close to the originals as possible. This is not to deny that the majority of B forms in most data sets show a good deal of phonological integration into the recipient language, or that multiple-word codeswitching utterances rarely show much phonological integration into the ML. But the point remains that one must conclude that far from all B forms can be distinguished from single CS forms on the basis of their phonological integration into the ML. Comparing Morphosyntactic Integration If phonological criteria do not clearly distinguish borrowing and code-switching, what about morphosyntactic criteria? If the issue is distinguishing single B forms and multiple-word codeswitching, it is a non-issue. Code-switching material of fullconstituent length or longer (i.e. EL islands) regularly shows no morphosyntactic integration into the ML, taking only the morphosyntax of the EL. But the problem is that much EL material in what many researchers call codeswitching consists of single lexemes or bound stems (in ML + EL constituents). These single CS forms are always integrated into the syntax and often into the morphology of the ML. (Recall, for example, u-na-m-time tu (SUBJ2ndS-PROG-OBJ3rds-time just) 'you were just timing her' in example (3) above. It is hard to see how time could be more morphosyntactically integrated than it is.) While earlier researchers often excluded single EL lexemes as instances of codeswitching, more recent studies now include them. For example, Nortier (1990: 140) states, 'By far the largest group of switches concerns insertion of single words in one language (usually Dutch) in sentences of the other language [Moroccan Arabic]. 402 such insertions were found.' Among others, Berk-Seligson (1986) also includes single lexemes.

< previous page

page_31

next page >

< previous page

page_32

next page > Page 32

Different Procedures for EL Islands Why do these singly-occurring CS forms resemble ML material in their morphosyntax and not EL islands? The answer is that they undergo different morphosyntactic procedures (those of the ML) because they are parts of different types of codeswitching constituents. Recall that the MLF model assigns different morphosyntactic requirements to these different types of constituents (those of ML + EL constituents vs. EL Islands). ML morpheme order and ML system morphemes are the rule for all material in ML + EL constituents. Thus, it is no surprise that Sankoff et al. (1990) find that English 'nonce borrowings' show the same morphosyntactic features (e.g. Tamil inflections) as indigenous Tamil word in their Tamil/English corpus. Nonce Borrowings The category 'nonce borrowing' originates with Poplack and her associates. These are singly-occurring EL forms which occur infrequently or even only once in a data corpus, but show morphosyntactic integration into the ML. According to Poplack, Sankoff & Miller (1988: 50), they are not part of the phenomena of codeswitching. But these authors also distinguish between forms 'that occur only once in our corpus ("nonce" borrowings) and those used by many speakers (widespread loans).' Two examples of 'nonce borrowings' appear in (10), car and drive: (10) anta car-ei drive paNNanum that ACC do must (we must drive that car.) (Tamil/English Sankoff et al., 1990: 80) The problems with creating such a category as 'nonce borrowings' for such forms are: (1) No explanatory value is gained in exchange for adding another category of description which is unrelated to other categories and constraints within the codeswitching model. (2) Creating a category of quasi-borrowings masks recognised similarities between either ML-origin or EL-origin material in ML + EL constituents in the production processes they undergo. (3) It also blurs distinctions between B forms and CS forms (assuming CS forms include 'nonce borrowings') as end products. They differ in that borrowing is a phenomenon open to monolinguals while codeswitching is not. More important, they differ in their freedom of occurrence.

< previous page

page_32

next page >

< previous page

page_33

next page > Page 33

Why Singly-Occurring Lexemes May be CS Forms Thus, the argument of this paper is that there is no reason to remove from the CS arena the single-lexeme EL material which cannot be identified as established B forms. True, these single CS forms are related to B forms. That core B forms arise through codeswitching has already been suggested; in the next section, the similarity of CS and B forms in the morphosyntactic procedures they undergo will be highlighted. But the point remains that single CS forms also are related to multiword CS utterances (EL Islands) in regard to their frequency and freedom of occurrence. Ultimately, this relation arises because CS material (i.e. single CS forms and EL islands) are not part of the ML mental lexicon, as are B forms. Incomplete Morphosyntactic Integration An examination of any codeswitching data corpus confirms that both established B and single CS forms do show morphosyntactic integration into the ML. Recall the example sentences cited above, with both single B and CS forms taking ML system morphemes and following ML morpheme order. But both also show incomplete integration in some cases. However, there may well be a difference between B and CS forms in this regard, with B forms showing generally more integration. Still, the difference seems to be one of degree, not kind. This subject requires further study; here I only offer some evidence and two tentative hypotheses, one concerning 'bare forms' and the other the relativity of incomplete morphological integration. The 'Bare Form' Hypothesis First, it seems that non-congruence between the ML and the EL in regard to subcategorisation in general (including that discussed above under the Blocking Hypothesis) sometimes results in single EL forms which are 'bare forms'. That is, especially nouns (but also some verb stems) appear without the system morphemes they would take in an EL utterance. In example (9) above, stadium is such a 'bare form'; it lacks a locative suffix (which it would have in Swahili). For another example, see escort in (11); it lacks any modifier. An article would be expected in English, but not Swahili; but some modifier would normally accompany it in Swahili, possibly mmoja resulting in escort mmoja (escort-one) 'an escort'. (11) Setting: Three young men from different ethnic groups are discussing problems in getting home at night. Luo man: Tu-ta-ku-pat-i-a escort mpaka nyumbani. (We will get for you (an) escort up to home.)

< previous page

page_33

next page >

< previous page

page_34

next page > Page 34

(Nairobi conversation No. 26) Such 'bare forms' have been reported in many codeswitching corpora (e.g. Berk-Seligson, 1986; Backus, 1990; they are discussed at length in Nortier, 1990 and also in Myers-Scotton, forthcoming, b). My speculation is that 'bare forms' are a type of 'speech error', resulting when an EL content morpheme is accessed under the command of an ML lemma. I suggest that the error (in the application of the ML morphosyntactic procedures) results because the congruence between the EL form and the ML lemma is less than perfect. This speculation leads to a 'Bare Form' Hypothesis: such 'bare forms' are much more frequent among single CS forms than among B forms. The motivation for this hypothesis is the claim that B forms have their own ML lemmas in the ML mental lexicon, an entry which normally includes access to all the relevant ML morphological procedures. In contrast, CS forms are only accessed through ML lemmas (at the content morpheme insertion stage). Whether a singly-occurring EL-origin lexeme is a CS or a B form would be determined by its frequency, an issue discussed below. The Morphological Integration Hypothesis Second, when morphological integration is incomplete, a Morphological Integration Hypothesis suggests this: Morphological Integration will be more incomplete on CS forms than on B forms, even though they have the same EL origin. Further, following Hopper & Thompson (1984), it is predicted that the more functional peripheral inflections are more likely to be lacking in general; however, if present, they will only occur on CS forms if they also occur on B forms. Two cases support this claim. 1. In Bantu languages, verbs carry a number of affixes. The two most central affixes are the subject prefix indicating agreement with the subject of the verb and the tense/aspect prefix. These are categorically present in all CS and B forms in the Nairobi and Zimbabwe corpora. However, the final vowel in the verb, an affix carrying little functional load in most cases, is often absent in CS forms, although it is almost always present in B forms. 13 For example, (12) from the Nairobi corpus and (13) from the Zimbabwe corpus illustrate CS forms: (12) matatu i- -li- -overturn na kugonga basi. small bus it-PAST-overturn and to-hit bus (The matatu overturned and hit a bus.) (Nairobi conversation No. 40) (13) . . .Ukaona achifamba apa u- -nga- -guess. . . if you see she-walking 2ndS-able-guess (. . .if you see a woman walking, you can guess. . .) (Shona/English Crawhall unpubl. data)

< previous page

page_34

next page >

< previous page

page_35

next page > Page 35

2. Additional evidence that B forms tend to show more morphological integration than single CS forms is supplied from an analysis of plurality marking on Shona nouns in the Zimbabwe corpus. Placing foreign nouns in Bantu class six (the prefix mamarks both the class and plurality) is a favourite strategy in southern Bantu languages in general. Bernsten (1990: 82) compared the number of B forms (defined as occurring in at least three different interviews) with CS forms (defined as occurring in two or fewer interviews) which received ma- and also showed the English plural suffix (e.g. ma-day-s). 14 She found that of the 136 tokens taking ma- which are B forms, only 17% occur with English -s as well. But 69% of the CS forms taking ma- also show English -s (86 out of 124). Both CS and B forms undergo the same ML morphosyntactic process, and both equally govern class six agreements on other elements in the sentence (i.e. agreements congruent with ma-). But one can say that the CS forms are less morphologically integrated because they are still open to EL morphological processes as well. Differences in Frequency On another front, there are several motivations to hypothesise that relative frequency should separate single B and CS forms. First, recall that the ML and Blocking Hypotheses predict that an important distinction exists between the two types of EL forms: the formulator's producing surface sentences puts no restrictions on B forms in ML + EL constituents, whether they are content or system morphemes, but it permits only content morphemes as CS forms and only those meeting the requirements of the Blocking Filter. Second, if one accepts the speculation that B forms are accessed via their own ML lemmas in the ML lexicon because they have become part of the ML (even while their originals, of course, may maintain membership in the EL mental lexicon), then B forms, as 'naturalised' ML forms, should have a similar frequency to indigenous ML forms. In contrast, if there is EL material which is accessed only through ML lemmas if there is a congruence between the ML lemma and the indigenous EL lemma, then these forms should have a more limited frequency (cf. Scotton, 1988b). The Absolute Frequency Hypothesis These observations/speculations motivate this prediction: not all EL material has the same frequency in a given corpus. This prediction is supported in the Nairobi and Zimbabwe corpora. Taking an admittedly arbitrary cut-off point, Bernsten (for the Zimbabwe corpus) and I (for both corpora) were able to distinguish EL-origin forms, based on whether they occurred in three or more conversations/interviews. Those forms occurring in three or more samples are considered B forms; others are considered CS

< previous page

page_35

next page >

< previous page

page_36

next page > Page 36

forms. That is, CS and B forms are defined by frequency. (In addition, all forms for objects/concepts new to the culture were considered cultural B forms; establishing their frequency is not at issue, as discussed above.) Relative Requency as a Criterion The frequency of a core EL form relative to its indigenous counterpart can be studied as another metric to differentiate single EL forms. Two cases from Shona, with forms from English which theoretically could be either B or CS forms, are discussed briefly. (For details, see Myers-Scotton, forthcoming, b.) Numbers in Shona: English-Origin B Forms In the Zimbabwe corpus, analysis showed that of all the numbers used by interviewees (N = 1, 257), 86% were in English. The use of numbers was studied by a number of variables; but in both urban and rural samples, for both males and females, for a variety of semantic functions, English numbers far outnumbered those in Shona. Based on this study of relative frequency, the conclusion that English numbers are B forms in Shona seems non-controvertible. Discourse Markers: More English-Origin B Forms In addition, the frequency of the discourse markers because and but was studied relative to their Shona counterparts. (Because of their discourse 'prominence' such markers often seem to be borrowed in many language contact situations (cf. Scotton & Okeju, 1973 on Swahili borrowings into Ateso).) The Shona/English discourse marker findings are much less dramatic than those for numbers, but still convincing: 8% (26 out of 367) of the encodings of 'because' were with the English morphme because; 9% (26 out of 315) for encoding 'but' are with the English morpheme but. Is Frequency too Arbitrary as a Criterion? There are clearly problems with using frequency to differentiate single B forms and CS forms, the main one being the issue of arbitrariness. But many thorny issues disappear by approaching the issue from the standpoint of CS forms. It is not that a B form must recur, it is that a CS form must not recur in order to be a CS form. This relates to my final point. ML Forms as on 'The Preferential Path' That morphemes from the two or more languages active in codeswitching are not equally accessed, that there is a 'preferential path', is the claim of the

< previous page

page_36

next page >

< previous page

page_37

next page > Page 37

MLF model. This argument is supported by the numerical and grammatical dominance of one language. One language is identified as the ML by the greater frequency of its morphemes; it turns out this same language sets the morphosyntactic frame of ML + EL constituents and also is accessed in the numerous ML Islands. Thus, there is ample quantitative evidence that both the morphemes and morphosyntactic procedures of one language are preferred. This language is called the ML. And, since B forms are, in fact, ML forms, it follows they are also preferred, with preference expressed by frequency. The outcome is that all EL material, whether single CS forms or EL Islands, is less frequent in CS utterances than ML material. This empirical finding indicates there is a clear link between all types of EL material as 'off the preferential path'. Conclusion This paper has argued that there are both similarities and differences between singly-occurring forms originating in an embedded language (EL) which occur in a recipient or matrix language (ML), whether they are B forms or CS forms. A crucial similarity is that both undergo similar, if not identical, ML morphosyntactic procedures; therefore, any CS model adequately accounting for single CS forms also necessarily accounts for B forms. The similarities in morphosyntactic treatment largely motivate the argument that core B forms arise originally as CS forms; that is, at any point in time, B forms and CS forms fall along a process continuum. Just as important, the paper claims that, for a variety of reasons, multiple-word codeswitching utterances also share a relationship with single CS forms. Both types of codeswitched material remain with entries only in the EL mental lexicon; their accessing in codeswitched utterances is possible only because the more dominant ML procedures are inhibited under special provisions outlined in the MLF model. In contrast, B forms have entries as ML lemmas and, accordingly, more accessibility to recurrence. Thus, what may be the most telling feature of all codeswitching material is its relative lack of predictability; specific codeswitching forms may appear in codeswitched discourse, but then surface again only in monolingual EL discourse. This absence of predictability supports the unity of all CS morphemes as 'aliens'. They are EL forms appearing in ML frames, distinct in this sense from B forms which have become part of the ML. Notes 1. Fieldwork for this paper was conducted under a Fulbright Research Grant to Kenya and Zimbabwe (1983) and a Social Science Research Grant to Zimbabwe (1988) as well as a University of South Carolina Research and Productive Scholarship Grant (1988); support is gratefully acknowledged. Claims here supersede a somewhat different argument in Scotton, 1988c. 2. My research associates in Zimbabwe were Janice Bernsten, who supervised much of the

< previous page

page_37

next page >

< previous page

page_38

next page > Page 38

interviewing for the Zimbabwe corpus, and also Nigel Crawhall, who collected some naturally-occurring conversations showing Shona/English codeswitching. Shem Lusimba Mbira has been my primary research assistant in Kenya. 3. The forerunners of the MLF Model appear in Myers-Scotton & Azuma, 1992; and in Myers-Scotton, 1991. 4. The ML is not fixed in any sense; that is, which language is the ML may even change within the course of a conversation (depending on a change in situational factors, such as topic or participants, in most cases). It also may change over historical time. 5. This verb form includes the stative extended form of a slang word for eating -sosal-soza, probably derived from sauce; it is not part of Standard Swahili. 6. Joshi, 1985 was one of the first codeswitching researchers to use the ideas of Garrett (e.g. Garrett, 1990) about differences in the ways 'closed class' vs. 'open class' items are accessed in monolingual speech. Basing his argument on evidence from speech errors, Garrett hypothesises that these two different types of morphemes may be accessed at different points in the language production process. I follow Joshi in using the terms 'Matrix Language' and 'Embedded language'. 7. For a fuller discussion, see Myers-Scotton (forthcoming). 8. Although their approach is different, Bentahila & Davies (1983) discuss subcategorisation restrictions as possibly limiting which forms could occur in codeswitching utterances. Azuma (1991) also discusses constraints on codeswitching material in terms of subcategorisation restrictions; his model is similar to the MLF model, but not identical. 9. I analysed Bernsten's (1990) listing of B forms in the Zimbabwe corpus. Out of 391 B forms, 68% are clearly cultural loans. Some of the remainder are borderline cases (e.g. [beg]/[begi] 'bag'; [oda] 'order'). 10. Claiming that cultural B forms enter the ML abruptly does not mean that all speakers immediately adopt such forms. Some B forms are always only part of the repertoire of only certain sub-groups within a speech community. 11. Their appearance in codeswitching utterances may facilitate the borrowing of cultural B forms; however, no evidence is available. 12. Treffers-Daller refers to Muysken (1990) for a more detailed discussion in regard to codeswitching of how syntactic properties may derive from the lexicon. 13. Of course those English-origin verb stems showing no final vowel also violate Bantu phonotactics which call for open syllables only. 14. In many codeswitching data sets, some single CS forms show 'double morphology'. Often, this means there are plural affixes present from both the ML and the EL (e.g. ma-watchmen in Swahili/English codeswitching). An argument as to why the EL affix is not 'active', but simply accessed as part of the EL stem, appears in Myers-Scotton forthcoming. References Azuma, S. (1991) Two level processing hypothesis in speech production: evidence from intrasentential code-switching. Paper at 27th Chicago Linguistic Society meeting (5/91). To appear in Proceedings. Backus, A. (1990) Turkish-Dutch codeswitching and the frame-process model. Paper at International workshop on ethnic minority languages in Europe. Tilburg, The Netherlands (12/90). Bentahila, A. and Davies, E. E. (1983) The syntax of Arabic-French code-switching. Lingua 59, 301-30. Berk-Seligson, S. (1986) Linguistic constraints on intrasentential code-switching. Language in Society 15, 313-48. Bernsten, J. (1990) The Integration of English Loans in Shona: Social Correlates and Linguistic Consequences. Ph.D. dissertation. East Lansing MI: Michigan State University. Eliasson, S. (1989) English-Maori language contact: code-switching and the free-morpheme constraint. Reports from Uppsala University Department of Linguistics 18, 1-28. (1991) Models and constraints in code-switching theory. Papers for the Workshop on Constraints, Conditions and Models. (pp. 17-50.) Strasbourg: European Science Foundation.

< previous page

page_38

next page >

< previous page

page_39

next page > Page 39

Gardner-Chloros, P. (1987) Code-switching in relation to language contact and convergence. In G. Ludi (ed.) Devenir bilingueparler bilingue. (pp. 99-113.) Tubingen: Max Niemeyer Verlag. Garrett, M. F. (1990) Sentence processing. In D. N. Osheron and H. Lasnik (eds) An Invitation to Cognitive Science, Vol. 1. (pp. 133-75.) Cambridge MA: MIT. Haugen, E. (1953) The Norwegian Language in America: A Study of Bilingual Behavior. Philadelphia: University of Pennsylvania Press. (Reprinted 1969 Bloomington IN: Indiana University Press.) Hopper, P. and Thompson, S. (1984) The discourse basis for lexical categories in universal grammar. Language 60, 703-52. Joshi, A. K. (1985) Processing of sentences with intrasentential code-switching. In D. Dowty et al. (eds) Natural Language Processing: Psychological, Computational and Theoretical Perspectives. (pp. 190-205.) Cambridge: Cambridge University Press. Levelt, Willem J. (1989) Speaking: From Intention to Articulation. Cambridge MA: MIT Press. Muysken, P. (1990) Concepts, methodology and data in language contact research: remarks from the perspective of grammatical theory. In Papers for the Workshop on Concepts, Methodology, and Data. (pp. 15-31.) Strasbourg: European Science Foundation. Myers-Scotton, C. (1990) Elite closure as boundary maintenance: the case of Africa. In B. Weinstein (ed.) Language Policy and Political Development. (pp. 25-42.) Norwood NJ: Ablex. (1991) Intersections between social motivations and structural processing in code-switching. In Papers for the Workshop on Constraints, Conditions, and Models. (pp. 57-82.) Strasbourg: European Science Foundation. (forthcoming, a) Social Motivations for Codeswitching: Evidence from Africa (in press). Oxford: Oxford University Press. (forthcoming, b) Accounting for Structures in Codeswitching: the Matrix Language Frame Model (in press). Oxford: Oxford University Press. Myers-Scotton, C. and Azuma, S. (1992) A frame-based process model of codeswitching. In K. Deaton et al. (eds) Papers from the 26th Chicago Linguistic Society Meeting. Chicago: University of Chicago Linguistics Department. Nortier, J. (1990) Dutch-Moroccan Arabic Code Switching. Dordrecht: Foris. Pfaff, C. (1979) Constraints on language mixing: intrasentential code-switching and borrowing in Spanish/English. Language 55, 291-318. Poplack, S. (1980) 'Sometimes I'll start a sentence in Spanish y termino en espanol': Toward a typology of code-switching. Linguistics 18, 581-618. Poplack, S., Sankoff, D. and Miller, C. (1988) The social correlates and linguistic process of lexical borrowing and assimilation. Linguistics 26, 47-104. Reyes, R. (1976) Language mixing in Chicano bilingual speech. In J. D. Bowen and J. Ornstein (eds) Studies in Southwest Spanish. (pp. 182-88.) Rowley: Newbury House. Sankoff, D., Poplack, S. and Vanniarajan, S. (1990) The case of the nonce loan in Tamil. Language Variation and Change 2, 71-101. Scotton, C. M. (1988a) Codeswitching as indexical of social relationships. In M. Heller (ed.) Codeswitching: Anthropological and Sociolinguistic Perspectives. (pp. 151-86.) Berlin: Mouton de Gruyter. (1988b) Code-switching and types of multilingual communities. In P. Lowenberg (ed.) Language Spread and Language Policy. (GURT '87). (pp. 61-82.) Washington: Georgetown University Press. (1988c) Differentiating borrowing and codeswitching. In K. Ferrara et al. (eds) Linguistic Change and Contact (NWAV-XVI). (pp. 318-25.) Austin: University of Texas Linguistics Department. Scotton, C. M. and Okeju, J. (1973) Neighbors and lexical borrowings. Language 49, 871-89. Sridhar, S. N. and Sridhar, K. (1980) The syntax and psycholinguistics of bilingual codemixing. Canadian Journal of Psychology 34, 407-16. Treffers-Daller, J. (1991) Towards a uniform approach to code-switching and borrowing. In Papers for the Workshop on Constraints, Conditions and Models. (pp. 259-79.) Strasbourg: European Science Foundation.

< previous page

page_39

next page >

< previous page

page_40

next page > Page 40

< previous page

page_40

next page >

< previous page

page_41

next page > Page 41

French in Urban Lubumbashi Swahill: Codeswitching, Borrowing, or Both? Marjolein Gysels University of Amsterdam, Oude Zijds Achterburgwal 185, 1012 DK Amsterdam, The Netherlands Abstract Contact phenomena are prominent in Lubumbashi Swahili and they belong to the core of the language. As such they are exploited creatively and they mean an important enrichment in the communicative tools available. The aim of this paper is to investigate the motives for the integration of French elements into Lubumbashi Swahili. Focusing on the mechanisms through which these elements are inserted, I argue that the mixing process is carefully regulated and controlled to serve several communicative functions. Fabian (1982) already found that loans are employed as stylistic devices. Here, I observe that they can also function grammatically. From this perspective, looking at the way these insertions are finally applied, two distinct mixed variants which serve different linguistic and social functions are discussed. Introduction In this paper I investigate the motivation for French elements occurring in Lubumbashi Swahili. The analysis is based on three different texts. 1 The first was taped by myself in Lubumbashi in January 1990 and consists of a story told by an elderly woman. This is followed by a discussion about the content of this story between the woman and a younger man and the rest of the listeners. The remaining two texts derive from material that was collected by J. Fabian during his fieldwork in Lubumbashi. I obtained one of them, an interview between Fabian and a cabinetmaker as a transcribed text. This recording was made between 1972-1974, as part of a project on language and work, in which Swahili was studied as it is used in the context of work. The third text is a recording of a political party speech made by the governor Duga ku Gbetoro on Radio Zaïre.2 There has not been much work on contact phenomena in Shaba Swahili. Kalunga (1979) deals with the occurrence of borrowings from different

< previous page

page_41

next page >

< previous page

page_42

next page > Page 42

languages in Lubumbashi Swahili, though he treats this from a perspective in which he opposes Lubumbashi Swahili to Standard Swahili. He maintains the view that Shaba Swahili deviates from the norm, and does not therefore recognise it as an autonomously functioning and fully fledged language, even though it has a considerable number of speakers (some three to five million in the southeast alone). Nkulu (1984) also demonstrates the incidence of French borrowing in Shaba Swahili in a survey and he focuses attention primarily on the foreign origin of these elements. 3 The Poetics of Borrowing I proceed from Fabian's (1982) argument with regard to borrowing in Lubumbashi Swahili. Fabian distances himself from previous approaches which made use of word lists and inventories, and showed only the final product in the receiving language. He explicitly rejects such taxonomic approaches which inevitably imply a referential semiotic focus. This leads to a view of borrowing as a necessary denotative matter. Fabian bases his research on recordings of spoken texts and directs his attention to the mechanisms involved in borrowing in order to understand the processes. From this point of view he is able to call into question the instrumental hypothesis which is normally accepted as accounting for borrowing. It is assumed that languages will adopt terms to fill lexical gaps. At the same time it implies a strong bias with regard to the uni-directional way in which borrowing, and thus cultural progress, takes place. In the analysis of a number of texts covering the domain of work and technologywhich should most obviously express the gap-filling nature of borrowingno direct confirmation could be found for the instrumental hypothesis: the French loans were not specifically related to work. A performative text on the other hand, demonstrated that borrowings fulfilled connotative and indexical functions. A test with the speakers of the texts in question revealed that they were able to provide a Swahili translation for two thirds of the French loans they had used in these texts. If this phenomenon is viewed as a part of a communicative praxis, the French borrowings finally prove to fulfil an expressive, performative function. The texts reveal a stylistic or poetic, in other words a creative, use of the French terms. Proceeding from these findings, and from a similar perspective with regard to the material, and given the assumption that the mixing of codes has a certain degree of regulation and some linguistic consistency, I will look at the functional and pragmatic reasons for the incidence of French elements in Lubumbashi Swahili and the inherent constraints on it. Historical and Sociolinguistic Background of Lubumbashi Swahili4 To be able to fully comprehend the actual status of Lubumbashi Swahili, I provide a rough sketch of its specific development. Initially it was assumed

< previous page

page_42

next page >

< previous page

page_43

next page > Page 43

that Shaba Swahili derived from the East Coast, and that its rise involved the development of a pidgin toward a creole. At least this was Polomé's (1971) opinion, which conforms to the more traditional (i.e. evolutionary) view about the origin of creole languages. According to Fabian, Swahili was introduced in Katanga for political reasons. He thus argues for a completely different process of development for Shaba Swahili (1986a and 1986b). On the one hand the promotion of Swahili fits in with the political power relations at that time, and it was seen as a way of creating a greater distance between the Congo and the southern British colonies. The latter constituted a danger because of their interests in the rich province of Katanga. On the other hand the introduction of Swahili was also thought to be very convenient for the creation of a stable work climate. By controlling the medium of communication, it was also possible to dominate other aspects of the life of the mineworkers. In this context the language was described in language guides (that were intended to serve as manuals for new European workers in the region) in the form in which it was used by the Europeans (Fabian, 1986b). It was presented as a deficient medium incapable of expressing complex ideas, with a restricted vocabulary and a rudimentary grammar. Although it was presented as a single general language form, which was supposedly spoken by the Africans, the descriptions in the handbooks diverged strongly. In fact, these were the whites' own improvised realisations of how they themselves communicated with the Africans. In this way several pidginised versions came into being. These variants were the consequence of a one-sided communication in a work context in which pronounced relations of power prevailed. The whites restricted their knowledge of Swahili to contexts of order and instruction and were thus able to determine and dominate the situation. Shaba Swahili did not start as a pidgin but originated at the same time as the sudden growth of the mining industry in Shaba. The African population created a popular form of Swahili, with which they could express the new urban culture that came into existence in Elisabethville. Apparently this form of Swahili, which existed in a creolised form, was not as limited as the pidgin form. The whites nevertheless only recognised the existence of two variants which they themselves created: on the one hand pidgin Swahili, that was supposed to represent the language as it was spoken by the Africans and which needed to be corrected, and on the other hand a grammaticalised variant, elaborated by missionaries with the aim of creating literacy. The latter, however, excluded most of the Africans from this elaborate and Latinised form. As a consequence, the existence of popular Swahili, together with the way in which its culture was experienced by the African urban population, was denied (Fabian 1986b: 136, 148). Today, as in many other urban contexts in Africa, the complexity of the multilingual profile of the city of Lubumbashi in which Shaba Swahili is spoken, has not diminished. Also, the linguistic situation that was con-

< previous page

page_43

next page >

< previous page

page_44

next page > Page 44

structed around Swahili during the colonial period still has its bearing upon the manner in which Swahili is perceived. In Zaïre, Swahili is one of the four 'national' languages next to Lingala, Ciluba and Kikongo. In Shaba it is still the main medium among the biggest part of the population in the cities. French, which remains from the colonial period, is the official language. It is used in education above the level of primary school, and it functions as the language of the administration and of almost all the written mass media. It is the language that implies a Western and modern orientation, and power and a career are associated with it. The prestige of French is, however, limited to those domains. Swahili carries prestige in completely different contexts, since this is the medium in which one is able to express the collective experiences of urban culture, and in the final instance it carries the identity of the population (Kabamba, 1979: 223). In addition to popular Swahili there is also a different variety that resembles the Swahili of the East Coast: 'Swahili Bora', which means 'superior Swahili'. This is the form of Swahili which was constructed by the missionaries for the purpose of writing. However, it had the effect that Shaba Swahili has always been banned from the official sphere, and that a standard form has never managed to develop. The 'Bora' form has been highly influenced by Shaba Swahili in the meanwhile. In its spoken version it is certainly not used consequently or as a full code. Rather, depending on the situation certain elements are adopted from it. Shaba Swahili is so deeply rooted in speech habits that even those who are supposed to fulfil 'gate-keeping' functions regarding the 'high' variant very often tend to give it up in the course of a conversation and continue in the local dialect. Other African languages are also present in Lubumbashi (Luba, Lunda, Bemba, Lingala, to name just a few). They were taken there by workers from various regions. Later they were passed on to subsequent generations who were born in the city. They function particularly in contexts in which ethnic backgrounds are salient (Kabamba, 1979: 56). Speech Repertoires and French in Lubumbashi Swahili As is the case in many urban contexts in Africa, an extremely complex linguistic situation is also prevalent in Lubumbashi, where a multitude of African and European elements are used. This contact situation is due to the great heterogeneity of the population, with a strong diversity of linguistic backgrounds. Most of the people of the city are multilingual. According to Kabamba (1979: 56) 79% of the population speaks Swahili and French, 78% knows an ethnic language, 21% an African, nonZaïrean language and 3% a non-African language. This situation of multilingualism, together with the lack of a standard form of Shaba Swahili, are both responsible for the most characteristic feature of this language, namely its great variability (de Rooij, 1989). In such a contact situation people draw selectively from different languages which are available. Depending upon background, a speaker

< previous page

page_44

next page >

< previous page

page_45

next page > Page 45

speaks one or more auTOC-hthonous languages, has Swahili at hand and, according to the level of education, a greater or lesser knowledge of French. As Nkulu (1984: 13) notes, speakers of Shaba Swahili classify themselves into three groups. The first group contains speakers who have been born in Lubumbashi or in the other urban centres of southern Shaba. They claim to have native-speaker command of the language and they are very sensitive to deviations from what they consider to be the norm. The second group consists of those who were born outside these centres but who have practically become native speakers in the meanwhile. They wanted to identify themselves with the first group and succeeded in adapting their language use to that of the native speakers. They constitute a group which is rather difficult to distinguish from the former with regard to performance in the language. The third group consists of those people who for all sorts of reasons (age, lack of motivation) speak a kind of Swahili that is not considered to be Shaba Swahili by the members of the two former groups. It is obvious that the speech repertoires of the Shaba Swahili speakers differ considerably. In this way people introduce their own characteristics into Swahili as a result of which their specific use of the language will differ from that of other Swahili speakers. But this also gives them the possibility of switching into their own language use, depending on the interlocutor and the situation. Lexical borrowing, codeswitching, phonological and syntactic interference and other contact phenomena may be considered as essential features of the language. They are utilised to achieve one's aims in daily communication. By simply looking at the process of mixing French and Swahili it is already possible to get a good impression of the subtlety that such language contact offers in shaping certain communicative situations. By delineating this mixed code externally to get an idea of the social role it serves we can observe that several motivations are involved in the adoption of French elements. On the basis of the analysed texts it is easy to detect two different patterns in the use of French loans and switches, which by consequence also signal a different message. What at first sight seems to be an unambiguous interference of French elements, can now be considered as two separate social and linguistic products. There are two mixed codes, which cannot be considered to be of equal importance. In the case of the first, French elements are only sporadically and rather unconsciously integrated into Swahili language use. We call this the colloquial variant. The other contains loans and switches which have grammatical functions, and here the mixed code serves as register. 5 The first text, for example reveals this explicitly. In this text an elderly woman tells a traditional story. She concludes the story with a moral lesson that it is not right to throw away everything from the past and only to live according to modern values. This is followed by a discussion between herself and a younger man and the rest of the listeriers. In connection with the moral message of the story they talk about concrete experiences from daily life, and even discuss politics.

< previous page

page_45

next page >

< previous page

page_46

next page > Page 46

In the extended parts of the text in which the elderly woman is talking, practically no codeswitches into French occur. When, however, the young man starts to comment upon the story, he makes his point by switching to a Swahili strongly interspersed with French elements. In other words with a mixed code as register. The decision to switch to a different code means that the speech situation is also defined on the basis of a number of characteristics that go together with this code. By this means the speaker dissociates himself from what was said before, which was mainly in Swahili with a single shift to French. All this comes down to the elaboration of the person's speech repertoire: in this case whether he has a good knowledge of French. The person in question is a student at the university, who therefore has to deal with French daily. Courses are given in French, the literature is in French and studying at the university already implies a certain training in the language during the greater part of primary education and throughout secondary school. The woman who speaks in the colloquial mixed code does not have this background and adds considerably less French instances in the conversation. The French elements that do show up occur in the form of stable loans, and indicate that French pervades daily life in Lubumbashi to such an extent that this mixed code is part and parcel of normal language use. The second text, is an interview between Fabian and a cabinetmaker. They discuss matters relating to work in the workshop, as well as autobiographical and more general subjects. This text shows that alternation from one code to another can also be realised intra-individually. The worker who speaks here, expresses himself in a Swahili in which French elements are almost unnoticeably integrated one moment. Later, however, he selects the mixed code as register, in which French switches are frequently applied and clearly consciously integrated. Situational and other broader social factors play an important part in this matter. The occupation of the speaker leads to regular contact with Europeans, and through this he had the opportunity to acquire a good knowledge of French, which of course indicates a certain social identity. The third text is a recording of a party speech made by the governor Duga ku Gbetoro on Radio Zaire. In this highly rhetorical speech the ideals of the country's only party, the 'mouvement populair de la révolution' are praised and justified. Here, the mixed code is seen as a suitable register for this political speech on the radio. The use of French elements is abundant and they give an even more formal touch to the whole argument. The mixed code is used here as a register implying authority and professional expertise. Both codes are distinguished by a quantitative difference in the amount of French. On the one hand the colloquial variant shows significantly less French instances than the register variant. The French words in the colloquial variant reflect normal language use while the words denoting register have grammatical functions and can be manipulated for specific effects (see below).

< previous page

page_46

next page >

< previous page

page_47

next page > Page 47

Otherwise the integration proceeds roughly in the same way. Both the mixed variants follow Swahili patterns complemented by French elements. French verbs, nouns and particles are generally preserved in their original form and the extent of fitting in the Swahili inflection system, morphology and phonology is relatively limited. French nominals are integrated into the agreement pattern very easily, mostly without, but often also with, a class prefix. The following examples can be found in the texts: (1) bibi wa patron alizala ule bébé ule (the wife of the boss gave birth to that baby) (2) inspecteur wa masomo (the inspector of the school) Bébé receives a demonstrative pronomen with a prefix of class 1, which indicates persons. Also, the connective following inspecteur is formed with a class 1 prefix. Other terms are placed in class 9-10. This is traditionally the class which absorbs foreign items. 6 (3) ile modèle ya zamani (that model from earlier times) (4) tunafanya calcul yetu (we make our calculation) I have found French words with a Swahili class 2 plural marker, but there are also examples in which French forms of the plural are preserved. (5) ya bapères Blancs ndiyo (from the white fathers yes) (6) baclients banakatala (the clients refuse) (7) meubles yote (all the furniture) In the texts there are also class 5 (ma-) and class 13 (ka-) prefixes which allow themselves to be connected to the French terms. Those prefixes do not posses the function of indicating the relation to a class that covers a certain semantic domain, for they get attached as facultative prefixes at places where plurality or collectivity (ma-) or a diminutive- or pejorative meaning has to be marked.7 (8) wanatrouver kila mara mamots difficiles ndani (they find every time difficult words in it) (9) kuna kabar kamoja (there is a little bar) French verbs are easily brought into the syntactic pattern of Swahili verbs. They can receive any possible prefix. When they are incorporated they behave like Swahili verbs, and we may accept that the concerted action between the aspect markers and the semantic content of the verbal roots

< previous page

page_47

next page >

< previous page

page_48

next page > Page 48

also follows the same patterns as in the case of Swahili verbs. In general, no suffixes are attached to French verbal roots, so in many cases the typical open structure of the Bantu syllables is disrupted. The French verbs do not receive a final -a, a general feature of Swahili verbal roots. Verbal derivation can occur, but in most cases the French verbs undergo a reduced inflection, which is restricted to the possibilities offered by prefixing. Thus the French verbs stay semantically stable and are not submitted to processes like transitivity, passivity etc. 8 Morphosyntactically the following constructions occurred: -nominal marker + root (infinitive) (10) kuemballer (to pack) -subject marker + aspect marker + root (present punctual) (11) anacommander (he/she commands) -subject marker + aspect marker + object marker + root (12) tulimudésigner (we showed him) -subject marker + root (subjunctive) (13) tuengrouper (let us organize ourselves) -copula + inflected form (14) iko coupé (it is cut) Finally we can conclude from this that it is the French elements which adapt to the existing Swahili patterns and not the other way round. This is also easy to observe in the cases where French clusters are formed: (15) njo kusema, hatufanye preuve ya militantisme ya conscience professionel' ya sens, n'est pas, de responsabilité de militant. (this means, that we do not manifest evidence of militantism, of professional consciousness, isn't it, of militant responsibility.) The French word group comes into existence in a hybrid form: the most important elements appear in French while the Swahili connectives or particles are preserved. Moreover, French elements appear not to disturb the general pattern of Swahili word order. The Pragmatic Consistency of the Two Mixed Codes Proceeding from the explanation given in more traditional studies for the occurrence of borrowing and codeswitching, namely that it should be the result of an instrumental necessity; we see that the material presented here provides evidence of a completely different perspective in the search for a reason for the use of French elements in Swahili. In the first place, we

< previous page

page_48

next page >

< previous page

page_49

next page > Page 49

distinguished between two mixed codes which we discussed on the basis of the nature of the incidence of the French elements. The French elements which appear in the colloquial variant are incorporated in a rather unconscious way. They are inserted in Swahili in a quasi-spontaneous way. This is the case for terms which are used almost idiomatically, or for terms which are an established part of the lexical register chosen by the speaker for that conversation. This comes down to semantic choices which are motivated by mechanisms like untranslatability, economy or habit. I consider this to be an illustration of the far-reaching influence of French on everyday Swahili language use. The pervasiveness of French in the language of Swahili speakers is a result of daily contact with French. Conversely, the mixing of French and Swahili elements in the variant which is used as register does not have anything to do with motivations such as linguistic economy. Here we see that switches are made in a rather conscious manner. The deliberate introduction of French lexemes is deduced from the fact that risks are apparently undertaken when these foreign elements are inserted. Codeswitches are regularly flagged by hesitations, false starts and corrections. They are often accompanied by 'eh', 'ehe', 'njo kusema' (this means) and other fillers which give the speaker some time to formulate his or her thoughts in the most appropriate way. (16) . . .série ya viti. . .waliviomba eh. . .douzaines mbili tatu (. . .a series of things. . .they asked eh. . .dozens two three.) (17) ah, njo kusema tena ba. . .batakutroubler. (ah, this means again that they will confuse you.) We can also exclude untranslatability as a motivation for switching here. There are of course terms which correspond to this pattern, like hôpital, inspecteur, radio. These usually concern very specialised terms or the names of organisations or associations which have been accepted and standardised (inter)nationally. Force Publique du Congo Belge, wafrères de Notre Dame, province ya Kivu etc. These can be considered as residual terms, from the colonial period. Reasons such as economy can also play a part in this, and then one can speak of partial untranslatability; a Swahili alternative is available but is judged unsuitable as it is a more difficult solution. In most cases, however, French elements make reference to simple everyday happenings. Moreover, we see that in many cases French and Swahili equivalents immediately follow each other in the same texts: (18) finissage/kumaliza kwa vitu. (the finishing of things.) (19) wanaadapter mentalité ya inchi/wanafwata hali ya inchi. (they follow the mentality of the country.) Contrary to the semantic choices which are integrated in a natural, self-evident manner and recur systematically, the intentional, planned choices are characterised by one single ad hoc appearance.

< previous page

page_49

next page >

< previous page

page_50

next page > Page 50

(20) kwa kugarantir eh social, kupata maisha mema (to guarantee eh social, to get a good life.) Example (20) confirms this conclusion: the nominal elliptic form we deal with here has been introduced into the discourse on an ad hoc basis. It only gets its meaning from the explanation which follows, and it is thus a case of intra-discourse convention, not established in the form of a stable loan word. Finally we can conclude that linguistic economy, which is sometimes offered as an explanation for codeswitching or borrowing, can only be accepted as just one of many motivations, and not even a particularly important one. As a consequence, French in Lubumbashi Swahili may be explained by an internal, functional discourse potential, which is responsible for the many ad hoc choices which occur. We illustrate this on the basis of a few examples: (a) Topic marking: The lexical choices that are made in a conversation strongly depend upon the subject and on the arguments that are introduced in that conversation. To gain some understandiing of this matter, we must also take into consideration the way in which the arguments tend to configurate inside the conversation. (21) A: Kwa hapa sasa hatufuate tena authenticité. (These days we do not follow authenticité anymore.) B: Mmm. A: Authenticité tunafanya nini, tunaesha kutupaayo. Sasa unaona, hatuishi muzuri. Unaona, hatufwate tena coutime yetu, hatuishi muzuri. Njo tule tumpanjo alitupaka, ile wakati alitupa, anapata yee ya sasa. Mbon, saa hii alipata, hakupate namna ya kuendelea nayo tena, imufukishe mbele. Unaona? Naweza kukupa mufano, saa shi Baluba eh?, coutime ya kwetu, bwana anaweza kuowa bibi, anamupa coutime ya kwabo, asema bibi yangu asikuwe tena na bwana fulani eh? bwana ni paka ule bwanake moya ule. Bon sasa, hapa sasa, batoto ba sasa habana vile. Kama na banamuowa yee anaanza kusema ho! tunesha kuwa indépendant, mitafwanya vile mitapenda. Unashikia? (What are we doing with authenticité, we have thrown it away. Now we do not live in a decent way. You see, we do not follow our customs anymore, we do not live properly. This also happened when he threw away these mice, the moment he threw them away, he received the things of today. Good, the moment he had this, he could not find a way to get on with it, it first had made him chase away. You see? I can give you a example, like with us the Baluba he?, our tradition, a man can marry a woman, he gives his customs to her, namely my wife, she cannot be with any man anymore he? A husband is only that one husband of her. Good now, today, the children of these days are not like this anymore. When they marry he starts to say ho! we have become independent, I do what I like. You understand?)

< previous page

page_50

next page >

< previous page

page_51

next page > Page 51

B: Mmm. A: Alitupa ile mambo ya zamani hakupende kuifwata hapana. Njo pale kama tunapata ya sasa, na ile ya zamani, tuengrouper pamoja, tutapata njia. Unasikia? Tunabeba authenticité na hii ya sasa tunafunda, tunatumika nayo, tutapata njia kuweza kuishi. (He threw away the things from the past, he did not want to follow them. So when we take the things of today and the things from the past and bring them together, we will find the way again. You understand? We carry authenticité and this what we are learning now, we work with it, we will find a way to live.) In this conversation the French terms authenticité, coutime, indépendant and engrouper appear throughout. These terms function topically in this text, where topic is used in its pre-theoretical sense of 'that which is talked about'. In other words these French elements mark the topic of the discussion and its continuity throughout the sequence. (b) Emphasis and foregrounding: French elements are flagged in many instances with high pitch or repetition. The French elements fulfil an emphasising function. (22) A: Leo hii, bintu binesha kuwa nguvu, ni kweri, ile mama anasema, même au niveau du pays, mu mukini, kule Mobutu chef. Kama tukonateswa leo, ni juu yee alitupa. (These days things have become difficult, it is true what this woman says, even at the level of the country, in the village, there where Mobutu is the chief. When we suffer it is because he has thrown the things away.) B: Alitupa bia zamani. (He has thrown away the things from the past.) A: Alitupa, byote byote. La preuve, anasema non, recour à l'authenticité. Njo, alitafuta ende akalokote. (He has thrown away everything everything. The evidence, he says no, back to 'authenticité'. That is what he looked to pick up again.) The expressions in French have a key-function here. In this part of the conversation the speaker makes sense of the whole discussion by expressing these considerations. The French fulfils a thematic function marking the core of the matter. (23) Quand tu agis comme ça et bien, unaonesha mapendo yako kwa jirani, grand commandement de dieu. (If you act in this way, well, then you show your love for your fellow-man, a big command of god.) This is an argument from a strongly rhetorical text. The speaker tries to persuade his public to think and behave in a certain way. This has to be justified and presented as favourably as possible. Grand commandement de dieu serves here as an authorityargument and receives a special marking by the shift to French.

< previous page

page_51

next page >

< previous page

page_52

next page > Page 52

Many of the French intrusions can be explained as emphasis. The occurrence of French instead of Swahili terms at certain places in the text clearly has to do with informative salience: the terms stand for those elements which carry the most important information in the sentence or the sequence. Because of the clear identifiability of French elements in a Swahili text, a contrast is realised. This contrast is, of course, a factor that stimulates an emphatic function. (c) The use of particles: finally, we also observed that French particles are used in Swahili. They do not appear by accident, but turn out to possess an obvious communicative function. (24) Bon, balianza kuona zamani, donc moi, baba alikuwa kama ana quinze ans, anamuowa mama iko na dix ans, donc batótó sana! (Good, they started to see in the past, so me, father when he was fifteen he married my mother who was ten, so very very young!) Bon is the pragmatic particle we find most often in our material, always preceded by a pause. Bon functions as a turn taker, it marks a new turn in the conversation or introduces a new subject where the preceding one is ended. We can consider the use of donc as a floor holder. The speaker wishes to make his point here and he announces this twice by donc. It signals that he wants to complete his turn. (25) Muke wangu c'est-à-dire bibi yangu atakuja kesho, huku utasema oh dis, bibi yangu atakuya kesho dis (laughs) alors (My wife that is to say my wife will come tomorrow, you should say here oh, I say, my wife will come tomorrow, I say, (laughs) come on.) The man who is talking here is a native speaker of the so-called Swahili Bora who mocks the Swahili as it is spoken in Lubumbashi, from a self-deprecating attitude which is frequently adopted toward this language. We are able to understand c'està-dire from this background information. This is the only time that this is used in French. Normally the Swahili equivalent njo kusema is used for this. It serves to explicitly oppose the two languages, the choice for the French alternative therefore has a contrastive effect. Dis cannot be considered as literally 'say', it stands outside the sentence but is communicationally far from unimportant. It gives a subjective, speaker-oriented evaluation of what has been said and here it is also an expression of a certain strategy. It functions as an attention-catcher, with which the speaker implicitly gives his opinion about this, in his view, odd language. Most of the particles are strongly emphatic when they occur in a Swahili text. The contrast with their environment certainly encourages this emphasis. The pragmatic particles in French express the attitude of the speaker toward what is being said. In this way there is a remarkable relation between information structuring and coherence on the one hand and the use of French on the other.

< previous page

page_52

next page >

< previous page

page_53

next page > Page 53

Finally, the French elements display a range of mainly pragmatic functions. Whenever a speaker wants to stress a certain piece of information, or to express an attitude French is available to expand communicative possibilities. Borrowing Versus Codeswitching The conclusions of the preceding analysis are similar to those of an earlier investigation of Campus Kiswahili, a mixed code spoken at the university of Dar-es-Salaam, where Tanzanian Swahili is mixed abundantly with English elements (Blommaert & Gysels, 1987 and 1990). This is at least the case for the way in which the integration proceeds from the embedded forms toward the matrix language. As is the case in Campus Swahili, we also see here that French items are accepted up to a certain level and that they still preserve their French integrity. In both cases French is used pragmatically and functions in discourse organisation. So far, I have not been explicit in the differentiation of the different language contact phenomena found in the texts. A precise delineation of what exactly borrowing represents and which types of codeswitching can occur are nevertheless of major importance in discussions of motivations and constraints. There has not been much discussion about how to distinguish established borrowings from codeswitching (a spontaneous insertion of multiple words of the embedded language into the matrix language). In this respect Scotton (1987) discusses Poplack's (1988) proposal regarding the basis on which a distinction between borrowing and codeswitching should be made. Poplack labels this the 'Structural Integration-based Definition' by which the distinction between borrowing and codeswitching is primarily drawn on grounds of the extent of morphologic and syntactic integration of the embedded language material in the matrix language. Scotton (1987) presents her own definition, namely the 'Community and Social Function Frequency-based Definition' in which the major difference between borrowing and codeswitching is constituted by 're-occurrence value', or what comes down to frequency. Although the two models generally concur, the difference lies in the classification of nonce forms, 9 which Poplack groups under borrowings, and in the status of borrowing and codeswitching in the matrix language. The results of my analysis of French in Lubumbashi Swahili lead me to take yet another position on the issue of the differentiation of both forms. I have shown that no categorical distinction can be made between borrowing and codeswitching, and as a consequence that neither Poplack's nor Scotton's model is adequate in this respect. Pure structural features or only quantitative arguments do not count as criteria. During the analysis it became clear that code choice is principally determined by discourse organisation, that shifts to the embedded language function to signal new elements in a spoken text or to mark importance or relevance. Bearing this in mind we can, for

< previous page

page_53

next page >

< previous page

page_54

next page > Page 54

instance, consider 'authenticité differently than would be the case in Poplack's or Scotton's models. I also regard it as an established loanword. No alternative term with the same meaning is available in Swahili. It is used everywhere and it has a general level of recognition and acceptance in this speech community. The actual motivation for its occurrence is to be found, however, in a code switch, for the word indicates and highlights the topic of the conversation. I, therefore, would argue for not categorising borrowing and codeswitching a priori but rather for examining the way terms from the embedded language are used in a text. Notes 1. I would like to thank Jan Blommaert for the invaluable comments which he made on this paper, and Robert Pool for correcting my English and making suggestions on style. I am also grateful to Carol Eastman for a careful editorial reading that greatly improved the paper. 2. Due to the length of the texts it is not possible to present them entirely in the context of an article. I therefore give short descriptions of the content where it is necessary, and selected examples as illustrations. 3. Recently Kazadi-Mukenge Kasefu (1985) has written a dissertation entitled 'An Investigation of Code Mixing: Exemplified from Swahili and French'. However, it deals with the structural regulating principles of the mixing of two languages. I do not discuss it here since this is beyond the scope of this paper. 4. Lubumbashi is the administrative capital of Shaba. The present population is estimated at approx. 500,000. For the history and urban geography of the city, see Chapelier (1957), Fetter (1976), Leblanc & Malaisse (1978). 5. I mean by a register a relatively bounded set of ways of speaking including topics, stylistic features and grammatical elements. The complex of available registers of groups or of individuals form a repertoire. 6. In Eastcoast Swahili we can already see this tendency: that class 9-10 is more tolerant vis-à-vis the acceptance of nouns with various meanings as a result of the more flexible behaviour of the concord prefix. This has undergone an even more radical evolution in Shaba Swahili, where this classification prefix takes on a syntactic surface function rather than possessing a semantic meaning. This prefix, which in Shaba Swahili acts rather as a 'cover form', is also capable of getting Swahili words normally belonging to semantically strongly marked classes, to agree with it. 7. We know that there are more of these so-called facultative prefixes which can be added to the French loans, but we have not found any in our texts. We have in mind the prefix bu-that refers to abstracta, the prefix ki- which is used to mark something as diminutive or pejorative, the prefix tu- which marks the plural of the latter and the prefix bi- which also marks the plural of diminutives. 8. In certain cases it is possible that the morpheme -aka is added. It is adopted from other Bantu languages and indicates a habitualis, durative, intensive. We did not find any examples of it in our texts but see for example: alitravaillaka sur les auxiliers 'he used to work on auxiliaries' (Kazadi, 1985). 9. 'Single forms or bound morphemes from an embedded language which are often morphologically and syntactically integrated into the matrix language, but which show little phonological integration and occur with no frequency, possibly only once' (Scotton, 1987).

< previous page

page_54

next page >

< previous page

page_55

next page > Page 55

References Blommaert, J. and Gysels M. (1987) Campus Kiswahili. Language planning, language attitude, and the emergence of a mixed variant among Tanzanian academic staff. Working Papers in Kiswahili 1. Gent: Rijksuniversiteit Gent. (1990) On the functionality of English interferences in Campus Kiswahili. Afrikanistische Arbeitspapiere (AAP) 21, 87-104. Chapelier, A. (1957) Elisabethville Essai de Geographie Urbaine. Brussels: Academie Royale des Sciences Coloniales. De Rooy, V. (1989) Kazi/kazhi: Creatief gebruik van fonologische variatie als middel tot vormgeven en manipuleren van 'situatie' in een Shaba Swahili tekst. Unpublished M.A. thesis, University of Amsterdam. Fabian, J. (1982) Scratching the surface: observations on the poetics of lexical borrowing in Shaba Swahili. Anthropological Linguistics 24, 14-50. (1986a) Simplicity on command: On pidginization of Swahili in Shaba (ZaÏre). In J. A. Fishman, A. Tabouret-Keller, M. Clyne, Bh. Krishnamurti and M. Abdulaziz (eds) The Fergusonian Impact: In Honor of Charles A. Ferguson on the Occasion of his 65th Birthday. Volume 1: From Phonology to Society (pp. 377-85.) Berlin: Mouton de Gruyter. (1986b) Language and Colonial Power: The Appropriation of Swahili in the Former Belgian Congo 1880-1938. Cambridge: Cambridge University Press. Fetter, B. (1976) The Creation of Elisabethville 1910-1940. Stanford: Hoover Institution Press. Heller, M. (1988) Introduction. In M. Heller (ed.) Codeswitching, Anthropological and Socioling-uistic Perspectives (pp. 1-24). Berlin: Mouton de Gruyter. Kabamba Mbikay (1979) Stratigraphie des langues et communications à Lubumbashi. Problèmes Sociaux Zaïros 124-25, 47-74. Kalunga Mwela Ubi (1979) Le lexique du swahili standard face au lexique du swahili de Lubumbashi. Africa (Roma) 34, 42442. Kazadi Mukenge Kasefu (1985) An investigation of codemixing: exemplified from Swahili and French. PhD thesis, University of York. Leblanc, M. and Malaisse (1978) Un Ecosystème Urbain Tropical. Lubumbashi: Université Nationale de Zaire. Nkulu Kabuya (1984) Alternative grammatical analyses of Zaïrean Copperbelt Swahili: some refutations and phrase structure proposals. PhD. thesis, University of York. Polomé, E. C. (1971) The Katanga (Lubumbashi) Swahili creole. In D. Hymes (ed.) Pidginization and Creolization of Languages. Proceedings of a conference held at the University of the West Indies, Mona, Jamaica, April 1968. (pp. 57-9.) Cambridge: Cambridge University Press. Poplack, S. (1988) Language status and language accommodation along a linguistic border. In P. Lowenberg (ed.) Language Spread and Language Policy: Issues, Implications and Case Studies (pp. 90-118). Washington: Georgetown University Press. Scotton, C. M. (1987) Differentiating borrowing and codeswitching. Paper presented at N-WAV 16 (New Ways of Analyzing Variation) Conference. University of Texas, Austin. (1988) Codeswitching as indexical of social negotiations. In M. Heller (ed.) Codeswitching, Anthropological and Sociolinguistic Perspectives (pp. 151-86.) Berlin: Mouton de Gruyter.

< previous page

page_55

next page >

< previous page

page_56

next page > Page 56

< previous page

page_56

next page >

< previous page

page_57

next page > Page 57

Codeswitching and the Exclusivity of Social Identities: Some Data From Campus Kiswahili Jan Blommaert IPrA Research Center, Antwerp, Belgium Abstract This paper investigates the macro-dimension of codeswitching, as observed in Campus Kiswahili (a sociolect of Swahili). Campus Kiswahili is used by academic staff from the University of Dar es Salaam (Tanzania), and it is distinguishable from other English-interfered Swahili variants spoken in the city of Dar es Salaam and elsewhere because of the grammatical and semantic correctness of English interferences. It is argued that the specific sociohistoric background of Tanzanian society accounts for the social valency of Campus Kiswahili. The identity constructed by speakers of Campus Kiswahili is, because of the specific social history, an exclusive identity, i.e. one which is inaccessible for a majority of the population. Conclusions are drawn with regard to existing theoretical models of codeswitching. Introduction 1 In this paper, I will be concerned with the question of what makes social identities so sociali.e. what makes them significant in terms of social relationships larger than that existing between individualsin codeswitching situations. Through the work of Gumperz, Scotton, Poplack, Heller, Woolard and many others, we have come to a fair understanding of the situational, interpersonal or micro dynamics of codeswitching. We know now that codeswitching is not just a type of imperfect speech, but that the appearance of elements from another language in the flow of speech of one language is both linguistically and sociolinguistically consistent. With regard to the sociolinguistic consistency of codeswitching (henceforth CS), a commonly held view associates the two codes with distinct symbolic sets of mutual rights and obligations (e.g. Scotton, 1983). As such, switching from one language to another, or using a mixed code, signals a new (and situation-specific) complex of mutual rights and obligations, and thus becomes a crucial element in the establishment of a role-pattern among

< previous page

page_57

next page >

< previous page

page_58

next page > Page 58

speakers. Mixing codes is therefore always a socially meaningful element of behaviour. I will take this as a given, and attempt to move to a higher level: given the importance of CS at the face to face level as a means of signalling social identities in terms of mutual rights and obligations, then what makes all this social? In other words, how can we understand the macro- or group dimension of an important face to face phenomenon such as CS? For this purpose I will use data from a mixed variant of Swahili, as spoken in Dar es Salaam (Tanzania). During fieldwork in 1985, I started recording the informal in-group speech of academic staff at the University of Dar es Salaam, because I was struck by the amount of English interferences in their Swahili, and by the systematic way in which this mixed variantwhich I dubbed 'Campus Kiswahili'seemed to be associated with certain speech situations (see Blommaert & Gysels, 1987). Further research in 1989 provided arguments for viewing Campus Kiswahili as a distinct sociolect (see Blommaert & Gysels, 1990). 2 Swahili in an Urban Context Swahili is the national language of Tanzania (East Africa). It is spoken as a second language by an overwhelming majority of the population. Ever since independence, and certainly since the federation between Tanganyika and Zanzibar in the United Republic of Tanzania (1964), Swahili has been a strongly symbolic medium. It is associated with values such as nationalism, self-esteem, Africanité and anti-imperialism. English, which used to be the official language during the pre-independence period, was until recently associated with colonialism, capitalist exploitation, and anti-African attitudes (see Akida, 1974; Mulokozi, 1991). Sociolinguistic complexity of a staggering degree is a feature Tanzania shares with most other African states. Yet, by African standards Tanzania is relatively well off, compared to, for example, Nigeria: according to Polomé (1980), well over one hundred languages are spoken, belonging to Bantu, Nilotic, Cushitic, Khoisan, Indo-European and Semitic language families. Obviously, this linguistic fragmentation caused severe communication problems, first to the colonisers, later to the independent government. There was, however, one exceptional asset: continental trade starting from commercial centres such as Zanzibar and going as far inland as East-Central Zaire had already established a lingua franca, Swahili, which was used by people from very different origins and linguistic backgrounds. The Germans, and following World War I the British, capitalised on this asset, and stimulated research into and the teaching of Swahili (Wright, 1965). The main advantage of Swahili, aside from the fait accompli of its spread throughout large portions of East Africa, was the unethnic character of the language. Despite the desperate search for the original 'Waswahili' (e.g. in Abdallah Khalid, 1977), the language was not associated with any ethnic group in particular, and was therefore assumedly not subject to ethnic

< previous page

page_58

next page >

< previous page

page_59

next page > Page 59

sensitivities. The early political movements in Tanganyika (especially Nyerere's Tanganyika African National Union) also adopted Swahili as their working language, and persistently pursued this line till after independence. In present-day Tanzania, primary education is conducted in Swahili, with English as a subject. Secondary and higher education are conducted in English, although the eventual goal of complete Swahilisation remains realistic. Swahili is the first language in the media, and there is a flourishing modern literature in Swahili (e.g. Bertoncini, 1989). In the large multi-ethnic urban centres (Dar es Salaam in particular), Swahili is used for most everyday tasks. The younger generation in these cities are now gradually becoming native speakers, for whom their parents' ethnic language is only a second language. Furthermore, the intense traffic of people from up-country coming into and leaving the city involves a strengthening of the position of Swahili, since most of these people arrive and settle in the large and extremely multi-ethnic slums around the city-centre, where Swahili is the predominant medium of communication. However, local ethnic languages are still used massively in the cities. They constitute the innermost core of the language repertoire of most families in Dar es Salaam, and especially among elderly and poorly schooled people they interfere in their Swahili. From independence (and certainly from 1964 onwards) until well into the 1970s, English was an almost forbidden language. The only social groups who had ample access to English, for professional reasons, were academic and secondary teaching staff, people involved in international politics and trade, and some intellectuals (writers, journalists . . .). Exposure of the masses to English was, in general, severely restricted. As soon as it became clear that the Ujamaa-experiment of massive village-based self-subsistent agriculture had failed, attitudes towards and exposure to English started to change. This can be judged from the barely motivated refusal to implement Swahili as a medium of higher instruction, scheduled for 1975 (Mulokozi, 1986). The need for foreign aid went hand in hand with the renewed acceptability of English in public life. At present, mass exposure to English through movies and (especially) pop music is considerable, though in general restricted to the cities. The consequence of this is that English-interfered Swahili is quite common in Dar es Salaam now. Not only are the traditional groups who had access to Swahili developing forms of codeswitching between English and Swahili, but also both youngsters with a very low level of schooling and children sometimes use English borrowings in their speech. One can roughly distinguish three parameters with regard to the speech repertoire of Tanzanian city-dwellers: (1) Generations: for members of the older generation, Swahili is hardly ever a mother tongue; for the younger generations it often is. Older generations more readily associate English with colonialism, and favour the use of Swahili more than youngsters, for whom English symbolises

< previous page

page_59

next page >

< previous page

page_60

next page > Page 60

modernity, nonconformism and 'Americanness' (wealth, progress, violence, machismo . . .) (2) Level of education: going through secondary and/or higher education automatically creates access to grammatical English and grammatical Swahili. (3) Profession: exposure to parts of the speech repertoire closely follows the requirements of certain types of profession. Factory workers obviously have less access to English than those who pursue a more mobile occupation. These three parameters constitute the conditions under which speech behaviour develops. It will be clear that the speech behaviour of Tanzanian city-dwellers, given the dynamic and variable nature of the three parameters, is a kaleidoscopic reality which is strongly intertwined with the place speakers occupy within the social world. Campus Kiswahili Campus Kiswahili is an English-interfered variant of Swahili spoken in Dar es Salaam. It is used by members of the teaching and research staff of the University of Dar es Salaam. Most members of staff live with their families on Campus, some 20 kilometres outside the city, although a growing number of staff are trying to move to the neighbouring cités of Mwenge, Ubungo and Sinza. Staff members have normally all gone through the Tanzanian education system (in the pre-independence and/or the independence period), and many of them have been abroad for further studies (mostly in Great Britain, the USA, the former German Democratic Republic and the Soviet Union). Since teaching is done in English, all of them have a good command of English, and are fluent in more academic and formal Swahili discourse. The staff have come to Dar es Salaam from all over the country, and they derive from all possible ethnic groups. There is no preponderance of any ethnic group, and as a consequence, Swahili is the normal medium of communication. Teaching is done in English, other academic business in standard Swahili; at home the ethnic language is used among parents and friends. Though children generally know their parents' native language, the dominant medium of communication between parents and children (as among children) is Swahili. Informal talk among staff members typically proceeds in Campus Kiswahili. Campus Kiswahili appears to be nothing special at first sight: just another case of Swahili containing abundant English borrowings and codeswitchings, as in example 1 3: (1) Mimi nina #two points // the first one# iko hivi. . . I I have it is like this (I have two points; the first one is like this . . .) The mechanism of English interferences is pervasive in the informal, ingroup

< previous page

page_60

next page >

< previous page

page_61

next page > Page 61

speech of University staff. Speaking 'pure' Swahili seems to require special attention and effort. When observed, people sometimes catch themselves and others switching to English. Sometimes this automatism gives rise to very funny utterances: (2) Sisi tunasema #practically# Kiswahili sanifu we we speak Swahili pure (we speak practically pure Swahili) Consider this fragment of a conversation held in Campus Kiswahili: 1 A:

Manake / shule hata *chakula hawapati siku hizi (because the schools don't even get standard 2 cha/ chakula #standard# hawapatikani kule // Basi (food nowadays, it can't be obtained there. So 3 imebidi *#wawithdraw# // Wanasema (hi)ooohh nijilimie (it was necessary for them to withdraw. They say Oh! 4 mwenyewe // na kila mwanafunzi wanacholima hakitoshi (I'll work in the field myself. But with all the students farming, even that isn't good enough 5 #(hi)even for *two months# // Basi / mambo (even for two months. So things have 6 yame#deteriorate# // Basi unakuta / ndiyo // (deteriorated. And so there you are, yes) 7 B: Sasa hivi wana/ wanaanza kuamini/ lakini sasa (right now they they start to believe. . . but now 8 haija#pickup# / haija#pick# / ni hali ambayo kwa kweli (it hasn't picked up yet, it hasn't picked yet; it's a situation which really 9 #it is *still deteriorating# // Lakini kwenye (is still deteriorating. But in a 10 #situation# kama hiyo hata kama umepata nafasi ya kusoma (situation like this, even if you get the occasion to study 11 nafikiri # it's just low // they (1o) can't go on# // (I think it's just low, they can't go on) 12 A: Nchi ambayo ina #population# inayoongezeka #by *three (A country that has a population which increases by three 13 percent# / lakini #intake# yake ya #university# (percent, but the intake of its university 14 inapungua #by (hi)*seventy percent# / na kuna wakati (increases by seventy percent, and there was a time 15 imefika hapa watu walikuwa wamechukuliwa mia tano // (when there came five hundred selected people here. 16 #(hi)intake# / ya University / #(hi)*five hundred# // (The intake of the university, five hundred! 17 #That wasn't what it was before# (That wasn't what it was before)

< previous page

page_61

next page >

< previous page

page_62

next page > Page 62

Both speakers use borrowings (B) and code switchings (CS). Borrowings, unlike CS, enter the Swahili flow of speech in a relatively frequent, idiomatic way, as preferred or 'normal' lexical choices. Terms such as 'standard' (2), 'withdraw' (3), 'deteriorate' (6), 'pick up' (8), 'situation' (10), 'population' (12), 'intake' (13,16) and 'University' (13,16) are clearly B, 'Situation' is used in its Swahili original ('hali') in the same fragment (line 8), so B does not always seem to be dictated by untranslatability. CS seems to be related to emphasis: English phrases carry stress in lines 5, 9, 12-13, 14, and 16. In 17, the speaker concludes his exposé with an English sentence. B appear as inflected verbs with Swahili subject- and aspect markers in 3, 6 and 8. On two occasions, the speaker starts off with an English phrase but seems to lose his way in it: in line 8 the correct B 'haijapickup' is incorrectly repeated as 'haijapick'; in line 11 the speaker's correct 'it's just low' is followed by a hardly audible and hesitating 'they can't go on'. This leads us to suspect that speakers insist on switching to English in certain situations, even at the risk of being ungrammatical or getting into other sorts of communicative trouble. In general, English interferences in Campus Kiswahili do not seem to be motivated by speech economy or untranslatability, and neither do they deviate from known linguistic restrictions and rules for codeswitching. Rather, the interferences fulfil a higher level function: that of discourse organisation. It is clear that English elements in a Swahili matrix flow of speech bear a potential for marking importance, newness or relevance in an utterance. This is not surprising, since the contrast between two languages intrinsically allows for contrastive emphasis effects. More relevant chunks of information, such as topic, may therefore be marked by means of B or CS. Discourse organisation is, however, not all there is to Campus Kiswahili. It accounts for structure but not for content. At this point, I can say that most of the code mixing apparent in Campus Kiswahili is stylistically, not linguistically, motivated. Style and Society What is meant by 'stylistically motivated'? Certainly, a narrow, structure-oriented definition of 'style' cannot be used here. Rather I would opt for a view on 'style' as a complex of recurrent patterns of coherence. And 'coherence', in turn, must be read here as being in line with the norms and expectations for thinking, speaking and acting operating in a society at a particular time (much in the sense of Östman, 1986: 185; see also Blommaert, 1990). 'Style' is therefore a primarily social concept, the linguistic side of whichi.e. the concrete forms of expression that make up the observable 'style'are epiphenomenal rather than crucial. When a Tanzanian University staff member uses Campus Kiswahili, he/she has made a stylistic choice. He/she has indicated his/her position,

< previous page

page_62

next page >

< previous page

page_63

next page > Page 63

and that of his/her interlocutor(s) in Tanzanian society. This choice can only happen because of the historical availability of such a style. The material for stylistic choice must have become available in the course of the social history of the larger group to which the interlocutors belong. I see two consequences of this view, one with respect to Campus Kiswahili (or variation within Swahili), one of more general applicability to CS theory: (1) Campus Kiswahili is a different, distinguishable variant of Swahili, not because of its structural peculiarities (which are fairly similar to those of other English-interfered Swahili variants), but because of its specific socio-historical genesis. The set of mutual rights and obligations, activated in face to face usage of Campus Kiswahili, is historically different from that of other mixed variants. An understanding of the pragmatics of Campus Kiswahili must therefore take account of its specific sociohistorical context. The next section will elaborate further on this point. (2) The study of codeswitching, or language mixing in general, is in my view therefore a type of social historiography, in which the object of enquiry is fundamentally historical in nature. Face to face analyses of codeswitching must take into account the fact that there is always a macro-dimension inherent in processes such as CSa dimension of group values. Every group is an historical object, be it a social class ('proletariat', 'white collars', 'politicians' . . .), an ethnic group, or a national group. Groups exist only by virtue of their historical reality, and history is always the primary source of justification for the existence of a group. Identities, by consequence, are never purely situational or dependent on the process of inter-individual negotiation alone. This is suggested in many approaches of CS that start from the assumption of an egalitarian (innate?) communicative competence. This communicative competence provides an intuitive knowledge of norms to the speaker, from which (to use Scotton's, 1988: 155-6 term) a 'grammar of consequences' is derived. This means that speakers are autonomously aware of what effects their communicative choices may engenderthat they 'are free to make any choices, but how their choices will be interpreted is not free' (Scotton, 1988: 155). In practice, this would mean that one could assume any type of social identity, as long as it is situationally appropriate. The error in this view is the implicit assumption that 'situations' are unstructured and absolutely variable, whereas this is not the case. Situations are part of the social world, and are therefore a limited and bounded set. Certain situations will never appear in certain societies, not because of appropriateness criteria, but because they simply do not exist in that society. If situations are not absolutely variable, then the set of available and negotiable roles for behaviour in certain situations is not unlimited either. The choice-potential for any individual speaker may thus seem to be unlimited, but it certainly isn't. Differences in the repertoires

< previous page

page_63

next page >

< previous page

page_64

next page > Page 64

of available choices may be precisely one of the main features of intercultural and inter-societal differentiation. Neglecting the macro-dimension of face to face processes of communication may lead to a trivialisation of some of the necessary macro-assumptions that form the basis of such analyses. I believe that 'communicative competence' is such a trivialised macro-concept. In many analyses, it stands variably for cognition, social norms and conventions, ethics, knowledge of politeness rules and so on. The problem is that all these features may indeed belong to the background knowledge of the speaker, but they can certainly not all be put on a par. Consider the following statement, in which an explanatory theory of CS is proposed: Speakers have tacit knowledge of this indexicality (of social roles) as part of their communicative competence (. . .) They have a natural theory of markedness. The result is that all speakers have mental representations of a matching between code choices and rights and obligations sets. That is, they know that for a particular conventionalized exchange, a certain code choice will be the unmarked realization of an expected rights and obligations set between participants (. . .) While the theory is universal, actual associations are speech community specific, with speakers knowing what code choice is unmarked and which others are marked for exchanges conventionalized in the community. (Scotton, 1988: 152) The question here is, how can speakers have mental representations for phenomena as unstable as (social) 'conventions'? In other words, how can a level of social behaviour which is highly susceptible to diachronic (sometimes revolutionary) change be explained by reference to universal, innate 'natural theories'? I see two possibilities: either Scotton proposes an extremely vague mental capacity, namely the capacity to distinguish between situations in an adaptable way (in which case 'markedness vs. unmarkedness' becomes a trivial notion); or the 'conventions' she presents are a very restricted set of deep-rooted and relatively stable routines of interaction, such as greetings (in which case the theory would hardly be applicable to CS, I presume). In both cases, the solution is an inadequate one. The gap between 'change' and 'choice' is not covered by theories that start from the egalitarian 'communicative competence' concept. The reality is that speakers, for one reason or another, seem to respond easily to sudden changes in social norms, and that they respond collectively (see Fabian, 1971; Ji Feng Yuan et al., 1990 for evidence). Social norms (with roles, possible relationships between speakers, and identities) can never be part of 'communicative competence' in Scotton's (and others') sense. Communicative competence can at best be defined, much in Verschueren's (1987) sense, as the human capacity to adapt to changing conditions for social conduct. In that case, however, the linguist's task is not so much oriented to the human mind, but rather to society.

< previous page

page_64

next page >

< previous page

page_65

next page > Page 65

In an otherwise very stimulating recent paper, Scotton (1990) tries to relate the fact that speech repertoires and their usages are not evenly distributed among members of the speech community to a tactic of boundary maintenance, called 'elite closure'. This tactic: involves institutionalizing the linguistic pattern of the elite, either through official policy or informally established usage norms in order to limit access to socioeconomic mobility and political power to people who possess the requisite linguistic patterns. (1990: 27) 'Elite' is narrowly defined as the bourgeois and petty-bourgeois classes (1990: 28). They prefer to use the former colonial language rather than the local ones because: . . . to use this language in certain situations becomes the unmarked choice if one wants to be recognized as a member of the elite. (1990: 29) This is the point where the markedness model comes in. Its precise relation to the dynamics of elite closure, however, remains to be demonstrated, since elite closure appears to be a conscious, political activity resulting in symbolic boundaries between social classes. Thus, elite closure should generate the 'consequences' or 'rights and obligations sets' which then become part of the mental grammar of consequences basic to the markedness model. Given the unstable nature of power relationships in all types of groups, this mental grammar should be so strongly subject to changes that the rights and obligations sets cannot pertain to anything more stable than customs, conventions or normsall of which are subject to challenge, renegotiation, and change. Consequently, these rights and obligations sets are wholly within the realm of the social, not mental, aspects of language and communication; they cannot be associated with any sensible notion of communicative competence. If the connection between the tactic of elite closure (as conscious social action) and the rights and obligations sets contained in the markedness model (as subconscious, 'natural' features of communication) is correct, then Scotton provides a strong argument against mentalist explanations of CS; one, however, which contradicts her own markedness model. The social meanings speakers convey when they use Campus Kiswahili can only be understood if the social origin and the development of their potential for code choice are taken into account. More specifically, Tanzanian national language policy will prove to be a highly relevant factor in understanding the pragmatics of Campus Kiswahili, as the next section will illustrate.

< previous page

page_65

next page >

< previous page

page_66

next page > Page 66

Identities and the Co-operative Nature of Codeswitching It is a common approach in much of the soiiolinguistics of CS to start from a co-operative maxim of conversation: the interlocutors establish a social relationship among themselves by means of a joint association between code choice and social context (see e.g. Heller, 1988: 3ff.). In other words, CS functions as an indicator of some kind of consensus about mutual rolerelationships between participants in an interactive event. Above, I argued that neglecting the macro-dimension of the face to face dynamics of CS may cause theoretical fallacies, and I think that this remark is equally relevant here. Apart from cooperative functions, CS also has a strongly negative dimension: that of the construction of identities by exclusion. In order to make this clear, I must first make a critical distinction between availability of and access to communicative resources. Both are very often put on a par, yet they are conceptually different. Availability refers to the historical presence of resources, while access indicates the way in which these available resources are distributed among groups and individuals, by means of mechanisms of control such as education, political ideology or law. A resource may thus be available in a society, but not accessible to all members of that society. Here again, one can observe that a strictly egalitarian view on communicative competence can hardly be valid for any society, since not only may a certain resource be unavailable (as argued above), but even if it is available it may not be accessible. One of the characteristics of Campus Kiswahili is the fact that English interferences are, in general, grammatically and semantically well-formed: terms keep their original English meaning and form. The overall grammatical structure of interfered speech displays very few grammatical errors, even in complex structures such as compound verbs: (3) Schule zilikuwa #nationalised# karibu zote schools were almost all (Almost all schools were nationalised) The copula verb 'kuwa' (to be) is correctly inflected, and followed by a well-formed, and semantically coherent, past participle of 'to nationalise'. Similarly, the Swahili used in Campus Kiswahili is largely well-formed in terms of grammar and semantics. Speakers thus blend two qualitatively high variants of languages: 'good' Swahili and 'good' English. Let us now take a look at what happens with English interferences in the speech of younger, poorly educated speakers: (4) Wewe unaji#praudipraudi# you you are showing off (you are boasting, you are showing off) (5) Mambo #fresh# things fresh (I'm alright)

< previous page

page_66

next page >

< previous page

page_67

next page > Page 67

(6) Kupiga #fix# to hit fix (to pull someone's leg) We notice that English lexemes receive metaphorical meanings, and become idiomatic or slang expressions. 'Proud', 'fresh' and 'fix' are not used in their appropriate English semantics, and 'proud' is reduplicated to 'praudipraudi'. Genuine CS is as good as absent from the speech of these youngsters, English interferences being restricted to relexicalised borrowings such as (4), (5) and (6). These seemingly trivial differences indicate a major point: it is the quality of the mixed languages that accounts for the sociolinguistic difference between Campus Kiswahili and other English-interfered variants of Dar es Salaam Swahili. University academic personnel are among the very few groups in Tanzanian society who have ample access to sophisticated English and Swahili. For the slum dwellers, these variants are availablethey are 'there'but inaccessible, partly because of the local education policy which is highly selective, and partly because of the national language policy which used to ban English from public life. The bits of English picked up by the uneducated youngsters therefore have a completely different social valency from the English used by the intellectualsit is not the same 'English', because it has entered their speech through completely different social mechanisms. When a University staff member switches to Campus Kiswahili, this choice automatically excludes all those who have not obtained the same degree of mastery in the two languages. In other words, the identity construction effected through Campus Kiswahili is one which makes use of an exclusive resource, one which excludes 90% of society. The relationships that will be constructed between the speakers, i.e. the micro-identities referred to in many theories of CS, develop within the pre-established macro-identity of 'University staff member'. This prerequisite for the face to face processes is largely anti-social in nature: it isolates the participants from the rest of society by means of the privileged use of a rare commodity. Identity construction, in my view, therefore has two phases: (1) a first phase consisting of drawing a boundary between the participants and the rest of society. This is an anti-social, uncooperative phase; (2) a second phase consisting of the establishment of mutual role relation- ships within the communicative event. This is the social, cooperative phase. Neglecting the first phase of identity construction may result in a distorted picture of the second phase. Code choice is not only guided by a cooperative conversational maxim, but also, paradoxically, by a tendency to isolate the participants through the use of hegemony over part of the repertoire which is accessible to them, but not to others. 4 Note, in passing, that although the data used here relate to an upper-class group and seem to corroborate

< previous page

page_67

next page >

< previous page

page_68

next page > Page 68

Scotton's (1990) findings on 'elite closure', I do not consider this process to be characteristic of 'elites' only. Any social group constructs its identity on the basis of the exclusivity of part of its speech repertoire. There is strong evidence that slang variants contribute to the construction of exclusive identities among impoverished youngsters, women and other non-elite groups (see e.g. Goyvaerts' observations on the Indoubil-variant of Swahili in Bukavu: Goyvaerts, 1988). 'Elite closure' is in my view one specific form of social identity construction, pertaining to the way in which exclusivity is created by one specific social group vis-à-vis the rest of society. It is therefore an ethnographic, rather than an interpretive-sociolinguistic notion. The Social Side of Codeswitching Why is CS such a socially meaningful act? What makes all this social? The micro-dynamics described by Gumperz, Scotton, Poplack, Heller and others are, I believe, but one side of it. The process of CS has always a larger dimensionone of group relations. This dimension has been largely ignored in studies of CS made so far. Very often assumptions are made that are untenable in the light of empirical observation, but that are necessary in order to safeguard the coherence of the theoretical framework. In the paragraphs above, I have emphasised the macro-dimension of CS, as observed in Campus Kiswahili. Campus Kiswahili, in its social relevance, can only be understood when one takes into account the particular social history of the speech community, especially with regard to the exclusivity of access to the resources that are drawn upon in Campus Kiswahili speech. The social identity thus constructed is not a purely situational, one-time phenomenonit is governed and bounded by larger-scale structures that have to do with the historical development of social differences in Tanzanian society. Everything which happens at the level of face to face communication becomes meaningful to the participants only by means of the coherence individual situations show with larger-scale social structures. If we want to understand the dynamics of CS, for instance in relation to language change, pidginisation and creolisation, the sociohistorical dimension will prove to be of paramount importance, as Thomason & Kaufman's (1988) attempt has (partially) demonstrated. Social identity, i.e. identity in the macro-terms outlined above, may become a crucial concept in understanding the dynamic nature of language. Notes 1. The gist of this paper was presented at the Annual Conference on African Linguistics, University of Georgia, April 1990. Thanks are due to Carol Myers-Scotton for numerous remarks on earlier papers on this subject, to Carol Eastman and Leigh Swigart for drawing my attention to some shortcomings in an earlier version of this paper, and to Jim O'Driscoll for stylistic comments. 2. During my stays in Tanzania, assistance and suggestions from Yohani Msanjila, Mugyabuso

< previous page

page_68

next page >

< previous page

page_69

next page > Page 69

Mulokozi, David Massamba, Daniel Mkude, Shaaban Mlacha and Herman Batibo were of great importance. Young Peter Msanjila was particularly helpful in introducing me to the 'tough' speech of young Tanzanians. 3. The following symbols are used: # marks the beginning and the end of the English interference. / short pause // longer pause (sentence boundary) * sentence stress on the word following the asterisk (hi) higher pitch (lo) lower pitch 4. I find similar arguments in Bourdieu (1982) and Mey (1985). Also, the discussion on Gramsci's concept of speech community in Helsloot (1989) is illuminating with regard to the sociohistorical dynamics of language. References Abdallah Khalid (1977) The Liberation of Swahili from European Appropriation. Nairobi: East African Literature Bureau. Akida, Hamis (1974) Language for the coming generation of the scientific age in Tanzania. Kiswahili 44/2: 1-6. Bertoncini, Elena (1989) Outline of Swahili LiteratureProse Fiction and Drama. Leiden: E.J. Brill. Blommaert, Jan (1990) Modern African political style: strategies and genre in Swahili political discourse. Discourse and Society 1/2: 115-31. Blommaert, Jan and Gysels, Marjolein (1987) Campus Kiswahili: Language planning, language attitude, and the emergence of a mixed variant among Tanzanian academic staff. Working Papers in Kiswahili 1. (1990) On the functionality of English interferences in Campus Kiswahili. Afrikanistische Arbeitspapiere (AAP) 21: 87-104. Bourdieu, Pierre (1982) Ce Que Parler Veut Dire. Paris: Fayard. Fabian, Johannes (1971) Jamaa: A Charismatic Movement in Katanga. Evanston: Northwestern University Press. Goyvaerts, Didier (1988) Indoubil: A Swahili hybrid in Bukavu. Language in Society 17, 231-42. Heller, Monica (1988) Introduction. In Monica Heller (ed.), Codeswitching, Anthropological and Sociolinguistic Perspectives (pp. 1-24). Berlin: Mouton de Gruyter. Helsloot, Niels (1989) 'Linguists of all countries . . .' On Gramsci's premise of coherence. Journal of Pragmatics 13, 547-66. Ji Feng Yuan, Kuiper, Koenraad and Shu Shaogu (1990) Language and revolution: Formulae of the Cultural Revolution. Language in Society 19/1, 61-80. Mey, Jacob (1985) Whose Language? A Study in Linguistic Pragmatics. Amsterdam: John Benjamins. Mulokozi, Mugyabuso M. (1986) Kiswahili as a medium of education in Tanzania: Some observations on recent policy shifts. Paper, International symposium on language standardization. Schmitten, March 1986. (1991) English versus Kiswahili in Tanzania's secondary education. In Jan Blommaert (ed.) Swahili Studies: Essays in Honour of Marcel Van Spaandonck (pp. 7-16). Ghent: Academia Press. Östman, Jan-Ola (1986) Pragmatics as Implicitness. Ann Arbor: University Microfilms International. Polomé, Edgar C. (1980) Tanzania, a sociolinguistic perspective. In Edgar Polomé and C.P. Hill (eds) Language in Tanzania (pp. 103-138). London: Oxford University Press. Scotton, Carol Myers (1983) The negotiation of identities in conversation: A theory of markedness and code choice. International Journal of the Sociology of Language 44, 115-36. (1988) Code switching as indexical of social negotiations. In Monica Heller (ed.) Codeswitch-

< previous page

page_69

next page >

< previous page

page_70

next page > Page 70

ing, Anthropological and Sociolinguistic Perspectives (pp. 151-86). Berlin: Mouton de Gruyter. (1990) Elite closure as boundary maintenance: The case of Africa. In Brian Weinstein (ed.) Language Policy and Political Development (pp. 25-42). Norwood: Ablex. Thomason, Sarah Grey and Kaufman, Terrence (1988) Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Verschueren, Jef (1987) Pragmatics as a Theory of Linguistic Adaptation. Antwerp: International Pragmatics Association (IPrA Working Document 1). Wright, Marcia (1965) Swahili language policies 1890-1940. Swahili 35/1, 40-48.

< previous page

page_70

next page >

< previous page

page_71

next page > Page 71

Codeswitching in Bukavu Didier L. Goyvaerts University of Antwerp, UIA, RUCA and University of Brussels, VUB Tembue Zembele I.S.P., Bukavu Abstract Following previous reports, the present paper contains additional information about the multilingual situation in Bukavu. The focus of attention is the phenomenon of codeswitching which emerges as an important characteristic of the overall dynamic picture of linguistic interaction in this multi-ethnic town. Relying on Myers-Scotton's markedness model we discuss in some detail the Bukavu scene and argue that this model may be relied upon as a useful heuristic to obtain comprehensive and rigid interpretations of the data. The present 1 paper constitutes another report about a general sociolinguistic investigation that is still in progress and which concerns the linguistic situation in the town of Bukavu. The ultimate aim of the survey is to end up with what might conveniently be called 'A Socio-Cultural Profile of Bukavu'. It is hoped that a careful description of this situation might lead to significant insights into problems related to multilingual or multiethnic communities in general and Third World cities in particular. Earlier research on this topic was presented in Goyvaerts et al. (1983) and Goyvaerts (1986, 1988, 1991). Almost everywhere in Africa, an important post-independence trend with considerable socio-economic repercussions has been the drift to the towns. This has invariably led to the growth of sprawling, insanitary shanty towns around the main urban centres and to an immense increase in the number of young people for whom jobs are not available because the economy cannot expand fast enough to absorb them. This drift has brought into African cities people of extremely diverse origins and cultures. Bukavu, the capital of the Kivu province, is located on the southern end of Lake Kivu. It is one of the principal towns of Zaire, with a population of 240,000 (in 1985).

< previous page

page_71

next page >

< previous page

page_72

next page > Page 72

Today, the linguistic diversity in Bukavu is immense for over forty languages are spoken within the city boundaries. The degree of multilingualism and multiculturalism is simply staggering. It should be obvious that, given the multiplicity of languages in Bukavu, the resulting multilingualism is bound to serve as an interactional resource for the multilingual speaker. If the opposite situation were to hold then one would have real cause for astonishment. Just as it is common for one language to have many alternate ways of saying 'the same' thing, its logical corollarythe use of a variety of languages to express a single messageshould not come as a surprise either. Therefore, we believe that in multilingual communities the phenomenon of codeswitching should be regarded as the norm rather than the exception. And although the layman may frown upon this state of affairs and regard conversational codeswitching as an instance of socalled 'poor grammar', the linguist should know better. To behave otherwise would lead him or her straight back into the prehistory of linguistics, more specifically to the hey day of unenlightened prescripfivism and prejudice. Typically, however, multilingual societies tend to assign different tasks to different languages. Sometimes the nature of these tasks is fairly obvious as in cases where two members of the same ethnic group conduct their official business in the official language but resort to the use of the tribal language 2 to discuss more trivial matters. Along the same lines one might easily envisage a situation where a government representative encounters members of his own ethnic group in an official context. It wouldn't at all be uncommon for this bureaucrat to make use of the official language to deliver the official message but to add some friendly advice in the tribal language. The former would testify to his impartiality as a civil servant whereas the latter would be an indication of his allegiance, for better or worse, as a fellow tribesman. The above examples are indeed fairly straightforward and can be readily understood from a mere glance at the overall language mosaic that exists in Bukavu (as described, for example, in Goyvaerts et al., 1983). On many occasions, however, examples of codeswitching are far more subtle and less self-evident than the ones mentioned so far: indeed, there's often more involved than meets the eye. Studies on codeswitching, therefore, need to be taken seriously because they are likely to reveal parameters that may be of utmost importance to students of society as well as to all those who, traditionally, have been fascinated by the workings of the human mind. As such, research in the domain of codeswitching would seem to be not only interesting but also significant. To detect universal parameters it is necessary to first elaborate a solid empirical basis since hypotheses are preferably tested against a reliable body of cross-cultural data. The present article should be seen as a modest attempt towards reaching this goal. In it we will discuss some examples of inter-and intra-sentential codeswitching which involve the use of French, Swahili, Shi and Lingala in the city of Bukavu.

< previous page

page_72

next page >

< previous page

page_73

next page > Page 73

To our mind, the clearest and perhaps most insightful account of code-switching presented to date is C. Myers-Scotton's markedness model as elaborated over the past few years in a number of publications (see Myers-Scotton, 1982, 1983, 1986a and b, 1988a and b). It is a model that makes sense to us especially in view of the fact that it constitutes an important step towards isolating significant parameters that are operative in multilingual communities. Our own material would seem to render support to the majority of her findings and, furthermore, yields some indications as to how this framework may be relied upon as a useful heuristic. Before turning our attention to this material it may be instructive to first present in a nutshell what we consider to be the most salient features of Scotton's model: (1) All speakers have a natural theory of the relative markedness of linguistic choices (Scotton's claim that this theory is innate cannot be supported on biological grounds; rather, we believe that it is the capacity to develop such a theory which is innate). (2) Part of the mental entry of any linguistic variant in a given community is its social indexicality (cf. conversational implicatures etc.). (3) As an unmarked medium of communication each linguistic variant is associated with a set of roles and obligations (RO set) which are recalled by the addressee's markedness theory each time a specific variant is used. (4) The use of an unmarked linguistic variant yields reciprocal identification in terms of the participants' RO sets; if the variant is marked it will be interpreted as a dis-identification with the expected RO balance and in this case the model will scan its data base in search of the RO balance for which the variant would be the unmarked choice. The addressee interprets the speaker's negotiation of an RO set accordingly. In this view then, codeswitching can be seen as a means to encode both power and solidarity. The direct outcome of this theory is that it forces one to assign immense importance to the speaker (rather than to the addressee). Why should it indicate superiority for the speaker? It does so because (i) marked choices are calls for a change; any initiation of change is an assertion of the right to initiate and hence of the superior position; (ii) codeswitching illustrates the speaker's multiple identities (each language or variety being the unmarked realisation of a different identity) and (iii) codeswitching may create ambiguity and hence uncertainty on the part of the listener (the reason for switching from the unmarked choice and the RO set it seeks to establish is not always clear). After this succinct recapitulation of the theory and some of its implications we may now turn our attention to the data that will constitute the focal point of reference for our discussion in the remainder of this paper. The examples are representative of a fairly large corpus that was collected in Bukavu during a period of several months of research. (1) Setting: A father and his 16 year-old son at home; the father asks about a particular report the son was supposed to obtain (Swahili-French)

< previous page

page_73

next page >

< previous page

page_74

next page > Page 74

Father: Je? (Well?) Son: Nilimukuta MAIS niliona kama CE N'ÉTAIT PAS POSSIBLE nimuPOSER tena ile QUESTION sababu alikuwa TRÈS OCCUPÉ. (I found him but I noticed it was not possible to ask him that question again because he was very busy.) Father: Sasa utafanya nini? Utarudi tena? (Now what will you do? Will you go back again?) Son: Nirudie? NON, MOI JE SUIS DÉJÀ FATIGUÉ. (Go back? No, I am feeling tired already.) (2) Setting: Two class mates on their way back home; they comment on an exam they took some days earlier (French-Swahili). Student A: JE PENSAIS QUE CE TYPE ALLAIT ME COULER, MAIS QUELLE SURPRISE, MON CHER! ON DIRAIT alidanganyika PARCE QUE sikujua kama atanipatia nusu. (I thought this guy was going to fail me, but what a surprise, my dear! One would say he was absent minded because I didn't know he would give me half of the marks.) Student B: MOIS AUSSI, SURTOUT kwa ile QUESTION ya mwisho. (Me too, especially for that last question.) (3) Setting: Two friends on the street are sharing some confidential information (French-Swahili). A: TU SAIS, QUAND JE TE DISAIS CA, C'ÉTAIT PARCE QUE ule mutu alikuwa pale, na sikupenda ajue bietu. (You know, when I told you that, I didn't want him to know our secrets). B: Kweli? Ndjo mana uliESQUIVER ile QUESTION? SikuREVER ata! (Is that why you avoided that question? I didn't figure it out at all!) (4) Setting: A pupil has come to see the headmaster in order to register (French-Swahili). Pupil: BONJOUR! JE SUIS VENU ME FAIRE INSCRIRE . . . (Hello, I've come to register.) Headmaster: BONJOUR! LES INSCRIPTIONS SONT DÉJÀ CLOTURÉES. IL N'Y A PLUS DE PLACES DISPONIBLES. (Hello! Registration is already over. There are no places available anymore.) Pupil: (pause) S'IL TE PLAIT, FAIS UN EFFORT VRAIMENT. NOUS POUVONS COOPERER kwa juu weye ndjo ESPOIR yangu ya mwisho. (Please, do make an effort. We might co-operate for you are my last hope.) (5) Setting: At home, a group of young people are watching TV (French-Lingala-Swahili; the Lingala utterances appear in italics) A: DIS, omonaki MATCH wana? . . . EN TOUTCAS po na ngai, IL

< previous page

page_74

next page >

< previous page

page_75

next page > Page 75

FALLAIT kaka ARBITRE abimisa JOUEUR wana . . . PARCE QUE TANTOT ni kucheza mubaya, TANTOT ni kuPROVOQUER benzake . . . Kobeta BALLE ango ayebi ata kobeta malamu te. (Hey you! Did you watch that soccer game? Anyway, as far as I am concerned, it was necessary for the referee to send off that player . . . he plays awfully, then he provokes his partners . . . Playing like this; he doesn't even know how to play. B:Ya solo! Ye azalaki na MISSION SPECIALE: koMARQUER SON VIS-À-VIS, C'EST TOUT. IL EST PROFESSIONNEL TU SAIS. (Right! He had a special mission: to closely watch his adversary, that's all. He is a professional you know.) C: Muache kwanza shie tusikie MATCH. Ile COMMENTAIRES mutaifanya nyuma. LE MATCH EST PLUS IMPORTANT QUE CETTE BLAGUE LÀ. (Be quiet, let us first listen to the match. You'd better leave those comments until later. The match is more important than that joke.) D: Bino bozalaka ndenge mini? Si muache fujo! (How are you behaving? Please, stop the noise!) (6) Setting: A landlord and a number of Shi workmen on a building site (Shi-Swahili-French; the Shi utterances appear in italics) Landlord: Kaci * kazi wajizire njo? (What did you do yesterday?) Bricklayer: Rhurhangi mala rhwakol'olukuta l'olulunda rhugali yigal'ololundi-lunda na . . . (We still have to finish the wall on that side; then, cover that part and . . .) Landlord: Basi. Ile ndio kazi ya siku muzima? Mufanye mbiyo tumalize hii nyumba. (Enough! Does that work take up a whole day? Hurry up and finish this house!) Bricklayer: Muhanyi, okele bunyibunyi, cikone* na bwinja. (Sir, it's necessary to work slowly but well.) Landlord: BIEN, OUI MAIS LE TEMPS C'EST DE L'ARGENT. Mie minapenda mumalize hii kazi mbiyo. (Well, yes, but time is money. As for me, I want you to finish this work quickly.) Bricklayer: Ndiyo PATRON. (Yes, Sir.) Code choices get readings of markedness in a given community by virtue of the salient features which in a specific exchange will determine the expected rights and obligations balance between specific participants for that exchange. Encircled specifications in Table 1 below, indicate interpretations that are in need of discussion for one might easily have expected the opposite characterisation here, as we shall see presently. The first piece of information emerging from Table 1 concerns the relative markedness of the specific codeswitching illustrated in the data. One might be inclined to characterise examples (2), (3) and (5) as instances of unmarked choices whereas examples (1), (4) and (6) seem to exemplify marked choices.

< previous page

page_75

next page >

< previous page

page_76

next page > Page 76

Table 1

However, for a number of reasons the 'marked' status of (1) turns out to be somewhat problematic. We shall return to a discussion of (1) below. To determine which language is the matrix language and which the embedded one we rely on well-known criteria. The matrix language is the language that sets the morpheme order or, rather, the overall frame; it is the main language of the speaker (his/her native language or the language he/she knows best); it is the source of all the system morphemes (function morphemes v. content morphemes). The information in Table 1 follows from the straightforward interpretation of those criteria and offers no problems. The next three entries in the table summarise succinctly the kind of switching one encounters in the corpus (i.e. they represent a scanning of the material for inter-sentential, intra-sentential and intra-word switches). Myers-Scotton (1986a) has argued that in addition to inter-sentential and intra-sentential switches, intra-word switches only occur in instances of codeswitching as an unmarked choice (which, typically, encode dual or multiple identities for the speaker). Inter-sentential switching is witnessed, for example, in cases of codeswitching of the sequential unmarked choice type where changes in situational features (new participant, new topic etc.) call for the switch to a new unmarked code without the usual switching back and forth taking place. Intra-sentential switching is typical of non-sequential (i.e. back and forth) switching of the unmarked choice type (for ingroup communication). On the other hand, given the Bukavu corpus, we have little hesitation to claim that codeswitching of the marked type can only be inter-sentential (or perhaps inter-clausal). Switching as a marked choice is switching away from the expected in order to indicate power (switching from the ingroup language; e.g. the switch from Shi to Swahili in (6)) or solidarity (switching to the ingroup language; e.g. the switch from

< previous page

page_76

next page >

< previous page

page_77

next page > Page 77

French to Swahili in (4)). In view of these remarks we must conclude that as far as those three attributes are concerned, examples (1) and (4) are not in line with our expectations and will need to be discussed in more detail below. The final entry in Table 1 concerns the kinds of constraints that are operative in both types of switching. In general terms the Bukavu corpus supports the hypothesis that switching as an unmarked choice shows looser structural constraints than switching of the marked choice type. However, given the six examples that make up the subject matter of the present paper we see that, again, it is examples (1) and (4) that are inconsistent with the general assumptions. Notice in passing that the Bukavu data contains several counter-examples to the free morpheme constraint (a constraint prohibiting intra-word mixing of morphemes from the matrix language and the embedded language) and the equivalence constraint (a constraint prohibiting switches that do not conform to syntactic rules in both languages). When all those variables are compared and evaluated we observe that both examples (1) and (4) correspond to matrices that are somewhat special for they violate some of our hypotheses. Let us first turn to example (4). The striking fact about example (4) is that the pupil's second reply contains intra-sentential switching as well as an instance of structural loose-ness. However, when we study this example more carefully we see that it is far more rigid than we originally thought. The phrase 'nous pouvons coopérer' is highly ambiguous for the word coopérer in students' jargon always refers to something illegal. In this case, the student is suggesting to the headmaster the possibility of a bribe before switching to Swahili as a means to appeal for solidarity. The marked choice of his reply becomes thus even more conspicuous. Furthermore, the use of the French word espoir in the Swahili sequence should not be interpreted as an instance of shifting back to French for the simple reason that it constitutes a borrowing and not a switch. Its frequency of occurrence in our corpus testifies to such an interpretation: the Swahili word for 'hope' viz. kitumaini is hardly ever used in Bukavu. Other instances of borrowing to be found in our data are question (in examples (1), (2) and (3)) and match (in example (5)). Also, observe that the student's reply shows a distinct shift of style for he starts using the pronoun of solidarity 'tu' instead of the expected, more formal, 'vous'. It seems as if the use of tu in a very natural way leads up to the somewhat dubious suggestion to 'co-operate' (sic!) and the marked switching at the end of his reply. The overall interpretation of speech event (4) is therefore fully in line with what a markedness theory would lead us to expect. An even more interesting case than the previous one is the speech event illustrated in (1). Our first inclination was to look upon this exchange as an instance of switching as a marked choice. Note that the unmarked choice can be identified empirically for it is always the most frequent choice in a given exchange. Therefore, relying on our corpus and bearing in mind the typical role patterns in the average African household in Bukavu (where the

< previous page

page_77

next page >

< previous page

page_78

next page > Page 78

hierarchical relationship involving grandparents v. parents v. adolescents v. younger children is rather strict) we almost automatically labelled this switching as being of the marked choice type. Some of the specifications in matrix (1), however, were not at all the ones we expected to occur given the alleged rigidness of such marked switches. It is these specifications that forced us to return to our sample for a closer investigation. Now, while it is true that in Bukavu one expects children to show a measure of respect towards their parents and that statistically speaking (i.e. given the overall corpus at our disposal) the son's language use could be interpreted as marked, we found out that for a particular age group such a code is not at all unusual in this context. It is common for young male adolescents to use this kind of language when talking to their fathers (not when addressing their grandfathers though). On the other hand it is not the code one expects to be used by girls in a close family context. In our present case, the son's language is indeed an instance of switching as an unmarked choice and should as such be added to the 'markedness dictionary' which the researcher must try to set up for the multilingual community he wishes to investigate. Given this knowledge, the sequence under discussion is clearly indicative of the close affection between father and son. The son approaches his father as if he were his peer, someone for whom no institutionalised barriers exist. The father on the other hand sticks to his role set and does not diverge from the use of Swahili as his medium of communication which is not at all surprising either. However, the next step in the exchange becomes very significant for the son first repeats the father's question in Swahili ('Nirudie?') and then switches entirely to French which is definitely a marked choice. The use of French at this point in the speech event is the linguistic equivalent of widening the gap between himself and his interlocutor. In this way he (the speaker) places himself in a more powerful position vis-à-vis his father (the addressee). Furthermore, whereas the first exchange concerned an issue that was of minor importance, the fact 'that he might have to return to school' is obviously not a trivial matter. It requires him to negotiate for power to replace the existing unmarked rights and obligations set with another one. We see then that example (1) actually consists of two parts which, independently, comply with the given specifications as is borne out by the representation of the relevant facts, as in Table 2. Having discussed examples (1) and (4) in considerable detail, we may now turn to some general observations regarding the Bukavu data. Given Myers-Scotton's markedness model a host of intriguing factstypical of multilingual settingsreceive a principled explanation and an unambiguous interpretation. An important finding in this respect is that in any conversation codeswitching as a marked choice is a strategy which negotiates power for the speaker; it signals the speaker's intention to change the social relationship with the addressee, i.e. to change the RO balance. In example (4) we observe that the speaker switches from French to Swahili as a marked choice to negotiate less social distance than is normative in this situation

< previous page

page_78

next page >

< previous page

page_79

next page > Page 79

Table 2 Speech Event Choice Matrix Lg. Embedded Lg. Inter-sentential Intra-sentential Intra-word LSC

(1) U M Swa Swa Fr Fr No Yes Yes No Yes No Yes No

whereas in example (6) the speaker switches from Shi to Swahili to negotiate greater social distance. Myers-Scotton (1986a) correctly observes that when switching occurs participants who already have relatively high power do most of the switching. In connection with this she asks: 'Does shifting make speakers powerful or does it only characterize powerful persons?' We think the Bukavu data favours at least a positive answer to the former question as would seem to be borne out by example (1). In the latter, the son's marked switch gives him more status. The remaining examples are all of the unmarked choice type. Speech events (2), (3) and (5) constitute conventionalised exchanges in which shifting itself is the expected, unmarked, choice for all participants. One could view this particular instance of codeswitching as a solidarity promoting feature, i.e. as a form of accommodation. Also, this kind of codeswitching is typical for some interactions among peers who use shifting to signal their mutually multiple identities. The speakers are boasting, symbolically as it were, of the range of their identities. One should not forget that in Bukavu, where the largest part of the population is to be situated within the 10-25 age group, youngsters in the shanty town lack any financial or material means to procure status symbols. However, status can be obtained through language use and at no cost at all for that matter! It seems to us there is no other explanation for the existence, for example in Indoubil (a Swahili hybrid used by Bukavu youngsters), of words taken from English and Italian. Furthermore, this kind of unmarked switching is predicted by the very nature of Bukavu's multilingual community: the potentially explosive inter-group competition (epitomised by the scores of tribal languages within the city boundaries) is neutralised by the existing lingua francas which in Bukavu are not indexical of ethnic rivalries. Societies might differ considerably with respect to which choices are regarded as marked and unmarked. Myers-Scotton (1986a) mentions 15 hours of recording in a rural bar in Kenya during which period hardly any codeswitching occurred. The use of a single variety in this context is the

< previous page

page_79

next page >

< previous page

page_80

next page > Page 80

unmarked choice. Here, not shifting could be viewed as a form of addressee-oriented accommodation. In Bukavu, however, one would expect the exact opposite in such a context (i.e. in bars, market places, taxis etc.): not shifting would be a marked choice. We believe this state of affairs to be a function of the degree of linguistic heterogeneity that pervades all layers of society in Bukavu. Codeswitching of the unmarked type further contributes to maintaining the ethnic balance. And, finally, we certainly do not exclude the straightforward neurological interpretation of unmarked codeswitching which, in neurophysiological terms, may be a function of the very make up of the synapses that link up specific neurons in the human brain (Goyvaerts, 1973). Resorting to common parlance, one might therefore wish to think of marked choices as representative of a speaker's quasi conscious attempt to show that he or she is different from or similar to the addressee. Codeswitching as an unmarked choice, on the other hand, could be regarded as indicative of the fact that whether or not the speaker finds the 'right' word in his or her matrix language is contextually irrelevant as long as some (as yet statistically undefined degree of) switching takes place. In one of the previous paragraphs we mentioned the notion of a 'markedness dictionary'. It may perhaps be useful to elaborate this idea a little bit further although, we must admit, it is somewhat beyond the scope of the present paper (cf. Guyvaerts, 1989). Ideally, we see this dictionary as an exhaustive listing of unmarked choices. Each entry in the dictionary would consist of a mapping of contexts and codes in terms of variables such as age group, ethnic affiliation, socio-economic class, sex, status, aspirations, family make up (extended, nuclear), superiors, strangers, peers, ingroup members, etc. Anybody who wants to become fully integrated in a multilingual community needs to develop such a dictionary. It goes without saying that given the dynamic nature of human society in general and multilingual societies in particular, this dictionary is under constant revision from the speakers. There was a time, for example, when Indoubilthe Swahili hybrid or jargon used by Bukavu youngsterswas frowned upon by the greater part of the population. Today, however, Indoubil has become slightly more respectable as the lingua franca of the younger generation due, undoubtedly, to the fact that many of its users are succeeding brilliantly in the underground economy. The tangible result is that Indoubil is already being used in codeswitching of the exploratory choice type! On a more mundane level of argumentation it might be tempting to claim that this provides additional evidence for the somewhat arrogant position according to which all human behaviour can be related to specific points on a purely economic scale of values. A similar observation about changing dictionary entries could be made with respect to the emergence and remarkable growth of Lingala in Bukavu (cf. Goyvaerts & Kabemba, forthcoming). By way of conclusion let us briefly recapitulate the essence of the present paper. Relying on Myers-Scotton's markedness model we have managed to obtain a deeper insight into the nature of codeswitching in Bukavu. In doing

< previous page

page_80

next page >

< previous page

page_81

next page > Page 81

so we have been able to further elaborateand give substance tothe overall picture of this multilingual town. Our particular approach has been dictated by a number of unique specifications. The heuristic value of these specifications has become obvious for it is these stringent criteria that forced us to question our preliminary conclusions and to study our corpus as well as the overall Bukavu scene in much greater detail. This has ultimately led to a more comprehensive and, we believe, essentially correct interpretation of the data. It may be intriguing to speculate on the extent to which these rigid variables can be used as criterial attributes for the characterisation of code choices. We hope further cross-cultural research will shed some light on this issue. Notes 1. The authors wish to express their gratitude to the Belgian National Science Foundation (NFWO) and the Belgian Ministry of Foreign Affairs without whose financial support this paper could not have been written. Furthermore, D. Goyvaerts should like to dedicate this article to Yolande De Royher shrewd observation on April 13th 1989 in some Nairobi hotel prompted him to take up the present research topic. 2. Although this happens to be a fairly common notion we should nevertheless like to make it clear that we are not at all happy with it. Indeed, the word tribe would seem to evoke fixity and atavism and, hence, the term ethnic group should be preferred as it allows for greater conceptual flexibility. Ethnicity is an essentially modern phenomenon, primarily rooted in urban settings and intimately tied with the processes of change introduced by economic and political modernisation. (cf. Goyvaerts & Kabemba, forthcoming). References Goyvaerts, D. (1973) Learning and some related aspects concerning psycholinguistic research. Linguistics 97, 5-45. (1986) Language and History in Central Africa. Antwerp: APIL 44. (1988) Indoubil: A Swahili hybrid in Bukavu. Language in Society 17, 231-42. (1989) Review of H. Haarmann: Language in Ethnicity. Language in Society 19, 414-19. (1991) Kibalele: form and function of a secret language in Bukavu. (to appear) Goyvaerts, D., Naeyaert, D. and Semikenke, M. (1983) Language and education policy in the multilingual city of Bukavu. Journal of Multilingual and Multicultural Development 4, 47-62. Goyvaerts, D. and Kabemba, A. (1992) Bukavu: Profil Socio-culturel d'une Ville en Afrique Centrale. (in preparation). Myers-Scotton, C. (1982) The possibility of switching. Anthropological Linguistics 24, 432-44. (1983) The negotiation of identities in conversation. International Journal of the Sociology of Language 44, 115-36. (1986a) Style-shifting and language-shifting as powerful language strategies. (Paper presented at the 1986 Minnesota Linguistics Conference, September 12-13.) (1986b) Code-switching as an unmarked choice and morpheme constraints. N-WAV Conference XV. (1988a) Code switching and types of multilingual communities. GURT'87, 61-82. (1988b) Differentiating borrowing and codeswitching. N-WAV Conference XVI, 318-25.

< previous page

page_81

next page >

< previous page

page_82

next page > Page 82

< previous page

page_82

next page >

< previous page

page_83

next page > Page 83

Two Codes or One? The Insiders' View and the Description of Codeswitching in Dakar Leigh Swigart Department of Anthropology DH-05, University of Washington, Seattle WA 98195, USA Abstract In describing the different types of Wolof/French codeswitching used in Dakar, the author questions the assumption found in many codeswitching studies that the use of two languages within a single conversation violates a norm and must therefore be explained. In Dakar, the use of 'purposive' codeswitching by bilinguals is relatively uncommon, as is the use of Wolof and French in their respectively pure forms. It is instead a fluid and unmarked switching between Wolof and French, 'Urban Wolof,' that has become the most common mode of speech among urban bilinguals. It is suggested that Urban Wolof be considered a third and separate code rather than the alternate use of two languages, and furthermore that the sociolinguistic models currently available in the codeswitching literature are not appropriate for the analysis of this bilingual phenomenon. Two aspects of Urban Wolof, central to an understanding of its sociolinguistic role, but surrounded by paradox and ambivalence, are then presented: 1) its status as a distinct 'language' and 2) the attitudes surrunding its use. In both cases, it is shown that a consideration of the insiders' view, that of Urban Wolof speakers themselves, provides a rich picture of the position of Urban Wolof and helps to predict its future role in the urban Senegalese community. The relevance of these findings on Urban Wolof to other urban codes spoken around the world is indicated throughout the paper. Introduction Although bilingualism is found in almost every country and estimates indicate that at least half of the world's population speaks more than one language (Romaine, 1989), approaches to the study of bilingualism have often implied that the knowledge and use of a single language is instead the norm. This view is reflected in the overwhelming number of terms, categories, definitions and models put forth by scholars, themselves often hailing

< previous page

page_83

next page >

< previous page

page_84

next page > Page 84

from monolingual societies, in their attempt to 'make sense of' how speakers go about their business using more than one language. Among the various kinds of contact phenomena, codeswitching has perhaps received the greatest amount of attention since it 'violates a strong expectation that only one language will be used at any given time' (Heller, 1988:1). But the question must be asked, whose expectation is this? It is not necessarily that of the many bilinguals to whom codeswitching is so natural that they fail to notice when they themselves practise it. Far from violating norms of linguistic behaviour, codeswitching can be the norm itself, the only appropriate mode of speech in many common situations. In this paper, I will describe a situation where the use of two languages within a single conversation has assumed this type of normative role in the communicative behaviour of bilinguals. In Dakar, the capital city of Senegal, two languagesthe indigenous lingua franca, Wolof, and the official language, Frenchare used together more often in a fluid and unmarked kind of codeswitching than separately in their respective 'pure' forms. The result is a distinct mode of speech that might be best described as a third code, occupying a central position in the speech repertoire of urban bilinguals. This 'Urban Wolof' has its own rules of appropriate usage, that is to say a 'sociolinguistic grammar', just like its constituent languages. However, it does not enjoy the same status with its speakers as those popularly recognised and sanctioned codes. This paper has two aims. The first is to describe a type of codeswitching that, although documented in a number of multilingual settings, has been largely relegated to the sidelines in explanatory models of codeswitching. This kind of codeswitching cannot, I will argue, be fruitfully analysed using most available models since it is essentially different from the bilingual patterns most commonly described in the codeswitching literature. Although my description will focus on this type of speech as observed in Dakar, the few details provided by scholars working in North America, Asia and Africa suggest that certain features of Urban Wolof typify bilingual behaviour in other urban settings as well. In fact, a number of parallel codes seem to have developed in cities where an educated elite speaks a European language, usually left over from the colonial period and now recognised officially, along with an indigenous language (Scotton, 1988). The second aim of this paper is to suggest some characteristics of this fluid type of codeswitching that are intimately tied to its modus operandi and which are consequently central for understanding how it functions sociolinguistically. These characteristics are its status as a 'language' separate from French or Wolof, or a third code, and the attitudes with which it is commonly viewed by its users. In both domains, considerable ambivalence and even contradiction are found. Urban Wolof is most commonly seen by its speakers as a variety of Wolof, although a debased and generally unacceptable one. But despite these overt criticisms, a consideration of the more deep-seated attitudes of the actual users of Urban Wolof, what I call here

< previous page

page_84

next page >

< previous page

page_85

next page > Page 85

the 'insiders' view', reveals that it occupies a separate and important place in the linguistic repertoire of its users. I will attempt to show that integrating the insiders' view into the description of this urban speech form provides insights that other approaches, because developed to describe a very different type of codeswitching, cannot give. The result is a picture of the complex and central role of Urban Wolof in the everyday communicative lives of its speakers as well as an indication of what its future place might be in the linguistic repertoire of urban Senegalese. The data on which the present paper is based was collected over a 27 month period between 1986 and 1989. 1 The speech community that I studied was comprised of multilinguals conversant in at least French and Wolof. Members of non-Wolof ethnicity sometimes spoke other Senegalese languages as well. I collected data using a combination of participant observation, audio recording, and interviews, using both Wolof and French to make and maintain contact with subjects. The first two activities, where I noted down and recorded the speech patterns of dozens of bilinguals, served as the basis for the development of an interview questionnaire and the selection of subjects. The interviews consisted of two parts, the first being a more standard questioning of subjects on their own linguistic patterns as well as attitudes toward the different codes used in Dakar. In the second part, subjects were asked to listen to and comment on a series of extracts of recorded speech. This interview technique was designed so that the formal and conscious linguistic attitudes offered by subjects in the first part might be compared with the more spontaneous ones prompted by hearing actual speech. I intentionally limited the number of subjects to 35 since my aim was to explore in depth their linguistic habits and views.2 The subjects may be broken down by the following characteristics which, through my earlier observation and recording, I found to influence language use: Gender:

22 men; 13 women

Native 13 Wolof; 4 French; 4 Diola; 3 Peul/Toucouleur; 3 Serere; 2 Soose; 2 language: Portuguese Creole; 2 Lebanese Arabic; 1 Soninke; 1 Sousou. Ethnic group:

12 Wolof; 5 Diola; 7 Peul; 3 Serere; 2 Soose; 2 Cape Verdian; 2 Lebanese; 1 Soninke; 1 Sousou

Age:

24 in the 20-35 category; 8 in the 36-50 category; 3 in the 51 + category.

Educational7 primary; 12 secondary; 2 technical training; 4 baccalau-réat the exam level: marking the successful completion of secondary studies); 10 (university. Place of origin:

12 Dakar; 17 another city, town, or country; 5 rural villages.

The underrepresentation of women and older people in this group was unavoidable but perhaps significant in itself. Older women are much less

< previous page

page_85

next page >

< previous page

page_86

next page > Page 86

likely to speak French than younger women due to a lack of educational and employment opportunities during their youth. Older men, on the other hand, are more often conversant in French (even if it has just been 'picked up on the job'), but it was somewhat awkward for me as a young woman to contact and interview them. 3 As for women interviewees, they were hard to interview for the reason that they were, in general, much less willing to be questioned than most men. When they did consent, their responses were usually more reticient. Interviews with women were also often shorter or interrupted since they almost always took place at the subject's home where children, visitors, and chores proved distracting. I tried to supplement my data by questioning bilingual women and older speakers in less formal circumstances whenever the opportunity arose. Codeswitching in Dakar Dakar is a city of a million and a half inhabitants located on the Cap Vert peninsula, the piece of land jutting out into the Atlantic Ocean to form the western-most point of Africa. As the capital of Senegal, it is the city where the official language is most often used. Yet despite the necessity of a knowledge of French for interaction in certain domains of activity, it is recognised that a knowledge of Wolof, the national lingua franca, is indispensable for the conduct of most activities in Dakar. This can be seen by the fact that while there are many Dakarois who do not speak French, those who cannot communicate in Wolof are now few. If it is not learned in the family as a first language, it is learned in the street since it is vital for everyday communication. My interviewees confirmed this assessment by almost unanimously agreeing that Wolof was the first language to be learned by someone migrating to Dakar and by conferring upon the offical language a place of secondary importance. Among those Dakarois who have a command of both French and Wolof, codeswitching has become the most frequently used mode of communication. This codeswitching may be divided into two separate categories according to the manner in which it occurs. The first and least used category, but that which is most often described in the literature, is the goal-oriented codeswitching that I will here call 'purposive codeswitching'. The second category can be best described not as the alternate use of two different languages but rather as a separate code itself. I will call this bilingual phenomenon an urban code or, in referring directly to my own research, Urban Wolof. Purposive Codeswitching Although recently scholarship has paid more attention to how codewitching can operate as a normative code in multilingual communities, the most frequently used approaches to codeswitching remain those which attempt

< previous page

page_86

next page >

< previous page

page_87

next page > Page 87

to find the underlying motivation for the shift from one language to another in the course of a speech interaction. The assumption is that speakers will 'normally' use a single language so that the use of two (or more) is an abnormal occurrence which must somehow be explained. As a recent example of such a view, Heller's work investigates the sociopolitical dimension of codeswitching, focusing particularly on the way that boundaries between language groups influence bilingual behaviour. Central to her interpretations is the idea that codeswitching 'can be approached as a function of what it accomplishes' (1988:10). Myers-Scotton, in a similar manner, sees most switches as an attempt to negotiate a social situation (Scotton, 1988). She views the meaning of switches as arising from their status as either a conventional (unmarked) or unconventional (marked) choice for the speech context in which they occur. A speaker may wish to comply with the rights and obligations obtaining between herself and her interlocutor by choosing the unmarked code during a speech exchange, or she may attempt to establish a new 'RO set' by choosing a marked one. The switches described by Heller and Myers-Scotton are generally ones for which a motivation can be specified, hence my term 'purposive'. In this type of codeswitching, which occurs most commonly at the intersentential level (Poplack, 1979; Scotton, 1988), a speaker may use the unexpected code for 'strategic' effect or, in Myers-Scotton's model, maintain the status quo in an equally intentional way by choosing the expected one. Purposive switching is dependent on the larger social context as well as the face to face one between interlocutors, since it is through the symbolic associations of various languages that the use of one or another carries different meanings. Heller (1988) has pointed out that some separation of domains of language use in a community is necessary for codeswitching to be considered a significant act. If there were no such separation, the fact that two languages were used within a single speech interaction would go unnoticed and would thus convey no message beyond the referential content. It is for this reason that purposive switching, particularly of the 'marked' variety, occurs in Dakar only in those situations, usually formal, where the use of a particular language (indigenous or official) is heavily sanctioned. The switch to an unexpected language thus takes on special connotations, and this is what the speaker manipulates in order to create a desired situation or meaning. Such switches may be heard, for example, in the context of political addresses where candidates attempt in the course of a single speech to appeal to both intellectuals and illiterates by alternating between French and Wolof. A single language, whether the official or an indigenous language carefully expunged of all loanwords, is usually considered appropriate for such a formal occasion. A vendor or customer in an outdoor market, normally an indigenous language setting, may also switch into French in the course of a transaction in order to negotiate with more authority the desired price for an item. A lycée director reported that he sometimes used Wolof in the classroom, a French-specific domain, as a marked code to

< previous page

page_87

next page >

< previous page

page_88

next page > Page 88

appeal to students on special topics such as discipline problems, or to capture the students' attention through humour. In Dakar, however, the domains in which a specific language is called for are few, and in most situations the use of either French or Wolof (assuming the interlocutors are bilinguals) is acceptable. 'Pure language' situations, according to older Dakarois, are decreasing in number, and it is the official language in particular which has lost many of what were once its exclusive domains. For instance, a lapse from the French language was considered a punishable offence anywhere on the premises of Dakar schools even as recently as 20 years ago, but now one hears both students and teachers conversing freely in Wolof (or Urban Wolof) unless engaged directly in classroom activities. At the same time, the situations that were once 'Wolofspecific', except for certain formal ceremonial or religious ones, have now become more characterised by the use of a Frenchinfluenced form of speech. The result of these shifts is that the use of codeswitching of the marked variety is not nearly as common as Urban Wolof, the type of fluid code-switching that Poplack has called an 'overall discourse mode' (1979:69) and that Myers-Scotton refers to as 'overall switching as the unmarked choice' (1988). My data thus suggest that the greater part of codeswitching in Dakar cannot be consistently explained in terms of a deliberate choice or the strategic negotiations of expected or unexpected codes. Although it cannot be denied that some instances of switching may be interpreted as purposive, most fall instead into the category of what have been called 'refractory cases of switching', or those that cannot be understood solely by analysing the immediate motivations or goals of the speaker (Pandit, 1986:64). This is not to contend that such switching is meaningless, but rather that each switch is not necessarily significant. As Myers-Scotton notes in her characterisation of switching as the unmarked choice 'it is the overall pattern of using two varieties which carries social meaning' (Scotton, 1988:162). Urban Wolof Urban Wolof, rapidly becoming the most widely spoken code in Dakar, is a type of Wolof-French codeswitching that generally takes place at the intrasentential (clausal, lexical, or morphemic) level, although it may also occur at the intersentential level. In this switching, Wolof is generally the 'matrix language' (Myers-Scotton, 1990) or the language to which the majority of structural features can be attributed. This codeswitching occurs without any pauses, meta-commentary, or hesitation phenomena that indicate that it is unusual or unacceptable to use Wolof and French in the same utterance (see Poplack, 1988 on Spanish/English codeswitching in the New York Puerto Rican community). The following is an example, drawn from a discussion between two young men about the sighting of the new moon that announces the end of the Muslim fasting month. (Here and throughout

< previous page

page_88

next page >

< previous page

page_89

next page > Page 89

this paper, Wolof words are shown in bold italic script and French words in plain italic script. Words of foreign origin that have been phonetically assimilated into the Wolof system are shown in 'bold script' between quotation marks. In the cases where a direct translation into English is given, the bold/plain distinction is retained for clarity. All Wolof words are written using the transcription system adopted by linguists at the Centre de Linguistique Appliquée de Dakar.): (1) . . . xam nga weeru benn jour, quelques minutes lay der, quelques minutes rekk et puis c'est petit, un tout petit kii la! Boo gaawul, doo ko mën a gis.

(. . .you know the first day's moon, it's only there for a couple of minutes, just a couple of minutes and then it's small, it's a really small thing! If you're not quick, you won't be able to seeit.)

This type of speech occurs as an 'in-group' phenomenon among educated urban dwellers, and it has become so frequent in African capital cities that Myers-Scotton has dubbed such mixed forms 'hybrids' or 'innovating' varieties. The use of two languages in this manner encodes a 'dual identity', showing that the speaker is both an African and a member of an educated elite (Scotton, 1988). It can be distinguished in several ways from Wolof into which many French loanwords have been inserted. First, loanwords may be used by all speakers, even those who do not have a command over the French language. Second, loans are often, although by no means always, used to fill lexical gaps in the borrowing language (as in the Wolof loan 'welo' from vélo 'bike') and this usage is generally consistent (Scotton, 1988). 4 Urban Wolof, on the other hand, is available only to relatively balanced bilinguals. The items or phrases drawn from French co-exist with Wolof equivalents and thus form, to use Sridhar's (1978) phrase in describing Kannada/English codeswitching, 'an additional lexical stratum'. The choice of a French item over a Wolof one is thus less predictable than in the case of loans, the result being that equivalents from both languages may occur alongside one another in the course of a single utterance, as prison and kaso do in the following example: (2) Journalistes yu journal boobu ñoo (The journalists from that newspaper went demoon def ay kii, quoi, enquête la, je to do, well, it was an enquiry, I don't know, ne sais pas, ci prison i Sénégal. Mais, on Senegalese prisons. Well, if you saw, if soo gisee, soo ko liree, dangay bëgg you read it, you would just feel like crying. jooy sax. Dangay wax ne jamais, jamais You would say that never, never will I do duma def loo xamene di nañu ma jàpp, something that they will get me for and take yobu ma ci Kaso bi. me to prison.) An essential characteristic of Urban Wolof is that it functions as the

< previous page

page_89

next page >

< previous page

page_90

next page > Page 90

unmarked or expected code in many urban speech contexts, primarily those of an informal nature. The manner in which Wolof and French items are intermingled is not calculated for strategic effect but is rather, as Dakarois explain the phenomenon, 'easier', 'faster', 'more convenient', 'less pedantic', 'almost instinctive', 'just a way speaking'. As such, Urban Wolof is often used by its speakers with very little conscious awareness and may even be used in the very sentence where its use is denied, as in the following phrase produced by a young Dakar woman: (3) Mën naa làkk olof sans làkk 'fararise'. 5 (I can speak Wolof without speaking French.) It is important to note that, due both to the frequency of codeswitching by bilinguals and a heavy borrowing from French by Wolof monolinguals, Wolof spoken without a trace of French influence is rarely heard in Dakar. It is, in fact, the use of pure Wolof that stands out, and Dakarois will call attention to it by remarking that a certain speaker uses Wolof bu set 'clean Wolof' or (borrowing from French) Wolof 'piir' (pur). Urban Wolof is thus considered a more normal or unremarkable choice than traditional Wolof. Paradoxically and, as we shall see below, significantly, this 'normality' is rarely openly recognised or admitted. Mention must be made here concerning the manner in which codeswitching may fulfil various conversational or discourse functions. Gumperz (1982) details the way in which switching from one language to another may highlight or otherwise set off passages of conversation during reported speech, message qualification, interjection, reiteration, etc. Despite the fact that such switches do have a specific motivation, I have not included them under the category of purposive codeswitching. This is because the shift from one language to another for discourse effect does not necessarily occur as part of a goal-oriented switching but may also characterise the unmarked urban switching style described above. In other words, a discourse effect may be achieved even without reference to the larger symbolic or connotative features of the language being used. Thus, Heredia-Deprez (personal communication) found that bilingual Maghrebian families in Paris switched to either French or Arabic to discipline their children, depending on the language in use at the moment. It was not evoking the association of Arabic as 'the language of tradition' or French as 'the language of authority' that made an impact, nor was it the use of the unexpected code for a family setting. It was simply the change from one language to another that served to call the children's attention and imply that things could not continue in their present state. Following are examples from Dakar where switches between Wolof and French serve to highlight or organise the discourse: (4) Mais xam nga, je parlais à un ami journaliste l'autre jour. Ma ne ko 'Pourquoi on ne peut pas nous laisser créer un journal durable avec un point de vue différent?' Mu ne ma 'Attend rekk, on travaille là-dessus'. (But, you know, I was talking to a journalist friend the other day. I

< previous page

page_90

next page >

< previous page

page_91

next page > Page 91

said to him 'Why can't they let us create a stable newspaper with a different point of view?' He said to me 'Just wait, we're working on it'.) (5) Sama grandpère, sama 'pàpp' u yaay. (My grandfather, my mother's 'father'.) (6) Le matin de bonne heure ngay jóge Medina 'pur' dem julli. Suba reel nga fa war a jóge. (Early in the morning you leave Medina 'to' go pray. Early in the morning you should leave there.) We find in these examples that several of the functions noted by Gumperz are fulfilled. In (4), the speaker sets off reported speech from the rest of his conversation by describing who says what to whom in Wolof. The conventionalised interjection xam nga 'you know' is also inserted into a dominantly French phrase. In (5), the speaker qualifies his use of the French word for grandfather with a more specific phrase describing his exact relation as called for by tradition. The description in (6) of the speaker's activities during a pilgrimage to Mecca includes the phrase le matin de bonne heure followed by its almost direct repetition in Wolof suba teel for emphasis. All of these examples are typical of the fluid urban code and were recorded during informal conversations between intimates. Switching to fulfil discourse functions may also occur as part of a purposive switch. In such a case, the various connotations of a language may be evoked at the same time that a particular phrase is being highlighted. In the following example, a customer shopping in a European-style Dakar supermarket makes a 'double' codeswitch of this kind. He asks politely in the more formal French how much he should pay, then afterward appeals to the cashier in Wolof to let him have a few more bags. His change of language not only evokes the tradition of helping out a co-ethnic but serves to mark the desired change from a business-like to a more friendly exchange. (7) C'est combien? Vous avez dit?. . .May ma ay mbuus waay! (How much is it? What did you say? . . .Give me some bags, would you?) The discourse function fulfilled here is referred to by Gumperz (1982) as 'personalisation' (vs. 'objectivisation'), where a speaker plays upon the connotations of a 'we code' (vs. a 'they code') to create a conversational effect. Much of the switching described by Heller and Myers-Scotton depends upon the same sort of connotations for the speakers to achieve their various social and political goals. My aim in briefly mentioning some approaches used in studies of code-switching is not to point out their shortcomings. Rather, I wish to suggest that a certain kind of codeswitching, what I refer to here as an urban code, falls outside the domain of purposive codeswitching and thus that its analysis requires another kind of model. Some scholars (Myers-Scotton in particular) have outlined insightful concepts that can be applied in describing urban

< previous page

page_91

next page >

< previous page

page_92

next page > Page 92

codes, but their attention remains primarily focused on the kinds of switches that have a specific motivation. 6 Scholars may have chosen to concentrate on those types of codeswitching to which a more definite meaning can be attributed. Myers-Scotton (1989:343) has suggested that the methodology used in many codeswitching studies has resulted in data exhibiting 'more marked codeswitching than unmarked; conversations, for example, are recorded in formal situations where 'flagged' switches are used for metacommentary or where fluid switching would be inappropriate. It is also clear that a certain number of conditions are necessary for the development of an urban code and that all multilingual settings will consequently not produce one (Scotton, 1987; Myers-Scotton, 1989; Poplack, 1988; Heller, 1988). But even in the places where an urban code has developed, it has often not received detailed attention. What I argue here is that a larger place must be made in the literature for the sociolinguistic description of this urban phenomenon, especially in a situation like Dakar where its use is more frequently encountered than codeswitching of the goal-oriented variety. I will not attempt, in the remainder of this article, to provide an alternative model capable of explaining urban codes just as others have attempted to explain purposive codeswitching. Much more research must be completed before such a project is undertaken. Instead, I will describe some of the aspects of urban codes that, through my own research and the information available in the literature, appear to be important for understanding their sociolinguistic role. These are (1) the status of urban codes as a distinct 'language' and (2) the complicated and often contradictory attitudes surrounding their use. In both cases, the insiders' view helps to provide a full and complex picture of the position occupied by urban codes in their respective communities. Urban Wolof as a Separate Code Several factors must be considered in order to classify the type of fluid and unmarked switching described above as a distinct code. First, the differentiation of an urban code from its constituent languages must be justified. Second, its lack of status as a distinct code in the minds of those who speak it must be explained. The notion of viewing this type of switching as a separate entity is not new. Ure (1972:226) found in Ghanaian mixed language that 'all the bits belong together situationally and linguistically as part of one whole'. Agheyisi (1977) states that 'interlarded speech', Nigerian languages and English switched at the clausal, lexical or morphemic level, is used in such a way as to be perceived as a single code. Indian scholars tend to describe the codemixing of South Asian languages and English as constituting either discrete registers in the repertoire of multilingual speakers (Kachru, 1978b; Sridhar, 1978) or separate codes themselves (Kachru, 1982; Singh, 1985). More recently, Gibbons (1987:42) has attempted to treat Hong Kong MIX,

< previous page

page_92

next page >

< previous page

page_93

next page > Page 93

a speech form widely used on the University of Hong Kong campus, as 'an alternative code choice rather than simply as rapid Cantonese-English code-switching'. And finally, Myers-Scotton has characterised the unmarked intrasentential codeswitching found in certain communities as 'hybrid' forms which may be such 'that the two components form something distinct from either donor system' (Scotton, 1988:158). Pandit (1986) has done a very detailed description of intrasentential switching as a single code in her work entitled Hindi English Code Switching, Mixed Hindi English. She notes that codeswitching studies have paid little attention to a mixed linguistic code like MHE which is used by speakers 'as an effective communicative resource in situations where they would not use either of the languages of which this variety is a mixture' (1986:69). Pandit describes MHE as having a 'distinct communicative role' (1986:71) and thus suggests that it be considered on a par with both Hindi and English. She goes to great length to distinguish MHE from pidgins such as (Indian) Butler English, noting that MHE, instead of exhibiting the reduced lexicon of most contact languages, has an expanded lexicon with terms drawn from both constituent languages and expressing subtle shades of meaning. Also, MHE has the properties of other 'full fledged' languages in that it fulfils 'communicative referential', 'expressive connotational' and 'social symbolic' functions (1986:76), the last being especially evident in MHE's association with the educated, urban Hindi-speaking community. There is even a small amount of creative writing being produced in this mixed form. Pandit ends her monograph by attempting to uncover the grammatical 'system' that MHE must possess as a distinct code, but due to the small corpus analysed utterances she is, by her own admission, unable to complete this task. While Pandit has gone further than any other researcher, to my knowledge, in attempting to show how the claims of a separate status for mixed codes can be justified, she focuses primarily on the formal aspects of this question, limiting her sociolinguistic descriptions to the (often repeated) characteristics of MHE as an 'informal code' used by 'educated' and 'urban' speakers of Hindi. In an interesting and rather unexpected observation, she states that speakers are aware of the fact that they are speaking neither English nor Hindi, calling it instead khircari or 'mixture'. This differs from the descriptions found of other mixed South Asian codes where speakers seem to consider them a variety of indigenous languages, although not of English (Fernando, 1977). For example, Annamalai (1978:239) writes that: This mixed variety is considered by the participants in the speech act as a variety of their mother tongue and not as a variety of English. When someone speaks the Tamil variety of this mixed language, another language conscious person may comment 'Are you speaking Tamil?' or 'You are murdering Tamil' and not 'Are you speaking English?' or 'You are murdering English'.

< previous page

page_93

next page >

< previous page

page_94

next page > Page 94

The importance of this recognition on the part of MHE speakers should not, however, be misinterpreted, as will be explained below. In justifying my own claim that the rapid and unmarked intrasentential codeswitching found in Dakar be considered a distinct code, I have used some of the same supporting details as Pandit. Urban Wolof, like MHE, plays a distinct communicative role in the repertoire of bilingual Dakarois. It is used in situations where the choice of pure Wolof or pure French would be inappropriate, these situations almost always being of an informal nature. Examples include friendly gatherings during meals or tea-drinking sessions and unofficial discussions about work between colleagues in the workplace. For certain speakers, namely young male Dakarois, the choice of Urban Wolof for conversation with peers is practically de rigueur. The distinguishing aspects of these young men's speech, such as the insertion of English lexical items and the use of the tag quoi, could even serve to qualify their speech style as a 'social dialect' of Urban Wolof. Unlike MHE, however, Urban Wolof cannot be said to have a recognised status as a separate form in the minds of its speakers. Much like the mixed form of Tamil and English, Urban Wolof is generally thought of as a variety of Wolof. My selecting a name for this code followed the most common descriptions given by Dakarois themselves of this speech form as le Wolof de Dakar 'Dakar Wolof', le Wolof de la ville 'city Wolof', etc. Manessy (1990) has noted that the development of 'urban varieties' of widely spoken languages in Africa is quite common, a few examples being Peul in Cameroon, Sango in the Central African Republic, and Arabic in the Sudan. These varieties have a simplified structure and a lexicon drawing on both their traditional forms and European languages. One of the primary features of these varieties is that they have undergone a process of 'devernacularisation' (see Thiam, 1990) whereby they become dissociated from the values traditionally linked to their use in rural village settings. In place of these values, urban varieties take on different urbanised values which are reflected, in part, by their lexical innovations of European origin. But such urban varieties remain true to their 'roots' in the sense that their grammatical structure as the vernacular is not compromised. The Peul spoken in Yaoundé is still recognisably Peul, the Sango of Bangui is still Sango, and so on. Although at first glance it might seem that assigning a distinct name to an urban variety or mixed code would carry a positive significance, it has been found that the attribution of names to 'non-traditional' forms of speech may, in fact, have a negative connotation. Instead of these names reflecting the separateness and thus legitimacy of mixed codes, some researchers have suggested that the use of terms such as 'Singlish' (Sinhala and English, Fernando, 1977) and 'Francarabe' (Moroccan Arabic and French, Moatassime, 1974), expresses the judgement that they are unsuitable forms of speech. Gumperz writes of Chicano speech that 'such language mixture tends to be disparaged and referred to by pejorative terms such as Tex-Mex' (quoted in Kachru, 1978a;111). Thus, it is the absence of a name that

< previous page

page_94

next page >

< previous page

page_95

next page > Page 95

may serve instead as an indicator of the acceptance of such mixed forms in the linguistic repertoire of multilinguals. That is, it suggests that the use of these codes is unmarked or considered a normal choice. Urban Wolof, significantly, has never been marginalised by a 'neither fish nor fowl' term. It must also be noted that although Mixed Hindi English is distinguished from its 'parent' languages by its speakers, such recognition does not protect it from negative sanctions. Like many of the mixed languages found around the world, it is regarded by its speakers as a debased code, and Pandit (1986:84) writes that the 'mixing of English lexical items and expressions has been considered so undesirable that it has been considered worthwhile to coin scientific and technical terms with a Sanskrit base rather than use familiar terms from English'. Although definitions of what constitutes a language may well include, in addition to certain formal features, an attitudinal component whereby its status is affirmed on the part of its speakers (e.g. Ferguson, 1978), I maintain that Urban Wolof does occupy an autonomous position in the Dakar speech repertoire. This position cannot be supported, perhaps, in the purely linguistic sense, but it can be unquestionably in the sociolinguistic sense. 7 That is, the rules of usage of Urban Wolof as well as a description of its community of speakers can be clearly specified, just as if one were to describe the communicative competence necessary to speak appropriately a more formally recognised language. Its particular features set Urban Wolof apart from both Wolof and French and the roles they play respectively in the repertoire of multilingual Dakarois.8 Practice and Perception In order that the preceding claimthat Urban Wolof be considered an autonomous code despite attitudes that deny it this statusmay not be seen as paradoxical, some discussion of linguistic attitudes and their relation to the actual use of languages is necessary. It has long been a recognised fact of sociolinguistic research, and the study of human behaviour more generally, that there is a disparity between how people act and how they say they act, or think they should act. Labov (1972) has observed that speakers may be unconscious of their use of low-prestige phonological patterns and tend to recognise their 'phonic intentions' instead. Gumperz (1982:62) writes that 'attempts to elicit such self-report information on bilingual usage regularly show significant discrepancies between speakers' descriptions of their own usage and empirical studies of tape recorded texts'. Poplack (1982:2) believes that speakers often do not notice subtle differences in the way they change their linguistic behaviour from situation to situation, and when they do they elevate such differences to 'categorical occurrences'. Similarly, Silverstein (1979) has described this tendency by contending that speakers form their idea of how language should be used, in other words their 'linguistic ideology', from

< previous page

page_95

next page >

< previous page

page_96

next page > Page 96

language situations where a choice is highly 'presupposed' and consequently do not recognise those acts of speech that are more 'creative'. These sociolinguistic patterns are significant here since they document human tendencies which are perplexing but may ultimately provide the researcher a key to understanding language use. Speakers 'codify' their ideas of how language should ideally be used by perceiving real language use in an (unconsciously) categorical manner. Subsequent failures to conform to this ideology then become conveniently inaccessible to conscious reflection, at least without a concerted effort. The result is the discrepancy described above, so salient to the sociolinguist and usually unnoticed by the community being studied, between how speakers think they should speak and how they do speak. It became apparent early in my study that Dakarois bilinguals exhibited such conflicting tendencies in their use of intrasentential Wolof/French switching to a high degree. Linguistic practice and perception were often at odds in two general ways: (1) speakers would actively denounce Urban Wolof although the greater part of their own utterances were made in precisely that code, and/or (2) speakers would deny or simply be unaware of their very regular use of Urban Wolof, believing their speech to be instead a purer form of Wolof or French. In the course of being interviewed, a young Wolof man observed that although much of the Wolof spoken in Dakar was unfortunately mixed with French, it was not impossible for him to produce a pure form of his native language. He spoke thus: (8) Mën nañu essayer léegi nii. Mën naa la tontu ci li nga may laaje ci olof sans boole ces mots tubaab. (We can try right now. I can answer what you ask me in Wolof without putting in those French words.) This phrase was uttered without deliberation and without the speaker afterwards realising that what he had just said was not, in fact, pure Wolof. Such 'self-contradicting' phrases frequently arose during discussions of language purity and the knowledge of 'deep' Wolof in Dakar. Other examples are (see also example 3 above): (9) Dégg nga olof?. Kon, c'est pas la peine ñu làkk 'faranse'! You understand Wolof?. Then it's not worth it for us to speak French! (10) Di naa la jàngal olof 'piir'! I'll teach you 'pure' Wolof! (11) Yaa ngiy làkk olof normal. You're speaking good 9 Wolof. The lack of awareness on the part of the speakers of any contradiction in these statements illustrates the other interesting disparity found between linguistic practice and perception in Dakar: the contradiction between the unremarkable fact of its use and the negative views that are regularly given of it. That is, at first glance it seems unlikely that the use of Urban Wolof could manage to pass unnoticed since Dakarois so often speak of it as an

< previous page

page_96

next page >

< previous page

page_97

next page > Page 97

objectionable form of speech. Such attitudes toward the mixing of languages do not characterise Dakar alone but seem to be found wherever urban codes are in common usage. Mixed codes are typically considered by members of bilingual speech communities, including their habitual users, to be deteriorated forms of local languages, evidence of the loss of cultural integrity and/or proof of a lack of linguistic competence in any language, whether this phenomenon be found in New York City (Poplack, 1979), Nigeria (Agheyisi, 1977), Thailand (Warie, 1977), India (Kachru, 1978a; Pandit, 1986), Morocco (Bentahila, 1983) or Hong Kong (Gibbons, 1987). Speakers in these various communities have not infrequently been 'caught' using a mixed code in the very utterances where they either disparage the form or deny their use of it (see Verma, 1976; Pandit, 1986; Romaine, 1989). It seems difficult, at times, to believe that such phrases could really pass unnoticed by speakers when to an outsider they can sometimes seem to verge on the absurd or at best hypocritical. But where exactly do such attitudes fit into the confusing relation between ideals of linguistic use and actual linguistic use just described? One view might be that the study of attitudes in sociolinguistic enquiry is the key to uncovering the linguistic ideology of a speech community. Since stated attitudes are presumably uppermost in the minds of speakers, such attitudes could be considered a reflection of what people think they ought to be doing. The reason, then, that the use of mixed languages is so often negatively viewed is that linguistic ideologies rarely seem to include an accepted place for these codes. Their use thus consistently violates speech community ideals. But this is, at best, a conservative interpretation of the role of attitudes. Since opinions are often elicited under very formal circumstances, the researcher is led back to the primary dilemma discussed above: that only a very codified part of sociolinguistic patterning is available to a speaker's conscious awareness. In my own study of the practice and perception of languages in Dakar, I sought to reach another, less consciously formulated level of attitudes. Much like those sociolinguists who use the 'matched-guise technique' for tapping certain underlying feelings about languages and their speakers that would be difficult to elicit in a more direct manner, my own approach aimed at getting below the formal linguistic ideology to a more 'hidden' set of attitudes toward languages and their use. These two sets of attitudes correspond more or less to what Edwards (1985:140) has described as the 'belief' and 'feeling' components of attitudes, the first reflecting the normative side, the second the affective side of language use. My approach revealed that linguistic attitudes in Dakar were multi-layered in this way, with beliefs referring to the shared linguistic ideology and 'gut reactions' to actual linguistic practice. Thus, when asked to comment on excerpts of conversations in which Urban Wolof was used, my interviewees reacted very acceptingly, even when they had earlier denounced the practice of speaking Wolof interspersed with French.

< previous page

page_97

next page >

< previous page

page_98

next page > Page 98

It is in this way that the use of mixed codes by those who often belittle them can be understood. Speakers are not being paradoxical or hypocritical by speaking in a way that they themselves believe to be unacceptable. The acceptability of mixed codes simply resides at the deeper level of feeling, less-accessible to both the researcher and the speaker. This idea is certainly not unheard of in the sociolinguistic literature, where it is usually referred to with the term 'covert prestige' (cf. Labov, 1966; Stevens, 1983; Trudgill, 1984; Gibbons, 1987). However the existence of this type of prestige, where 'the overt downgrading of non-standard varieties may co-exist with more latent positive connotations' (Edwards, 1985:149), has more often been deduced than illustrated. The exception is Trudgill's study which shows statistically that for working class men in Norwich, England, 'non-standard speech is in a very real sense highly valued and prestigious' (1984:172). In my own study of attitudes toward Urban Wolof I have similarly tried to show, although using a more qualitative approach, that its 'covert' acceptance is real and can (and will) be expressed by the speakers themselves. Such elicitations, however, require an approach that first accepts and then bysteps the speaker tendency to use, in a moment of conscious reflection, only their linguistic ideology or belief as a basis for opinions and judgements. Besides serving as concise illustrations of the 'practice/perception split', the self-contradicting phrases cited above can also be taken as proof of the unmarkedness of Urban Wolof as a code choice in Dakar. Little thought is given by most speakers to which language the lexical items or clauses they utter derive from. When asked immediately after speaking which language or languages were used, speakers will typically remember the referential content but may well repeat what was said using a different combination of French and Wolof. Given this level of unmarkedness, it is not surprising that speakers should make such 'illogical' slips, especially since pure Wolof is the single appropriate code in only a few speech contexts. In fact, it would be more surprising if multilingual Dakarois committed such lapses in a situation calling for pure French since the exclusive use of the official language in the classroom has accustomed most bilingual speakers to a stricter control over that language. Pandit (1986) and Valdès-Fallis (1978) have similarly noted that pure English is more easily produced than pure Hindi and pure Spanish respectively by the Indian and Chicano populations they studied. In contrast Wolof, since it has never been formally studied by its speakers, is rarely thought about by the average Senegalese as a linguistic 'system' in the same way. The result is that attempts to speak absolutely pure Wolof often fall short, despite a stated concern with knowing and safeguarding the language. Such concern was obvious, for example, through the Dakarois' avid watching of a game show called Baati Olof 'Wolof Words', which had been aired on television several years before my arrival in Dakar but was still spoken of with great delight. The aim of the show was for contestants to speak on a given topic for several minutes without including a single word

< previous page

page_98

next page >

< previous page

page_99

next page > Page 99

of French, even one phonologically assimilated and thus usually an accepted part of the Wolof lexicon. Topics were ones that rendered the use of pure Wolof extremely difficult, e.g. a description of school activities where Wolof equivalents had to be found for tableau 'blackboard', dictée 'dictation', etc. Dakarois were at the same time amused and impressed by contestants' attempts to 'do the impossible'speak Wolof as if they lived in an isolated rural village rather than an urban area whose population is exposed to multiple linguistic and cultural influences. Although the disparity existing between linguistic ideals and use has been recognised by sociolinguists working in a number of settings, I believe that this split is exacerbated in Senegal by the fact that, like much of Africa, it is undergoing rapid social change. French, once the language of prestige and success, is being pushed out of many of what were once its exclusive domains. Many feel that the French language has passed its heyday in Senegal and that francophony can no longer maintain its former privileged position. This disillusionment with the possibilities of French has not, however, produced a reactionary return to Wolof or other indigenous languages among Dakarois. The linguistic future seems to lie, rather, between the two, in Urban Wolof. Yet attitudes supporting this development of the role of the mixed code seem to be evolving more slowly than its use. The result is that there exists perhaps an even wider gulf between perception and practice than might be found in a less rapidly changing society. Conclusion In these pages, I have attempted to present some of the difficulties inherent in describing an urban code. Used without any immediate goal in mind and often passing unnoticed by speaker and interlocutor alike, Urban Wolof is an innovative mode of speech that connotes informality, bilingual-ism and participation in the life of a large city. Yet its characterisation is not simple since the ambivalence surrounding its status and its unconscious usage by many speakers complicates a straightforward evaluation of its position. It is the complexity provided by the insiders' view, however, that suggests that Urban Wolof is not a passing trend but a serious contender for the place of most useful and, now, even the most sociolinguistically well established code in Dakar. As such, it is important that it be further studied and described. Now to answer the question posed in the title: should an urban code be considered the use of two codes or should it be viewed as a distinct one in and of itself?. My data indicate that Urban Wolof is linguistically a variety of Wolof which has earned itself a separate sociolinguistic status. This position will undoubtedly be reinforced in the years to come as the use of Urban Wolof spreads to those sectors of the population not possessing a strong control over the French language. It has already been remarked that many illiterates are able to use Urban Wolof with a surprising skill by

< previous page

page_99

next page >

< previous page

page_100

next page > Page 100

imitating their more educated peers (Thiam, 1990). Given the declining prestige of French and the weakening of ties between city and rural populations, it seems reasonable to predict that Urban Wolof, despite the campaigns of certain 'purists', will become the native language model of many Dakar children in the years to come. The question remains as to whether the Senegalese government will be able to take advantage of this new urban lingua franca in order to more effectively disseminate information to the greater population through both written and oral media. For the moment, Urban Wolof occupies too ambiguous a position in the attitudes of its usersif not in their sociolinguistic habitsto be recognised in any official manner. As for the relevance of the observations made here for the study of other urban codes around the world, the brief descriptions available of such codes suggest they they share many features in common with Urban Wolof, both in their roles as informal and in-group varieties and in the ambiguity of their status. What is needed now are in-depth sociolinguistic studies so that the various functions fulfilled by these codes in their respective communities, their various rules of usage, the attitudes surrounding them, the characteristics of their speakers, and the ways in which they are transmitted can be better documented and understood. First, the essential differences that exist between urban codes and the more commonly described kinds of codeswitching need to be acknowledged. Only then can they be studied as a separate phenomenon instead of a series of codeswitches that are largely ignored because they are unable to fit the sociolinguistic model at hand. Notes 1. This research was funded by a Fulbright Research in Africa Grant from October 1986-September 1987 and by a Foreign Language and Area Studies Fellowship from February 1988-May 1989. 2. A broad sociolinguistic survey was made throughout Senegal by questionnaire in the early 1970s (Kane, 1974). While it pointed out certain patterns in the relation of language use to ethnic background, it offered very little detail on the more personal attitudes and usages of its subjects. 3. Older men spend a large amount of their free time in prayer and engaged in other religious activities, and male retirees in any neighbourhood (there is mandatory retirement from civil service at the age of 55) will be found in large groups chatting, saying their prayer beads, and reading the Koran. It was more difficult for me to get to know older men than members of any other age or gender group in Dakar. One 76-year old man became the perfect research subject, however, when he unfortunately broke his leg and had to remain immobile for several months. 4. Myers-Scotton (1988) has also pointed out that phonological or other types of assimilation to the matrix language may occur in either loanwords or switched items but is necessary in neither. This criterion is therefore not a sufficient one for distinguishing between these two phenomena. 5. The word olof is a variant of wolof, used especially in reference to the language (as opposed to the ethnic group). 6. I am referring here to their work that has focused on the sociolinguistic aspects of codeswitching. The structural models that they have developed for codeswitching may well be relevant.

< previous page

page_100

next page >

< previous page

page_101

next page > Page 101

7. Linguists Ndiasse Thiam and Alioune Ndao of the Centre de Linguistique Appliquée de Dakar have arrived independently at the same conclusions concerning the status of Urban Wolof (personal communication, December 1990). 8. A more complete description of the sociolinguistic grammar of Urban Wolof will be found in Swigart (forthcoming). 9. Olof normal in this context means Wolof spoken as if by a native speaker. References Agheyisi, Rebecca (1977) Language interlarding in the speech of Nigerians. In Kotey and Der-Houssidian (eds) Language and Linguistic Problems in Africa. Columbia: Hornbeam Press. Annamalai, E. (1978) The anglicized Indian languages: A case of code mixing. International Journal of Dravidian Linguistics 7(2), 239-47. Bentahila, Abdelâli (1983) Language Attitudes among Arabic-French Bilinguals in Morocco. Clevedon: Multilingual Matters Ltd. Edwards, John (1985) Language, Society and Identity. Oxford: Basil Blackwell. Ferguson, Charles A. (1978) Multilingualism as object of linguistic description. Studies in the Linguistic Sciences 8(2), 97-105. Fernando, Chitra (1977) English and Sinhala bilingualism in Sri Lanka. Language in Society 6, 3, 341-60. Gibbons, John (1987) Code-Mixing and Code Choice: A Hong Kong Case Study. Clevedon: Multilingual Matters Ltd. Gumperz, John J. (1982) Discourse Strategies. Cambridge: Cambridge University Press. Heller, Monica (1988) Codeswitching: Anthropological and Sociolinguistic Perspectives. Berlin: Mouton de Gruyter. Kachru, Braj B. (1978a) Code-mixing as a communicative strategy in India. In J. E. Alatis (ed.) International Dimensions of Bilingual Education. Georgetown University Round Table. Washington D.C.: Georgetown University Press. (1978b) Towards structuring code-mixing: An Indian perspective. International Journal of the Sociology of Language 16, 2746. (1982) The bilinguals' linguistic repertoire. In Hartford, Valdman & Foster (eds) Issues in International Bilingual Education: The Role of the Vernacular. New York: Plenum Press. Kane, Francine (1974) La Sociologie des Langues au Sénégal. Thèse de Troisième Cycle. École des Hautes Études, Paris. Labov, William (1966) Hypercorrection by the lower middle class as a factor in linguistic change. In W. Bright (ed.) Sociolinguistics. The Hague: Mouton. (1972) Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press. Manessy, Gabriel (1990) Modes de structuration des parlers urbains. Paper presented at the International Conference Des Langues et des Villes, Université Cheikh Anta Diop de Dakar, December 1990. Moatassime, A. (1974) Le bilinguisme sauvage au Maroc. Revue du Tiers Monde 15(59-60), 619-70. Myers-Scotton, Carol (1989) Codeswitching with English: types of switching, types of communities. World Englishes 8(3), 33346. (1990) Building the frame in codeswitching: evidence from Africa. Paper delivered at the 21st Annual Conference on African Linguistics. University of Georgia, Athens, Georgia. Pandit, Ira (1986) Hindi English Code Switching, Mixed Hindi English. Delhi: Datta Book Centre. Poplack, Shana (1979) Sometimes I'll Start a Sentence in Spanish Y TERMINO EN ESPANOL: Toward a Typology of CodeSwitching. Language Policy Task Force 4. New York: Centro de Estudios Puertorriqeços. (1982) Bilingualism and the vernacular. In Hartford, Valdman & Foster (eds) Issues in International Bilingual Education: The Role of the Vernacular. New York: Plenum Press. (1988) Contrasting patterns of codeswitching two communities. In M. Heller (ed.) Codeswitching: Anthropological and Sociolinguistic Perspectives. Berlin: Mouton de Gruyter.

< previous page

page_101

next page >

< previous page

page_102

next page > Page 102

Romaine, Suzanne (1989) Bilingualism. Oxford: Basil Blackwell. Scotton, Carol Myers (1987) Code-switching and types of multilingual communities. In P. Lowenburg (ed.) Georgetown University Round Table on Language and Linguistics 1987. Washington D.C.: Georgetown University Press. (1988) Codeswitching as indexical of social negotiations. In M. Heller (ed.) Codeswitching: Anthropological and Sociolinguistic Perspectives. Berlin: Mouton de Gruyter. Silverstein, Michael (1979) Language structure and linguistic ideology. In P. Clyme, W. Hanks and C. Hofbrauer (eds) The Elements: A Parasession on Linguistic Units and Levels. Chicago Linguistic Society. Singh, Rajendra (1985) Grammatical constraints on code-mixing: evidence from Hindi English. Canadian Journal of Linguistics 30(1), 33-45. Sridhar, S. N. (1978) On the functions of code-mixing in Kannada. International Journal of the Sociology of Language 16: 10917. Stevens, Paul (1983) Ambivalence, modernization and language attitudes: French and Arabic in Tunisia. Journal of Multilingual and Multicultural Development 4(2 & 3), 101-14. Swigart, Leigh (forthcoming) Practice and Perception: Language Use and Attitudes in Dakar. Doctoral dissertation, University of Washington, Seattle, Washington, USA. Thiam, Ndiasse (1990) L'évolution du wolof véhiculaire en milieu urbain sénégalais: le contexte dakarois. Plurilinguismes 2, issue entitled 'La dynamique des langues au Sénégal': 10-37. Trudgill, Peter (1984) On Dialect: Social and Geographic Perspectives. New York: New York University Press. Ure, Jean (1972) Code-switching and mixed speech in the register systems of developing languages. In Proceedings Volume II of l'Association Internationale de Linguistique Appliquée, Third Congress. Heidelberg: Julius Groos Verlag. Valdès-Fallis, Guadalupe (1978) Code-switching among bilingual Mexican-American women: towards an understanding of sexrelated language alternation. International Journal of the Society of Language 17, 65-72. Verma, S. K. (1976) Code-switching: Hindi-English. Lingua 38(2), 153-65. Warie, Pairat (1977) Some aspects of code-mixing in Thai. Studies in the Linguistic Sciences 7(1), 21-40.

< previous page

page_102

next page >

< previous page

page_103

next page > Page 103

Socialising Multilingualism: Determinants of Codeswitching in Kenyan Primary Classrooms 1 Marilyn Merritt with Ailie Cleghorn Jared O. Abagi Grace Bunyi 2706 Key Boulevard, Arlington, VA 22201, USA Abstract Individual multilingualism in a multilingual society is often associated with a mass education system. The Kenyan situation illustrates the intertwined complexity of the pedagogical and socialising aspects of language contact in such a system. Using ethnographic observation of classroom interaction in three primary schools, determinants of teachers' language choice and codeswitching among English, Swahili, and mother-tongue are explored: official school policy, cognitive concerns, classroom management concerns, values and attitudes about societal multilingualism. Analogously to monolingual modality switching, codeswitching between languages often provides a resource to focus or regain students' attention, or, to clarify, enhance, or reinforce lesson material. Instructional demands and inconsistent patterns of use, however, may also lead to linguistic insecurity and the mislabelling of content or conceptual problems as language competency problems. Understanding the complex role of language, and how teachers informally provide models for behavior and attitudes in the context of formal instruction is important for educational policy. Introduction Like many others, Kenya is a nation struggling to combine the revered values of a traditional largely rural way of life with the demands for modernisation and economic development to support its rapidly growing population. Children lead lives that are very different from their parents and grand-parents; the learning experiences through which they are socialised into

< previous page

page_103

next page >

< previous page

page_104

next page > Page 104

becoming accepted and competent members of adult society are concomitantly different. This difference indexes a process of cultural and societal transition that has salience for most developing countries today. It is also resonant with the processes of social change and assimilation that are taking place in large pluralistic developed countries like the United States. Socialisation is the experiential process through which an individual informally learns how to comport him/herself as a socially accepted and competent participant within some social domain. It takes time and practice, and involves modelling on recognised 'members' or 'experts' for the absorption of socially acceptable knowledge, attitudes and habits of association, as well as particular behaviours. Typically one or more of the models will be imbued with authority to direct and evaluate behavioural outcomes of the new member or 'novice'; early in the socialisation process negative sanctions are usually absent or mild. Language socialisation refers to the experiential process through which an individual (usually a child) learns how to use the communicative resource of one or more languages, within socially defined domains of activity; this is an integral part of the process of learning how to use the entire range of communicative resources (e.g. nonverbal modalities). (See also Hoggart, 1957; Cook-Gumperz, 1986; Heath, 1984; Schieffelin & Ochs, 1986.) The capacity of individuals to perform successfully in new situations depends heavily on what they have learned informally through socialisation processes as well as what they have been formally taught. In trying to understand better how individuals and societies can more readily adapt to the pace of social and technological change, research must seek to find ways of 'seeing' these processes and how they are interrelated. A better picture of any single situation or socialising institution will contribute to the general knowledge that is needed by policy makers and those responsible for planning social and economic development. The situation of Kenya is particularly relevant because social and technological change has been dramatic, population growth has accelerated at an unprecedented rate, and the complexity of a multilingual society is present. In Kenya's recent history there has been a basic shift from a traditional locally-based non-literate approach to educating the young, to using a time consuming government-regulated mass educational system. The resulting primary schools are recognised as socialising institutions that promote values and attitudes as well as technical skills and abilities. These are explicitly stated in the National Goals of Education printed in the Syllabuses for Kenya Primary Schools (1986, ix-xi). However, although policy makers know the student outputs that they hope to achieve and are largely aware of the range of student inputs to the system, very little is known about the classroom processes that transform the inputs into outputs, or the complexities involved in achieving the educational objectives (see Gorman, 1970; Whiteley, 1971; Eisemon, 1991). Classroom teachers are, in effect, expected to perform miracles, while important reasons for poorer-

< previous page

page_104

next page >

< previous page

page_105

next page > Page 105

than-desired student outputs may be overlooked or misidentified (see discussion below on linguistic insecurity). An important aspect of the situation in Kenya is societal multilingualisma condition shared by many developing countries. The availability of more than one language provides a resource for universal classroom objectives that must be met differently in monolingual settings. As with any resource, there are the costs of acquisition, maintenance, and management. Here we will skirt the complex issues of acquisition and maintenance and focus rather on management. How do teachers actually do this? In particular, what determines the teacher's choice of language for a given context and incipient shifts from one language to another? This paper explores the determinants of teachers' language choice and codeswitching in classrooms of three Kenyan primary schools. We use ethnographic observation of classroom interaction as a tool for probing the complex dynamics involved (see also Cazden, 1988; Martin-Jones, 1988; Scotton & Ury, 1977; Scotton, 1988). We argue that, although official policy is the major factor in language choice and lesson objectives is the major factor in teachers' codeswitching, there is overlaid a subtle permeating factor of language socialisation that reflects and reinforces the positioning of English, Swahili, and 'mother-tongue' in the collective repertoire of Kenyan society. Determinants of language choice and codeswitching in the classroom are necessarily more complex than can be 'legislated' by language policy on medium of instruction. Recognising this may help policy makers to realistically rethink their models of how language will be used in the classroom. It may also help teachers to more accurately identify communicative strategies for educational effectiveness (as well as constraints), and to thereby become more aware and deliberate in their use of language. The Kenyan Situation: History, Policy, and Practice In Kenya today most children who attend primary school are confronted with living in a bi- or tri-lingual environment. This reflects the fact that Kenya is a nation with a multilingual repertoire and a national standardised system of government sponsored education. Though the geographical territory has historically always encompassed linguistically diverse groups, widespread multilingualism on the part of individuals, especially in the rural areas, is relatively recent. Many individuals from the larger ethnic groups have traditionally been monolingual. Kenya became an independent nation of East Africa in 1963, when the colonial British rule was relinquished. English has remained the official language of government. Swahili, designated national language and first language of a relatively small coastal group, has long been a lingua franca of the region, and with increasing urbanisation and movement between regions has come to be learned by more and more people, at earlier and

< previous page

page_105

next page >

< previous page

page_106

next page > Page 106

earlier ages. Though not all Kenyans speak either English or Swahili, all learn to speak their 'mother-tongue' during early childhood and continue to use it with family and friends from the same linguistic group. There are more than 30 of these mother-tongues, nine of which are used in official radio broadcasts, and most of which have little if anything in written form (Eastman, 1983; Whiteley, 1974). Individual multilingualism has been both cause and effect of political unification, modernisation, rapid social change, and significantly increased levels of intergroup communicationall important indices of Kenya's national development. In conjunction with the informal learning that always accompanies social neede.g. the on-the-street and in-the-bazaar learning of Swahili for interethnic communicationprimary education has been a major factor in furthering both national development and widespread multilingualism. The Kenyan government provides primary teachers (at no cost to local communities) and has established national standards for curriculum and terminal year (standard 8) examinations (for the Kenya Certificate of Primary Education). Policy makers are concerned with equity between regions and ethnolinguistic groups, and with establishing a curriculum that meets national needs for development. Having considered a number of factors, the national government sets all terminal year examinations (the KCPE) in English (except the now compulsory Kiswahili language examinations). In accordance, the official language of instruction from standard (grade) 4 onwards is English; the first three standards may be taught in various of the three languages, depending on the homogeneity and location of the school. Educational practitioners have the job of implementing this policy while teaching students to learn modern scientific concepts and how to read and write. From the point of view of the communities they serve, their main task is to teach students to pass the KCPE exam, which opens the door to higher education and perceived opportunities for economic advancement. It is not often recognised that at least two important complexities are involved in the ideal outcome: (1) Teachers must guide students to learn specific skills and facts, many of which are experientially disconnected from their home environments. (2) Because students are young and because often these students are being exposed to literacy and languages other than mother-tongue for the first time, teachers provide the major expert models for socialisation into the use of English and into a 'culture of literacy' in a society that is basically triglossic (with an orientation to diglossia between English and Swahili) (Abdulaziz, 1972; Merritt & Abdulaziz, 1988). The policy selection of English as the major medium of educational instruction in Kenya reflects a number of practical facts: the designation of English as the official language of Kenya, extensive use of English that began during the colonial period, the stature of English as an international language, the regional neutrality of English as the first language of only a small number of Kenyans, the availability of textbooks and other resources in English,

< previous page

page_106

next page >

< previous page

page_107

next page > Page 107

and the widespread use of English to express concepts in scientific and technologically salient areas throughout the world. These facts shape a political decision that is designed to minimise material expenditure and inter-regional inequity while maximising the desirability of successful student outputsstudents who can operate fully with modern scientific terminology. However, since most beginning primary students do not come to school already knowing English, Standard 1 teachers must usually provide English language instruction before or at the same time that they begin to use English as language of instruction. For many students, even though English is taught as a subject, this is tantamount to a 'sink or swim' English immersion for learning other subjects, although this is mitigated by the fact that teachers, in the lower primary grades especially, may be authorised to use mother-tongue or Kiswahili to facilitate content learning. The language used between teacher and student outside the formal lesson usually reflects the prevailing local pattern for the use of English, Swahili or mother-tongue. Such a multilingual situation effectively means that teachers make individual and moment-to-moment decisions about language choice that are only partially dictated by formal language policy. Teachers' language choice can thus be seen to be accountable to national and school policy, as mitigated by the immediate cognitive and management concerns of the classroom lessons, and by an orientation to the teacher's perceived norms of local and societal multilingualism. This paper reports some of the findings of a two year study of Kenyan primary schools. The general aim of the study was to make close observations and provide analytic descriptions of how languages are used in school, in ways that might have important implications for how languages are used out of school. An important objective has been to reach a better understanding of the overall sociolinguistic environment in which teachers operate and the nature of communicative demands and resources in this setting. In our study an effort was made to select three schools that would represent the three different linguistic situations that have been addressed by educational policy makers: (1) a school that was English medium through-out, (2) a school that was Swahili medium for the first three years, with a switch to English medium at Standard 4, continuing with Swahili being taught as a subject, (3) a school that used mother-tongue as medium of instruction for the first three years with English and Swahili being taught as subjects, and then switched to English medium at Standard 4 with Swahili continuing to be taught as a subject every year. In fact it was difficult to find a school of the second type. In all the Nairobi schools we surveyed there seemed to be an effort to use as much English as possible as early as possible. In School 1, though not explicitly stated there seemed to be a kind of track system whereby some students, at extra expense, began a pre-primary year in which there was heavy emphasis upon learning some English and preparing for Standard (grade) 1; all students who attended preprimary were grouped into the class of one teacher, who was generally viewed as the most experienced; these students

< previous page

page_107

next page >

< previous page

page_108

next page > Page 108

progressed at a much more rapid rate than students in either of the other Standard 1 classrooms. Our school 2 used English as the dominant medium of instruction, used Swahili occasionally, especially in the lower standards, and on some occasions Kikuyu, which was the dominant mother-tongue and school ground language. Both school 1 and school 2 are located in Nairobi. Our school 3 is located in Kisumu district, outside the city council limits. This is a rural and linguistically homogeneous area in which Luo is mother-tongue. School 3 used Luo as medium of instruction for the first three standards, and occasionally used Luo in addition to English even in the upper levels. Swahili was not much used outside the context of Swahili language lessons. In each school we directly observed numerous classroom lessons from Standard (grade level) 1 through Standard 8, made audio recordings that were transcribed, conducted group interviews with teachers, conducted a few experiments in reading and writing ability, and collected information on textbooks, teacher resource materials, and students' performance on nationally set examinations. In analysing the language choice and codeswitching that we observed in teachers' instructional language, we tried to interpret functions for the patterns we found. We looked first for functions that were motivated by cognitive or management concerns of the classroom. These indeed seemed to be dominant. Not surprisingly we also found that social habits and attitudes seem to play a significant role. Although these attitudes were confirmed in interviews with the teachers, our analysis is based on patterns of actual usage that may or may not reflect the speakers' conscious impressions of how they communicate. It is this kind of habitual doing that creates the model for socialisation processes. The Socialiser Role of the Teacher Throughout the lesson corpus there was ample evidence of teachers putting themselves forward as socialisers, not only in terms of teaching rules and behaviour specific to the classroom, but also in terms of social values and concerns generally. For example, when teaching the lesson on 'air' in School 3, the confusion between cooking gas and gas as a state of matter arose. This led into a discussion of the dangers of cooking gas. Concern about the conservation of resources is often invoked when teachers make evaluative remarks. For example: Example 1 School 3, Standard 8 (English language lesson) T: (to student whose paper he is examining) Why didn't you use this space? You are wasting. Tomorrow your father is going to buy another for you. end 1

< previous page

page_108

next page >

< previous page

page_109

next page > Page 109

Admonishments of classroom behaviour include explicit references to the use of language. For example: Example 2 School 1, preprimary T: Yes, somebody was talking in Kiswahili. Now that is what I heard. I heard somebody speaking in Kiswahili. end 2 Example 3 School 3, Standard 5 T: (after asking a question in English and getting a Luo answer) English wachi (Luo: speak English) end 3 Language Choice as Modality Choice: The Importance of Consistency for Informal Learning Language socialisation does not take place in a vacuum, but rather occurs through social interaction. Earlier research on classrooms in monolingual schools (Merritt, 1982a, 1982b; Merritt & Humphrey, 1979, 1980) found that good teachers display communicative patterns in which the communicative resources available to them were differentiated or 'split' in accordance with differing communicative needs or demands within the classroom. For example, the verbal modality was typically used for attending to the content of the classroom lesson, while if there was a need for classroom management the non-verbal modality would often be used. These modalities might be further split as classroom needs dictated; for example, by differentiating between full voice when addressing all the students and soft voice when addressing only one student. Students thus had to learn to make these differentiations in order to know which communication signals were relevant to them, and/or for what. That study found that the most effective teachers used the resource of modality splitting very consistently. Students' thereby informally learned the significance of the various modalities so that their attention would be appropriately directed for content learning and classroom participation. For example, a reading group using the oral-verbal modality learned to 'tune out' the teacher's non-verbal attention to math students (such as checking homework). This kind of informal teaching and learning does not take place consciously. Even the most effective teachers seemed to be unaware of what they were doing. One less experienced teacher sometimes attributed problems to student sloppiness during just those periods in which it was she who had first been inconsistent in her use of modality splitting, perhaps inadvertently generating a sense of classroom disorder. In the Kenyan schools that we observed classrooms were far more rigorously structured than in the North American school setting: in Kenya most class time was spent with the whole student group simultaneously involved in listening to the teacher and participating in established routines. Despite these differences we found that Kenyan teachers occasionally use

< previous page

page_109

next page >

< previous page

page_110

next page > Page 110

the non-verbal modality for managing interaction that is not part of the lesson, much as the U.S. teachers did. For example, a Standard 1 teacher in School 3, while continuing the lesson orally, took one young boy by the arm and walked him around to the other side of the room where he was no longer able to chat with his neighbour. Similarly, the issue of the teacher appropriately directing students' attention, and giving signals that allow students to process and interpret the information appropriately, is as applicable to Kenyan classrooms we observed as to the ones studied in the US. For example, teachers (in the lower primary grades especially) often use Swahili or Mother-Tongue to speak to one student privately (e.g. while checking homework) even though the lesson may be in another language. Just as it occurred in the monolingual US classrooms, in the Kenyan classrooms it seemed that changes in modality were used to indicate changes in what the teacher was communicating about. Consistency in the teacher's patterns of modality use may be the key to informally teaching students how to interpret any significance in a switch from one language to another. In the Kenyan study we found that older students were rarely confused about which language to use, or what was being said when there was a switch from one language to another. However, in the lower primary grades this did occasionally happen, indicating that being able to interpret the switch is something that is learned in addition to learning each language. For example, when a standard 3 teacher in School 3, who was teaching a lesson in mother-tongue Luo, suddenly introduced the English term 'gas', and the key (i.e. given in the curriculum) phrase 'air is a gas', one of the students heard this as the Luo word ngas (which means 'ladder') and offered in class participation that 'gas' was 'for climbing a house (child's statement made in Luo)'. This example clearly illustrates that the student had not grasped the switch to English. This particular teacher did not point out the source of the confusion but simply answered 'No'. By way of contrast, one of the most highly respected Standard 1 teachers in School 1 often pointed out differences between Kiswahili and English, the differences in how letters were pronounced, when one or the other was being used, and other ways of helping students to make the necessary distinctions and avoid confusion. An ability to guide students to quickly acquire this competency in interpreting codeswitching may be one of the 'hidden' teacher characteristics that mark outstanding teachers. Variation and Repetition; Linguistic Competency and Insecurity 2 The issue of consistency gives us another perspective on codeswitching, or what we may think of as variation in language choice. In almost any communicative setting variation and repetition can be seen as contrapuntal

< previous page

page_110

next page >

< previous page

page_111

next page > Page 111

factors in how communicative resources are used. They can be seen as opposites, each of which has positive and negative aspects. Variation in expression can sometimes be seen positively as elaboration, as demonstrating fluency with both subject matter and communicative modality. Variety and the use of metaphors and figures of speech have positive cultural value in Kenya. On the other hand variation can sometimes be viewed negatively as instancing inconsistency. For example, in Kenyan primary schools teachers, the curriculum, and national examinations often seem to be more demanding of students' use of prescriptively correct English grammar than is consistently used in the classroom. Many teachers we observed were inconsistent not only in oral use but in some blackboard use as well. These inconsistencies are mostly lack of noun and verb concord for singular and plural, inconsistent use of capitalisation in writing a list, and other 'errors' that do not significantly affect the semantic content of the lesson. Such 'errors' do not usually affect the content of the instruction (content is more reflected by the display of basic concepts and logical organisation of the lesson), nor do they indicate a lack of fluency in English. However, such variation in teachers' language provides a model of inconsistency which may make it difficult for students to perform well on the highly prescriptive KCPE examinations that are offered at the end of Standard 8. This inconsistency in the model that teachers provide to students is likely to be something that teachers are largely unaware of. Yet to the extent that they are aware of it, it probably contributes significantly to whatever sense of linguistic insecurity they may have about English language competence. Many teachers are the products of a similar classroom environment. Linguistic insecurity is probably linked in part to the difficulty East African speakers have in mastering grammatical consistency in constructions with semantically low value such as noun verb agreement in number (a feature that is morphologically quite different in East African languages). The interaction of the evolving standard of East African English and the current approach to testing English language competency according to British or American standard usage may have a negative impact on teacher confidence and effectiveness. As Mwanzi (1984:34) points out in discussing language policy, 'Confidence in the use of the language of instruction is necessary for conquering inhibitions in the school environment.' Most teachers did express concern about their competence in using English. However, all teachers displayed a great deal of fluency in English, as exemplified by the ability to say things in different ways, and use syntactic and morphological variability. Example 4 Standard 8, School 3 T: It is a plant or an animal that cannot live on its own. That is why we say it depends on another plant. When it is living on a plant we say it

< previous page

page_111

next page >

< previous page

page_112

next page > Page 112

is a plant parasite. When it is living on an animal we say it is an animal parasite. end 4 Example 5 Standard 4, School 2 T: How many have got the new book now? Put up your hands. Those of you who do not have these books please try and buy before the end of this week, do you hear? end 5 Linguistic insecurity is probably also related to the difficulty teachers experience in relating new concepts, especially scientific concepts for which there is vocabulary only in English. When generic terminology (such as 'temperature' to speak of either 'hotness' or 'coldness'; or 'gas' to speak of one of three possible 'states of matter') is not available in the indigenous language, teachers experience difficulty because there is no local translation readily available. When English-language scientific metaphors (such as 'cycle' in 'water cycle' and 'chain' in 'food chain') are not a familiar part of the concept structure of the indigenous language culture, teachers may experience difficulty if their habit is to provide a translation rather than an explanatory discussion. When a University of Nairobi linguist and native speaker of Kikuyu was asked to give a Kikuyu translation of an examination passage on the water cycle, his first words were, 'I wish you would choose something else, because we don't talk about that in Kikuyu.' Understanding science, in particular, often involves understanding a set of interrelated concepts that requires a vocabulary for talking about the interrelationships. When the English terms of the required vocabulary put together semantic features of the domain in a way that differs from local expression, the teacher will find difficulty in making direct translations. The English word 'air', for example, has no semantically congruent equivalent in Luo. Translations must choose from the Dholuo words muya, which means approximately 'wind', and yamo, which means approximately 'breath'. A translation approach to teaching curricular objectives like 'air is a gas' and 'air is everywhere' can lead to teachers uttering semantic anomalies like 'wind is a 'gas'' and 'breath is everywhere'. Not surprisingly, we observed a Luo-speaking teacher in Standard 3 insert the English phrase 'air is a gas' with no attempt to translate (or explain). (See also Abagi, Cleghorn & Merritt, 1988.) Similarly, in discussing the concept of erosion, the teacher we observed in School 3 never translated the English word 'land', perhaps because the closest Luo words (meaning roughly 'soil', 'earth', and 'plot') did not incorporate the topographical aspect included in the meaning of the English word 'land'. Labelling these kinds of difficulties as a 'teachers' language competency' problem per se is a misidentification of the problem. Rather, what would seem to be needed are better strategies for teaching concepts across linguistic and cultural boundaries: (1) Research and workshops are needed to build bridges between indigenous

< previous page

page_112

next page >

< previous page

page_113

next page > Page 113

language expressions and the scientific phenomena being taught in the curriculum, perhaps even re-examining the choice of scientific phenomena for instruction in light of this consideration. Teachers must have a firm foundation of scientific knowledge in order to communicate it successfully. Further, as we have suggested elsewhere (Merritt, Cleghorn & Abagi, 1988), teaching practices that achieve cultural congruence through familiar communicative routines (those with culturally salient patterns of rhythmic repetition) can enhance the teaching of new and unfamiliar conceptsespecially to young students who are just learning English and Kiswahili, and who have immature vocabularies in their own mother-tongues. (2) With complex sets of new concepts that are not part of the indigenous cultural expression, teachers can be encouraged to develop the practice of definition, example, and explanation rather than always relying on translation. For example, to explain 'local market': Example 6 Standard 4 Agriculture, School 2 T: . . . a local market is that type of market which is situated within a village and the kinds of goods sold there are the ones produced locally. For example, maize, beans, potatoes . . .. Now, in these local markets, those people who are found with this type of market here are people within this areapeople from this locality. People don't come from very far . . . those people who buy and sell goods in a local market are people who come from nearby or who are near the market . . . (. . . indicates words not included) end 6 Repetition also has both negative and positive aspects (Merritt, 1992). As demonstrated in an earlier paper (Merritt, Cleghorn & Abagi, 1988) repeated formats for lessons contributes to cultural congruence between the Kenyan primary classroom setting and home settings for many of the students. Repetition is often communicatively positive in helping to establish rhythm and achieve synchrony, and in providing redundancy and extra processing time (for the learner to comprehend and assimilate the message). Repetition is also associated with the communicatively negative, however: repetitively boring because of lack of new information; indicative of the speaker's inability to engage full meaning or express himself or herself articulately; exemplified by rote learning practices that don't inspire creative manipulation of new information. In Cleghorn, Merritt & Abagi (1989) we illustrated how Standard 8 teachers' use of metaphors for elaboration (positive variation) when explaining scientific concepts sometimes resulted in logical or factual inconsistencies (negative variation). This suggests one reason some Kenyan teachers seem to overuse exact repetition in presenting lesson material; by doing so they avoid inconsistency (negative variation). However, this tactic may also instantiate boring, cognitively unengaged classroom participation

< previous page

page_113

next page >

< previous page

page_114

next page > Page 114

(negative repetition). Achieving a desirable balance is what every effective teacher does. In the Kenyan primary classroom, access to another language provides a resource for variation in expression, in many cases the means to restate (repeat) the content of (some portion of) the lesson. The message content that has already been said in one language can be reformulated with the form of another language. Types of Codeswitching Our data suggest four possible types of codeswitching between English and Swahili or mother-tongue. For switching that involved a whole sentence or interactional move we have found two basic categories, Types I and II. Type III codeswitches involve translation or word substitution and Type IV codeswitches involve one or a few words that function as discourse markers or interactional particles. Our interest is not so much in conceptualising a rigid typology of codeswitching as in pointing to kinds of examples that demonstrate the multiple functioning of codeswitching in these classroom settings. All non-English words in the examples are in Swahili unless other-wise noted. Type I consists of reformulation across codes, with no new information and no new instructions. This usually occurs in a regular sequence: from English to Kiswahili to mother-tongue, suggesting a probable intent to make the con-tent progressively more easily understandable. Or to be more insistent that some response on the part of the student(s) is required (see Examples 9 and 10). Example 7 Standard 3, School 2 T: What's the date today? Leo ni tarehe ngapi? (Today is date which/how many?) S1: Twenty-three. end 7 Note that there is no return to the Swahili form, but that the syntax of the answer matches the syntactic formulation of the Swahili. English would be '(the) twenty-third'. Example 8 Standard 3 science, School 2 T: We said that air is strong. Hewa ina nguvu. (Air has strength.) end 8 Example 9 Standard 1, School 2 (Early in the lesson the teacher wants all the students to put away their books and listen.) T(to all Ss): Close the books you have. Shut your books. (Swahili): Funga vitabu kila mtu. (Close books everyone) T (to one child) (Kikuyu): Grace, hinga ibuka. (Grace, shut book) end 9 Example 10 Standard 1, School 2 (Another instance in which the teacher wants all students to follow instructions, here to participate and answer the question that S1 has addressed to the class.)

< previous page

page_114

next page >

< previous page

page_115

next page > Page 115

S1 (to T): Not all are saying. T (to all Ss) (Swahili): Wengine hawasemi. (Some others are not saying.) T (to all Ss) (Kikuyu): Kwimundu utararia guku na aroria inyuothe. (There is somebody who is not speaking here and s/he is all of you) end 10 Example 11 Standard 3, School 2 (science lesson on 'air') T: (at the end of a long set of instructions for writing in their exercise books) . . . So you write there air. We say air has got strength. Air has got /^ __? Ss: Strength. T: Hewa ina nini? Ina / ^__? (Air has what? It has /^ ?) Ss: Nguvu. (Strength) T: Hewa ina nguvu. (Air has strength.) end II Type II consists of codeswitching as the content of the activity or the textual instruction is moving along. Here there is always something new, some progression in the discourse, that is contained in the codeswitched message. It seems that the non-English code is used not so much to clarify what might have been said in English, but rather as a communicative strategy to focus or redirect the attention of the students by changing modality. Example 12 Standard 3 science, School 2 (in the course of an experiment) T: Mnasikia hayo makelele? (Do you (pl) hear those noises?) T: What is giving us that noise? end 12 Example 13 Standard 1, School 1 (in the course of a science lesson in which some seeds are being planted at the front of the room) T: Siweka maji mingi. (Don't put a lot of water) T: Just so the soil is wet. end 13 Example 14 Standard 3 science, School 2 (T has asked them to write in their notebooks) T: Mimi nakuja kuiona. (I am coming to see it.) T: And then I will come see who is writing like a hen. end 14 Example 15 Standard 3 science, School 2 (in the course of an experiment) T: If we throw them you will see how they will fall. Angalia vile vitaanguka. (Look (there) at them (they) will fall). (Note that the English syntax of concordant future tense in the dependent clause of the first sentence parallels that required by Swahili grammar. Note that the second sentence in Swahili starts with a directive verb, angalia 'look'.) end 15 In all the Type II examples Kiswahili is used at the beginning of a behavioural restructuring of the lesson activity. It is as though the teacher chooses to speak in Swahili in order to be sure of getting the students' attention, and then once attention and the next phase are underway, the teacher switches back into English.

< previous page

page_115

next page >

< previous page

page_116

next page > Page 116

In the Type I examples English is used first, after which the content is repeated in Swahili when there seems to be insufficient uptake on the part of the students, or when the teacher wants to reinforce what is being said. It can be argued, especially in the lower grades, that these reformulations help to teach or reinforce the English that is used as well as the lesson content (see also Cleghorn, in press). Type III consists of translation or word substitution within a sentence. Here the pattern seems to ensure that the content of some portion of the lesson is clear. Sometimes this seems to be a technique for teaching or reinforcing the meaning of English words. Example 16 Standard 8, School 2 (speaking of a container) T: mkebe any type of mkebe. (Mkebe is the Swahili word for pot, can, or mug). end 16 Example 17 Standard 1, School 1 (in a science lesson on plants) T: Haya, seeds nisema mbegu (so, seeds I say/call mbegu). For example, we have beans, we have maize . . .. They are called /^ __ (waits for choral fill-in-the-blank response and then says herself) seeds. end 17 Example 18 Standard 8, School 3 (in a science class) T: How do you call tapeworm in Dholuo? Ss: Jofi. end 18 A special case is the use of words which seem to have no equivalent in the language that is otherwise being used. For example, in Standard 3 (all three schools) English 'gas'; in School 3, Standard 5 (agriculture) English 'land'; in School 3, Standard 8 the Kiswahili word 'matumbo' (special food made of intestines). Type IV consists of interactional particles: discourse markers, classroom management routines, and terms of address: Discourse markers: Dholuo:

Koro (roughly 'now then' or 'so then')

Kiswahili:

Haya (roughly 'O.K. then')

English:

O.K.; All right; Now

Classroom management routines: Dholuo:

Kendo ('again')

Kiswahili:

Jaribu ('try') Sauti kubwa ('big voice') Tena ('again') Mwingine ('someone else')

English:

Try Speak loudly Again Someone else

< previous page

page_116

next page >

< previous page

page_117

next page > Page 117

Come on (i.e. participate) Yes? (calling on a volunteer) Terms of address: Dholuo: woud/nyar ('son of/daughter of') Kiswahili:kijana (roughly 'young man') my sister, our brother, my friend, my dear friends, English: etc. Patterns of Multilingualism Our data suggest the following general patterns of usage: 1. English is more formal, Swahili less formal, mother-tongue the least formal. Admonishments are often softened by using an affective term of address shifted to a less formal code (though English softeners also occur) e.g. 'Speak loudly, kijana.' 2. English is the most official and literary modality. Even where mother-tongue is heavily used along with English there is patterned differentiation. For example, words written on the blackboard at the upper levels are almost always in English only, even when Kiswahili or mother-tongue is used orally within the classroom lesson. 3. Linguistic markers that indicate topic shift are very fluid. 'Now', 'O. K.', 'Haya' (Kiswahili 'O.K. then'), 'Koro' (Dholuo 'Now then') all frequently occur. Language shift does not necessarily signify an attempt to translate or 'simplify' meaning. 4. Turn selection devices and classroom routines are also relatively fluid across languages in a similar way. 'Yes?' 'Jaribu' (Kisw. 'Try'). 'Try'. 'Come on'. 'Mwingine' (Kisw. 'another person'). 'Someone else'. 5. When English is medium of instruction, translation or word substitution is sometimes used instead of giving a-proper definition. 6. When Swahili or mother-tongue is medium of instruction an English word may be used when there is not an equivalent word in the language of instruction, especially when the English term is seen as a key term in lesson curriculum (e.g. gas, land) and one that may be tested (in English) later. 7. Codeswitching or language shift often functions as an attention getting or attention focusing device. This may occur when there is a behavioural objective such as having the students shift their focus of attention or change their activity. 8. Shifts from English to Swahili or mother-tongue across sentence boundaries are often summaries, restatements, or reformulations that emphasise the major points in the instructional material. 9. English as subject lessons are often characterised by drills and competitive games (even wrong guesses to the 'hide and guess where' game were treated as wrong test answers), exercises that focus on difficult aspects of English grammar. 10. Swahili and mother-tongue subject lessons, by contrast, use more narra-

< previous page

page_117

next page >

< previous page

page_118

next page > Page 118

tive, especially story-telling (as opposed to story reading), and work more explicitly on vocabulary building. 11. The availability of more than one language is a communicative resource. The expectation that students already know, and are not simply learning, more than one language is evidenced by the fluidity of codeswitching with interactional particles and discourse markers, as well as by the occasional pattern of discourse in which codeswitching is not used to repeat or reformulate what has just been said, but rather to simply continue the content of the discourse. Concluding Remarks Those who are interested in schools and societal adaptation to technological change know that what the teacher does in the classroom is a major determinant of what students learn and how creative they will be in meeting the challenges of their society. How teachers use language is a major part of what they do. In the Kenyan primary classrooms we observed there were four basic determinants of language choice and codeswitching: (1) official school policy, (2) cognitive concerns, (3) classroom management concerns, (4) values and attitudes about the appropriate use of English, Swahili, and mother-tongue in society at large. A general orientation to the bi- or tri-linguality of the school educated Kenyan is evidenced by the fluid use of interactional particles and classroom management routines in any of the three available languages. Codeswitching provides an additional resource for meeting classroom demands. The avail-ability of more than one language as communicative resource can result in different languages being used as differing modalities might be used in monolingual contexts. As we might expect, codeswitching seems most often motivated by cognitive and classroom management factors: usually either (1) a need to focus or regain students' attention, or (2) a need to clarify, enhance, or reinforce lesson material. A consideration of teachers' language behaviour must take into account the role of the teacher as socialising agent. As an adult bi- or tri-lingual member of Kenyan society, these teachers not only teach lessons and inculcate values having to do with conservation of resources; they, perhaps unconsciously, are socialising students into the prevailing accepted patterns of triglossic multilingualism. Socially significant variation between the two or three languages involves not only choice of which language to use but also choice in how each language is used. By and large we have found that more affectively positive genres and routines are used with mother-tongue or Swahili, more formal and less elaborated routines with English, sometimes with Swahili. This, of course, reflects societal attitudes about what we may think of as the 'affective content' of the three language choices. However, a possible educational concern is that for some teachers these socialising tendencies may result in

< previous page

page_118

next page >

< previous page

page_119

next page > Page 119

a cognitively dysfunctional restriction of English during classroom lessonsfor example, (a) by not fully developing metalinguistic devices like word definition (a sometimes needed alternative to translation or word substitution), or (b) by not fully using affectively positive genres such as narrative to develop student interest in the subject matter (insisting on rote formulations geared to phrases from the teacher syllabuses and expected in the KCPE exam). Many teachers did express linguistic insecurity about their own competency in English. Our impressions are that teachers' levels of fluency are actually quite high and more than adequate for primary teaching, especially when lessons are developed in a logical and organised way. One factor in this pervasive linguistic insecurity may be the history of repeatedly testing English grammatical features that are incongruent with grammatical features of the indigenous African languages and for which classroom models (in both teachers' oral speech and written blackboard material) are inconsistent. Another factor in classroom linguistic insecurity may be the misidentification of cross-linguistic conceptual problems as language competency problems (see earlier discussion). A related issue is the teacher's occasional lack of control in subject matter knowledge, especially when dealing with culturally unfamiliar subjects, a condition that is exacerbated in the poorer schools by the scarcity of texts and reference materials. Though the semantic and cultural difficulties in applying English to the African context are very real, it seems that an attitude towards English as being complicated and difficult is often furthered by testing difficult aspects of syntax, which in turn fosters linguistic insecurity on the part of students. Though we have argued that this pattern may derive from natural processes of language socialisation to societal triglossia, it may have undesirable effects on academic learning so long as English is the major medium of instruction. Many Africans have expressed concern over the relationship between language use and cultural identification (Gachukia, 1970; Wanjala, 1980; Omondi, 1983; Ngugi, 1986; Rubagumya, 1990; Eisemon, Hallett & Maundu, 1986; Mazrui, 1975; The British Council, 1986), and many advocate a more central role for indigenous African languages. Attitudes and beliefs about the appropriate place of English, Swahili, and mother-tongue within Kenyan society vary by individual, social class, and region. Shared concern about intergenerational communication and appreciation of mother-tongues might be addressed by curriculum changes that allow separate enrichment in this area, or by a major reconsideration of the overall goals of the educational language policy. Research that leads to better understanding of the processes, and programs that encourage general language aware-ness (see Martin-Jones, 1989) would seem to be helpful. We emphasise that our study is based on observations of only three Kenyan primary schools, and that any generalisation to the whole of the Kenyan primary educational system should be carefully considered. At the same time we feel that the issues and parameters, the directionality of

< previous page

page_119

next page >

< previous page

page_120

next page > Page 120

concerns, are comparable, not only to the broader situation in Kenya, but also to similar settings in other multilingual countries. We hope that the questions we have raised will interest practitioners and policymakers in education, as well as others who work with social change and human development. Notes 1. This paper grows out of research funded by the Social Science and Humanities Research Council of Canada, carried out in Kenya in 1986/87 under the direction of the principal investigator, Ailie Cleghorn. Marilyn Merritt served as senior associate and project linguist; Jared Abagi was full time research assistant and Grace Bunyi was part-time research assistant for School 2. Portions of the analysis have been presented by Marilyn Merritt at meetings in Kisumu, Kenya (1987), Montreal (1988), Washington, D.C. (1990), and New York (1991). All correspondence should be directed to Marilyn Merritt at 2706 Key Boulevard, Arlington, Virginia, 22201, USA. The authors gratefully acknowledge the support of George Eshiwani and the Kenya Bureau of Educational Research at Kenyatta University, as well as comments and observations from Mohamed Abdulaziz, Thomas Eisemon, Karega Mutahi, Judith Nyagah, Anna Obura, Okoth Okombo, and Lucia Omondi. 2. For what is probably the earliest invocation and analysis of a notion of 'linguistic insecurity' see Labov, 1966 (1982): Chapter XII. References Abagi, Jared O., Cleghorn, Ailie, and Merritt, Marilyn (1988) Teachers' language use in standard III lessons on 'air' in urban and rural Kenyan schools. Kenya Journal of Education. Abdulaziz, M. H. Mkilifi (1972) Triglossia and Swahili-English bilingualism in Tanzania. Language in Society 1, 197-213. Cazden, Courtney (1988) Classroom Discourse. New Hampshire: Heinemann. Cleghorn, Ailie (in press) Primary level science in Kenya: constructing meaning through English and indigenous languages. International Journal of Qualitative Studies in Education. Cleghorn, Ailie, Merritt, Marilyn and Abagi, Jared O. (1989) Language policy and science instruction in Kenyan primary schools. Comparative Education Review 33, 1, 21-39. Cook-Gumperz, Jenny (ed.) (1986) The Social Construction of Literacy. London: Cambridge University Press. Eastman, Carol (1983) Language Planning, an Introduction. San Francisco: Chandler and Sharp. Eisemon, Thomas O. (1991) Language issues in scientific training and research in developing countries. Washington, D.C.: World Bank report. Eisemon, Thomas O., Hallett, Martin and Maundu, John (1986) Primary school literature and folk tales in Keyna: What makes a children's story African? Comparative Education Review, 10, 2, 220-40. Gachukia, Eddah (1970) The teaching of vernacular languages in Kenya primary schools. In T. P. Gorman (ed.) Language in Education in Eastern Africa, 18-24. Nairobi: Oxford University Press. Gorman, T. P. (ed.) (1970) Language in Education in Eastern Africa. Nairobi: Oxford University Press. Heath, Shirley Brice (1984) Ways with Words. Cambridge University Press. Hoggart, Richard (1957) The Uses of Literacy. London: Chatto & Windus, also Pelican. Labov, William (1966, 1982) The Social Stratification of English in New York City. Washington D.C: Center for Applied Linguistics. Martin-Jones, Marilyn (1988) Codeswitching in the classroom: A review of research in bilingual education programmes. Lancaster: Center for Language in Social Life Working Paper Series No. 22.

< previous page

page_120

next page >

< previous page

page_121

next page > Page 121

(1989) Language education in the context of linguistic diversity: Differing orientations in educational policy making in Britain. In J. Esling (ed.) Multicultural Education and Policy: ESL in the 1990s, 36-58. Toronto: OISE Press. Mazrui, Ali A. (1975) The Political Sociology of the English Language: An African Perspective. The Hague: Mouton. Merritt, Marilyn (1982a) Repeats and reformulations in primary classrooms as windows on the nature of talk engagement. Discourse Processes 5, 127-45. (1982b) Distributing and directing attention in primary classrooms. In Louise Cherry Wilkinson (ed.) Communicating in the Classroom, 223-44. New York: Academic. (1992) Repetition in situated discourseexploring its forms and functions. In Barbara Johnstone (ed.) Repetition in Discourse, Vol I. N.J.: Ablex. Merritt, Marilyn and Abdulaziz, Mohamed H. (1988) Swahili as a national language in East Africa. In Florian Coulmas (ed.) (pp. 48-67) With Forked Tongues: What are National Languages Good For? Ann Arbor: Karoma. Merritt, Marilyn, Cleghorn, Ailie and Abagi, Jared O. (1988) 'Dual translation' and cultural congruence: Exemplary practices with English, Swahili, and mother-tongue in three Kenyan primary schools. In Kathleen Ferrara, Becky Brown, Keith Walters and John Baugh (eds) Language Change and Contact, 232-39. Austin: Texas Linguistic Forum, Vol 30. Merritt, Marilyn and Humphrey, Frank M. (1979) Teacher, talk and task. Theory into Practice, XVIII, 4, 298-303. (1980) Service-like Events During Individual Work Time and their Contribution to the Nature of Communication in Primary Classrooms. Final Report for Grant No. NIE G-78-0081 of the National Institute of Education. Available through ERIC Clearinghouse (248 pages ) # ED196277 FL01209. Mwanzi, Helen O. A. (1984) Language policy in primary school in Kenya. KERA Research Report No. 1.5. Nairobi: Kenyatta University Bureau of Educational Research. Ngugi wa Thiongo (1986) Decolonising the Mind: The Politics of Language in African Literature. Nairobi: Heinemann. Omondi, L. N. (1983) Language handicap in everyday life. Journal of Eastern African Research and Development 13, 104-25. Proceedings, Conference on English in East Africa, 24-27 March 1986. Nairobi: The British Council. Rubagumya, Casmir M. (1990) Language, social values and inequality in Tanzania: Reinterpreting triglossia. Lancaster: Center for Language in Social Life Working Paper Series No. 26. Schieffelin, B. and Ochs, E. (eds) (1986) Language Socialization Across Cultures. Cambridge: Cambridge University Press. Scotton, Carol Myers (1988) Differentiating borrowing and codeswitching. In K. Ferrara, B. Brown, K. Walters and J. Baugh (eds) Language Change and Contact. Austin: Texas Linguistic Forum, Vol 30. Scotton, C. and Ury, W. (1977) Bilingual strategies: The social functions of code-switching. Linguistics 193, 5-20. Syllabuses for Kenya Primary Schools (1986) Volume II. Nairobi: Ministry of Education, Science and Technology. Printed by Jomo Kenyatta Foundation. Wanjala, Chris (1980) For Home and Freedom. Nairobi: Kenya Literature Bureau. Whiteley, Wilfred (ed.) (1971) Language Use and Social Change: Problems of Multilingualism with Special Reference to Eastern Africa. London: Oxford University Press. (ed.) (1974) Language in Kenya. London: Oxford University Press.

< previous page

page_121

next page >

< previous page

page_122

next page > Page 122

< previous page

page_122

next page >

< previous page

page_123

next page > Page 123

The Politics of Codeswitching and Language Choice Monica Heller Ontario Institute for Studies in Education, 252 Bloor Street West, Toronto, Ontario M5S 1V6 Canada. Abstract Using data from ethnographic studies of the use of French and English in Ontario and Quebec (Canada) in a variety of settings (hospital, factory, school, etc.) over the period 1978-1990, I discuss language choice as a political strategy, especially as a strategy of ethnic mobilisation. More broadly, I argue that codeswitching must be under-stood in terms of individual communicative repertoires and community speech economies, particularly as these are tied to a political economic analysis of the relationship between the availability and use of linguistic varieties, on the one hand, and the production and distribution of symbolic and material resources on the other. Introduction 1 The purpose of this paper is to discuss language choice as a political strategy, in particular its role in processes of ethnic mobilisation. The major point I want to make is that the study of the distribution and use of language choices in multilingual communities (choices which include but are not limited to codeswitching) can reveal not only the extent of stability of intergroup relations, but, perhaps more importantly, it can reveal the ways in which the regulation of access to symbolic resources is tied to the regulation of access to material ones. In a given setting, at a given historical moment, codeswitching may be conventional, or, on the contrary, anti-conventional. In other words, it may represent a normal, routine way to use language, or it may violate expectations about how to behave. In any case, it represents part of a range of linguistic resources upon which people can draw to define the value of the other resources they control and to regulate access to them: resources are distributed by specific groups in specific situations through the provision and evaluation, among other things, of symbolic, including verbal, performances. Conventional language practices represent relatively stable relations of power, while violations can be seen as forms of resistance.

< previous page

page_123

next page >

< previous page

page_124

next page > Page 124

Such conventions are created, maintained or changed through social interaction; by examining such processes, we can understand how (and possibly under what conditions) groups mobilise to alter prevailing modes of regulation of distribution of valued resources, and, indeed, may contribute to the alteration of prevailing modes of assigning value to specific forms of symbolic and material resources. In other words, we can discover how dominant groups control who can participate in situations where important things happen and how they control what is to count as important; in addition, we can discover how subordinated groups acquiesce in or in different ways resist these processes. Methodologically, this implies examining codeswitching not in isolation, but as part of a range of linguistic practices which people create and deploy to accomplish social goals. The absence of codeswitching can be as significant as its presence, or as the various linguistic forms which it may take. In order to support this argument, I will discuss material from my own work on French-English codeswitching in a variety of settings in Quebec and Ontario (Canada). 2 This example is designed to stand as an illustration, rather than as a model for universal processes of development and change in practices of codeswitching and language choice, since a crucial assumption here is that the processes in question, while they may possess common elements, are historically contingent. Further, any generalisations can only be based on comparative research across time and space (Gal, 1988); I am limited here to comparisons within a few regions of Canada and with relatively shallow historical depth. Broader generalisations require broader bases of comparison, and I hope that the framework I offer here might be useful for just such future undertakings. Codeswitching, Resources, Repertoires and Ideology The picture I want to draw here hinges on a notion of codeswitching as a means of drawing on symbolic resources and deploying them in order to gain or deny access to other resources, symbolic or material. This picture builds both on Bourdieu's concepts of symbolic capital and symbolic market-places, and Gumperz' concepts of speech economies and verbal repertoires (Bourdieu, 1977, 1982; Gumperz, 1982). I will use Bourdieu's concepts in the following way. I take codeswitching as a means of calling into play specific forms of linguistic and cultural knowledge, forms which conventionally possess certain kinds of value. That value is linked to the extent to which those forms facilitate access to situations where other kinds of symbolic and material resources are distributed, resources which themselves have value based in the prevailing modes of organisation of social life in the community (and who controls them). Certainly some resources have a concrete, functional basis to their value (like food); but most are related in more indirect ways to the methods people have of not only acquiring the basic things they need to survive, but of also

< previous page

page_124

next page >

< previous page

page_125

next page > Page 125

acquiring various forms of power and solidarity (in the Weberian sense of acquiring the means to mobilise and allocate resources). Groups which control valued resources (of whatever kind) also control the 'marketplace' (in Bourdieu's terms) in which they are exchanged, the set of social relations in which the value of resources is defined and resources themselves are exchanged. Beyond sheer force, such marketplaces operate through hegemonic practices, through symbolic domination, through convincing participants that the values and modes of operation of the market-place are immutable and universal. To change metaphors slightly, specific groups set the rules of the game by which resources can be distributed. In other words, it is necessary to display appropriate linguistic and cultural knowledge in order to gain access to the game, and playing it well requires in turn mastery of the kinds of linguistic and cultural knowledge which constitute its rules. Buying into the game means buying into the rules, it means accepting them as routine, as normal, indeed as universal, rather than as conventions set up by dominant groups in order to place themselves in the privileged position of regulating access to the resources they control. Bourdieu has insisted over and over again that it is precisely through appearing not to wield power that dominant groups wield it most effectively (Bourdieu, 1982; Gal, 1989). These notions tie into those of Gumperz in a number of ways. First, both have noticed that linguistic and cultural capital are not equally distributed in any given community, despite the fact that all members of the community might share (at least along some dimensions) the same scale of values, that is, they all might agree on the fact that it is the capital (and other resources associated with it) concentrated in the hands of one group that is what is really valuable in life. In Gumperz' terms, forms of language are distributed unequally across a speech community, that is, any individual member will have a verbal repertoire which draws on part, but rarely all, of the forms in circulation. Further, it is this unequal distribution and the way in which resources are deployed which drives the operation of the marketplace, and hence the reproduction of relations of power. Only some members of a population are in a position to decide what will count as appropriate behaviour in situations where resources are distributed, and to evaluate performances there; normally, it is the symbolic capital dominant groups already possess which is the key to participation and success in the situations they control. As Gumperz and others have shown, an inability to bring to bear appropriate conventions of behaviour on key situations in daily life where crucial decisions about one's access to resources are decided (a job interview, an exam, a courtroom trial, etc.) can result in the systematic exclusion of segments of the population from the resources distributed there (cf. also Erickson & Shultz, 1982; Heath, 1983; Michaels, 1981). Many critics have pointed out that this thoeretical framework is helpful in exploring how relations of power are sustained, but not very helpful in exploring resistance and change. First, it assumes a form of collusion on

< previous page

page_125

next page >

< previous page

page_126

next page > Page 126

the part of subordinate groups in their own subordination, whether conscious or not; it does not allow us to explore ways in which subordinate groups may oppose hegemony, whether directly or through the setting up of alternative systems of value, alternative marketplaces (Williams, 1976; Woolard, 1985). Second, it limits our ability to locate the creative power of individuals in social interaction to define and redefine social relations through communication (Shirley, 1986; Giddens, 1984). Put more broadly, the framework makes it difficult not only to discover where symbolic capital comes from, but also to account for (and hence to influence) processes of change. 3 Various researchers have taken up the challenge of responding to these critiques. Some, notably Mehan and Cicourel, explore these dimensions in institutional settings (Mehan, 1983, 1987; Mehan et al., 1986; Cicourel, 1975, 1978, 1980; cf. also Collins, 1987, 1988; Davis, 1988). Others focus on ethnic boundaries in the distribution of resources (cf. e.g. Woolard, 1985, 1989; Gal, 1988; Hill, 1985). The approach oulined here is situated in this second line of research, and takes codeswitching as a point of entry into the exploration of processes whereby dominant groups use conventions of language choice to maintain relations of power, while subordinate groups may (at times simultaneously) acquiesce to or resist them, and may even exploit conventions of language choice to redefine them (whether by substituting one group for another in the hierarchy, or by more radically altering the value of the symbolic and material resources in the marketplace). In what follows, I will describe first the ways in which language was historically involved in the establishment of anglophone dominance in Canada. This will serve as a backdrop to the discussion of the ways in which linguistic practices, codeswitching among them, then became available to some as a means of altering (or coping with altered) relations of dominance in order to achieve particular social and political goals, specifically, participation in, acceptance of or resistance to the collective mobilisation of francophones to gain access to the valued resources hitherto the privileged reserve of the anglophone élite. French and English in Quebec and Ontario (Conquest to the 1960s) We possess limited knowledge about the social distribution of bilingualism in Canada, particularly in the period before the 1960s, when other matters, such as religion, obsessed us more than the language questions which preoccupy us today, or when tensions which we now think of as language questions were thought of more as matters of nation or race. There is, of course, no consensus among historians regarding the exact nature of language contact over the last 250 years or so, but a general picture emerges of two groups aligned in shifting, but basically unequal, relations of political and economic power, the result of the British conquest of New France.4 Points of contact were limited, confined probably to the francophone élite

< previous page

page_126

next page >

< previous page

page_127

next page > Page 127

and to those who organised the participation of francophones in British-run economic activities (as coureurs des bois in the fur trade, or later as loggers and draveurs in the lumber industry); possibly a small urban commercial class may have been involved in contact as well, especially in Montreal, where the majority of English-speakers were concentrated (indeed, until the end of the nineteenth century, Montreal's population was predominantly anglophone; Lachapelle & Henripin, 1980). While some anglophones settled in rural areas (mainly in the wake of the American Revolution and the War of 1812), the nature and conditions of that settlement precluded much contact; whatever contact there may have been most likely involved the use by francophones of English (Cartwright, 1987). In the nineteenth century, and then into the first part of the twentieth, francophones, originally predominantly involved in agriculture and forestry, became increasingly involved as labour in English-controlled manufacturing activities, or in other primary-resource extraction and transport industries (notably railway-building and mining). The few literary and sociological treatments of intergroup relations in this period, while focussing on other matters, give the impression that social stratification continued to limit actual contact between individual members of each group. Small businessmen and foremen in factories were probably the most likely to act as brokers, and hence to be bilingual, along with service personnel (shopgirls and maids, for example); most of these were probably drawn from among the francophone population, although the Irish working class, and later the Jewish commercial and service class, were probably also involved (MacLennan, 1945; Hughes, 1943). Most typically, francophones had to learn English when they entered the workforce in order to interact with anglophone superiors, and even then, this was most important only for those who occupied positions at the interface between the groups. This picture is confirmed by a demographic study of bilingualism in Montreal in the early 1960s, which showed that FrenchEnglish bilingualism was confined to francophone males between the ages of 18 and 60, that is, only those members of the subordinate group who were active in the commercial/industrial, anglophone-owned and run, workforce (Lieberson, 1965). The prevailing convention of language choice, then, was to use English in interaction with anglophones, at least anglophone members of the ruling élite. Anglophone managers in factories, and their wives running households staffed by francophones at home, probably spoke some French, of a syntactically, morphologically and lexically limited variety. Anglophone (principally Irish)-French interaction within the working class was undoubtedly quite different; the high rate of intermarriage, and the existence today of people with English or Irish names who speak no English and French names who speak no French attest to a different, but largely undocumented, dynamic (cf., however, Fennario, 1980 for a literary treatment of French-English workingclass relations in Montreal at that time). 5 Any codeswitching would

< previous page

page_127

next page >

< previous page

page_128

next page > Page 128

have had to have been here, in interaction between Irish and French, or among francophone bilinguals in the workforce (although not necessarily in the workplace). Somewhere, in this distribution of linguistic proficiency and linguistic practice, as it is linked to participation in and control over economic (and political) activity, lies one key to understanding the dynamic of the maintenance of anglophone domination of francophones until the 1960s. What is certainly clear is the ability of anglophones to dominate through the establishment of a hierarchy of language values in which English was valued over French (and European French over Canadian French), and through maintenance of the hegemonic norm of English as the language of private enterprise and of public (including but not limited to intergroup) communication. These conventions of language use and language value were indexical of an ideology of ethnic relations in which 'national' character suited each group to its place in the economic and social order. My purpose here, however, is not so much to explore that dynamic (which would require in any case empirical data which I do not possess), but rather to use it as a backdrop to understanding its destabilisation through francophone ethnic (national) mobilisation from the 1960s through to today. In the rest of this paper, I will explore, on the basis of the (admittedly limited) evidence available, the relationship between the speech economy of language contact in Quebec and Ontario, its manifestation in individual verbal repertoires, and the redefinition of distribution of access to symbolic and material resources through ethnic mobilisation. In particular, I will focus on the distribution and functioning of codeswitching as a means of reconstituting social relations of power, specifically through a re-evaluation of the value of the symbolic resources which are the French and English languages (more precisely, the various varieties of those languages in circulation) as linked to their role in regulating access to other valued symbolic and material resources. Codeswitching and Francophone Ethnic Mobilisation A variety of economic and political processes, too complex to enter into here, underlay the beginnings of francophone nationalist mobilisation in the 1960s. 6 While Canadian francophone nationalism has a long history, the most recent manifestation of it is distinguished by its major focus: that of gaining francophone access to the economic resources hitherto controlled by anglophones, without sacrificing francophone identity. In other words, while through the 1950s and 1960s (and even to some extent, in some places, today) francophones who wished to gain access to management positions in private enterprise had to do so through assimilation, the 1960s saw the beginings of a collective mobilisation designed primarily to achieve that access for the group as a whole, and used both a sense of collective identity and evidence of collective oppression to achieve that mobilisation. Of course,

< previous page

page_128

next page >

< previous page

page_129

next page > Page 129

many nationalists have contributed to that movement for other reasons, with goals that have more to do with social transformation than with upward mobility within the current structure of society, but their goals have been largely eclipsed by the economic and political benefits for an emerging francophone private and public sector élite of the mobilisation processes of the last 30 years. Economic and political conditions have been key in creating the possibility of francophone nationalist mobilisation (notably the expansion of the industrial base of Quebec and Ontario in the post-War period, a shift in focus among the anglophone élite from Montreal to Toronto, and the existence of a state apparatus at the provincial level controlled by the numerically-dominant francophones). However, the realisation of that mobilisation in the experience of individuals in their everyday life can be seen as having been equally accomplished through their ability to manipulate linguistic resources in order to alter the rules of the game. Two dimensions of the political significance of codeswitching need to be addressed. The first has to do with its distribution in the community: who has access to what kinds of linguistic resources? The second has to do with the way in which people who are in a position to do so exploit those resources for the accomplishment of a variety of social goals. The following discussion is drawn from data that I have been able to collect in an out-patient clinic and in a large company in Montreal over the period 1977-1979, in French-language minority schools in Toronto (Ontario) over the period 1983-1990, and among Franco-Ontarian women married to anglophone men and living in three different regions of Ontario in 1989. Political mobilisation of francophones really began in Quebec, and only later began to have an impact outside the province. The period of my fieldwork in Montreal was particularly interesting, since a nationalist government had been elected in Quebec in 1976. Among their initiatives was the Charter of the French Language (commonly known as Bill 101), enacted in 1977 with the intent of promoting the French language as an instrument of mobilisation, an instrument for redefining the system and locus of distribution of economic and social resources. Among its many provisions, it made French the language of work, in the private as well as in the public sector. My work in Ontario has allowed me to focus on an area which is bound up with the development of Québécois nationalist ideology in a number of complex ways. Some Ontarians are isolated from it; others are caught up in it, but must re-orient themselves within a new framework. While in the past francophone nationalism embraced francophones across Canada, the association of Québécois nationalism with a specific territory (more or less) contiguous with the current borders of the province of Quebec means that a distinction must now be made between francophones inside and those outside Quebec. In addition, the economic developments of the last 15 years have drawn increasing numbers of francophones from Quebec to Ontario. What I possess, then, are in-depth views of specific social locales (Marcus,

< previous page

page_129

next page >

< previous page

page_130

next page > Page 130

1986), distributed across time and space, but linked in a variety of ways within broader social processes (Giddens, 1984). Comparison of language practices across those locales can therefore tell us something about the processes in question. Indeed, several patterns stand out both in terms of the distribution of codeswitching and in terms of its use. These patterns are directly concerned with the extent to which the individuals involved are in contact with both francophones and anglophones, the nature of that involvement, and most importantly the political response of individuals to their relationship to the language boundary. What counts most of all is the extent to which individuals can (because of the verbal repertoire they develop by virtue of their social position) exploit linguistic resources for fulfilling social aims, and, secondly, whether they see their interests as lying primarily in strengthening or rather in neutralising the boundary between the groups. The patterns will be described in terms, first, of those who do not codeswitch, because they do not have access to the necessary linguistic resources or because their overriding concern is for the maintenance of the ethnic boundary, even if they do have access to the necessary linguistic resources. Secondly, I will discuss those who not only have the right kind of verbal repertoire, but find that it is in their interest to exploit it (through codeswitching, notably) in order to cross the boundary, level it, or neutralise the tension across it. I will focus mainly on the presence or absence of codeswitching, and less on the more subtle dimensions of variability across that continuum, although some mention will be made of different forms of codeswitching encountered, and the ways in which that formal variation can be understood both in terms of individual repertoires and political ideologies. Non-Codeswitchers Two kinds of groups (whose membership may overlap) generally fail to codeswitch. The first kind consists of people who are too far from the language boundary to possess the resources (and the motivations) upon which codeswitching is based. Many of these are anglophones, especially those in Ontario, but there remain francophones both in Quebec and in Ontario whose contact with English is limited or nonexistent. This is true not only of rural populations and certain segments of the working class, as in the past, but, in Quebec at least, also of the increasing numbers of francophones who work in private and public sector enterprises and organisations now owned or at least controlled by francophones and where the language of work is French. It can also be true of people (so far, generally francophones) who are assimilated to such an extent that over the course of their lives they have removed themselves far enough from the boundary so as to make codeswitching impossible and in any case meaningless. The second kind consists of people who are caught up in processes of mobilisation, whether to participate in it or to resist it. Whereas many of them possess bilingual verbal repertoires, their political strategy entails a

< previous page

page_130

next page >

< previous page

page_131

next page > Page 131

reinforcement, not a levelling, of the linguistic boundary. (Indeed, for many, it is precisely their experience of the boundary which underlies both their bilingualism and their mobilisation.) In order to challenge the dominance of English in Canada, it was felt to be necessary to challenge prevailing conventions of language choice, more specifically, to violate them. Rather than speaking English, then, where expected, it became necessary to speak French (and only French). As an explicit strategy, this was current among mobilising francophones in Quebec in the 1970s, and remains current among mobilising francophones in Ontario today. In our recent interviews of Franco-Ontarian women, several talked about how they do this. One woman, Louise, gave this example: 7 Example 1 (Ottawa 1989) . . . dans les magasins . . . je fais ma naïve jusqu'au dernier degré 'je ne sais pas l'anglais moi' j'ai pour dire à Orléans à Ottawa tu te fais servir en français [oui c'est vrai] point final [tu peux] c'est moi qui perds du temps je veux dire je perds énormément de temps parce que je là il faut qu'ils aillent me chercher quelqu'un que là je fais ma naïve je vais en tout cas si cela ne fait pas je vais protester 'je veux me faire servir en français' l'épicerie ici e c'est supposé être bilingue tu sais puis quelquefois il y en a qui ne le sont pas puis ils sont insultés parce que je là je me rends justqu'à la direction je leur dis 'ça me prend quelqu'un bilingue' pour {pause} ils sont supposés d'être bilingues tu sais ils sont supposés [mhum] d'avoir quelqu'un lô toujours une qui parle tu sais je perds énormément de temps . . . (in stores . . . I act naive to the utmost 'I don't speak English' to say that in Orelans in Ottawa you can be served in French [yes it's true] full stop [you can] it's me who's wasting time I mean I waste enormous amounts of time because I then they have to find me someone I act naive I'll anyway if that doesn't work I'll protest 'I want to be served in French' the grocery here e is supposed to be bilingual you know and sometimes there are some who are not and they're insulted because I then I go all the way to the management I tell them 'I need someone who's bilingual' to {pause} they're supposed to be bilingual you know they're supposed to [mhum] have someone there at all times who speaks you know I waste an enormous amount of time) But it is important to note a few things about Louise. She lives in Ottawa, a city right on the language border, and one where, as a direct result of francophone mobilisation, speaking French can be a major key to upward social mobility, through the federal civil service (Ottawa is the federal capital) or through service industries spinning off the civil service. Indeed, Louise's husband is in the Canadian Armed Forces, and although Louise herself was not working at the time of interview, she was accustomed to working periodically as a teacher of French (as well as sometimes of English, depending on the part of the country in which they are stationed). For Louise, and others like her with whom we spoke, French has become valuable, and the source of that value has to do with the creation of resources

< previous page

page_131

next page >

< previous page

page_132

next page > Page 132

which francophones exclusively control. It is in her interest to make sure the boundary is maintained. It is clearly not that Louise can't speak English, it is that it is not always in her interests to do so. Indeed, she also signals this symbolically in her language choice within the interview: she displays her stance by using only French throughout, except for a few, highly flagged occurrences limited to proper names or fixed expressions. On the other side of this mobilisation effort are anglophones whose privileged access to and control over resources are being threatened. They react in two major ways. The first set of strategies involves resistance to mobilisation through flight from boundary disputes or through attempts at reinforcement of existing boundaries. In the brewery in which I did field-work in Montreal, it was clear that some anglophones reacted by removing themselves from the boundary, whether by moving to anglophone areas of the country or by isolating themselves within the company. The most poignant example of this was a man in his thirties whose geographical mobility was restricted, largely because he did not hold a management-level job. Instead, he found himself a job within the brewery where he could work alone in a cubicle in the basement or make the rounds of the equipment and rarely talk to anyone. In other cases, the reaction is one of passive or even overt resistance, for example, speaking English where others would now insist on French, and, in extreme cases, the organisation of English-rights movements (such as the Association for the Preservation of English in Canada, or the Confederation of Regions party). In Example 2, a routine exchange is rendered problematic by such political concerns. The hospital where this exchange took place is in downtown Montreal; as with almost all social institutions in that city it is historically affiliated with one of Montreal's two major language groups, in this case English. Whatever the first language of the actual participants in exchanges there (and when I worked there in the mid-1970s most lowranking staff and patients were in fact non-anglophone), the convention of language choice had always been English. When francophone mobilisation began to call this convention into question, staff members were presented with a problem. We saw ourselves as being at the service of the public, and tried to find ways to engage in interactions without incurring the wrath of patients through inappropriate language choice. Since it is impossible to tell at a glance whether someone speaks French or English (let alone what their political position on the issue of language choice might be), we had to find ways around our dilemma. One was to codeswitch (see Heller, 1982a); another was to look at the name on the hospital card patients presented to us, and guess. As the example demonstrates, however, this did not always work smoothly; the patient's silence and demeanour signalled that a reanalysis of the political frame of reference was in order: Example 2 (Montreal 1977; out-patient clinic): Clerk: Lombard, Anne-Marie?

< previous page

page_132

next page >

< previous page

page_133

next page > Page 133

Patient: (Silence, Glare.) Clerk: c'est bien ça votre nom? (That's correct isn't it, that's your name?) Patient: (Silence. Glare.) Clerk: is this your name? Patient: yes. The next example shows more overt forms of resistance. The exchange took place in Montreal in 1978, in the reception area of the provincial government office responsible for implementation of the controversial Bill 101. One of the provisions of the law was that members of certain professions had to demonstrate proficiency in French in order to be able to exercise their profession. One way in which this was evaluated was through government-designed and administered language tests. The man who comes up to the receptionist is, presumably, there in order to take one of these tests. Example 3 (Montreal 1978; provincial government office): Man: Could you tell me where the French test is? Receptionist: pardon? Man: Could you tell me where the French test is? Receptionist: en français? Man: I have the right to be addressed in English by the government of Quebec according to Bill 101 Receptionist (to a third person): Qu'est-ce qu'il dit? (what's he saying?) Codeswitchers The second broad set of strategies are ones which are designed to enable individuals to cross or level the boundary. Some anglophones attempt to gain access to the resources controlled by francophones (and distributed through the use of French) by learning French. These are the parents who line up all night to place their children in French immersion programmes, who themselves spend weeks in intensive language study in Quebec, or who learn French the way francophones used to learn English: on the job. For the most part, these are members of the middle class. They are the ones who have held the jobs in which French has now become important, and to which they do not want to lose access. They also have the material resources and the cultural knowledge necessary to enable them to acquire French, this new form of symbolic capital which has changed its value in the marketplace and its place there. 8 In the brewery, this was the reaction of some of the anglophone managers, who, for a variety of reasons, preferred to stay and adapt rather than fight or flee. While their proficiency in French may have been poor, they deployed it strategically and symbolically (mainly through the use of French greeting routines and other fixed expressions) in situations in which they wanted to be able to participate and with francophones with whom they wanted to maintain good relations. They were joined in this strategy by a small group

< previous page

page_133

next page >

< previous page

page_134

next page > Page 134

of younger employees who had more to gain and less to lose by becoming bilingual, and, indeed, whose bilingual proficiency and use of codeswitching exceeded those of older anglophones. They were also joined by some of the young and newly-arrived francophone managers, in some situations. While the position of these young francophone managers depended for its legitimacy on the exclusive use of French, they were frequently prepared to use some English in order to include individual anglophones in interactions. They were motivated in this in part by a need to gain access to the anglophones' expertise (based, usually, on years of experience), and in part by interpersonal considerations: there are, just the same, often limits to the extent to which people are prepared to allow group conflict to influence interpersonal relations. It sometimes makes more sense to maintain contradictory behaviour than to live with the consequences of being consistent. In the following example, Albert and Bob are taking part in a regular weekly meeting of one sector of the brewery. The sector manager and management staff had always been mainly English-speaking, and the meetings had always occurred in English. The one francophone who had worked there for some time used only English with his colleagues. About six months before this exchange took place, this situation changed in a number of ways. Most importantly for our purposes, the sector manager received a transfer to Ontario. Bob was the next-highest ranking member of staff, and the one with the most seniority. He was, however, passed over for promotion in favour of Albert, a young, educationally-qualified francophone. While this situation certainly contained all the seeds of conflict, Albert and Bob adopted a different approach, and tried to work together in a context of mutual respect. One important way in which they constructed this context was through strategic use of codeswitching, often framed as light-hearted joshing (despite the seriousness of the content of what was said): Example 4 (Montreal 1979; management meeting, manufacturing company): Albert: Uh it's like passing the buck the somebody but uh (laughs) can you spend some time with Pierre {xx} Monday it could be a good thing. Bob: avec plaisir . . . okay I'll do that uh I charge Anne rien but spécial pour toi forty-five dollars an hour. In my data, two other groups not only had access to bilingualism but also used it in this boundary-levelling way. The first group consists of some of the Franco-Ontarian women married to anglophone men whom we inter-viewed in 1989. Unlike Louise, these women tended to live in parts of the province where there were fewer white-collar jobs in which French was valued, or to not possess the qualifications necessary for them to gain access to those jobs even where they existed. They were, however, conscious of the importance of French for their children's future, and, while English dominated their lives in their nuclear families and in the workplace, French

< previous page

page_134

next page >

< previous page

page_135

next page > Page 135

remained an important link to the emotional and material support provided by their extended families of origin, and to whom they retained close links. Within that francophone world, conventional linguistic practice was defined as bilingue; the normal way to talk among francophones is to mix (mélanger) languages, and the normal way to identify members of the group is as bilingues. As one woman says of talk in gatherings of her birth family: Example 5 (Ottawa 1989) . . . on parle français c'est comme bilingue tsé mélangé (we speak French it's bilingual y'know mixed) This practice is also reflected in the discourse of the interview, in which English is used much more extensively, and in a more smoothly integrated fashion, than it is in interviews with women like Louise, Further, English is used strategically in the interviews to symbolically signal a double affiliation. Annette says: Example 6 (Ottawa 1989) . . . même si ça change c'est tout anglais je serai toujours française comme {pause} that's me (even if it changes it's all English I'll always be French like {pause} that's me) Julie says: Example 7 (Sudbury 1989) je suis une Canadienne-française I guess (I'm a French-Canadian I guess) Later, talking about the fear among monolingual anglophones that all government jobs will be bilingual, she says: Example 8 (Sudbury 1989) c'est ça qui choque tout le monde asteur it's creating hard feelings avec les Anglais pis j'aime pas ça because moi j'en connais trop des Anglais . . . (that's what's upsetting everyone these days it's creating hard feelings with the English and I don't like that because I know too many English people) Codeswitching allows them to participate in both French and English worlds, and, in the French networks, it is a sign of that ability to participate in activities controlled by anglophones. It is the codeswitching of the bon vieux style, with a twist: in Ontario, more so than in Quebec, social change has opened up opportunities for assimilation to English. Many have gone that route. But some have become aware of the importance for their children of retaining French for purposes of upward mobility; the value they place on their mother tongue now goes beyond the local in-group resources of a subordinated group to which they have access, to open up new vistas of opportunity, if not for themselves, then for their children. Indeed, their children are among those who frequent the French-language minority schools where I have done fieldwork since 1983. In these schools it is possible to see in microcosm many of the issues I have just discussed.

< previous page

page_135

next page >

< previous page

page_136

next page > Page 136

There are students who in fact speak only one language well (whether English or French), because they just arrived from a monolingual French region in Canada or abroad, because they come from families where language transfer has already taken place, or because they are children of the anglophone middle class seeking bilingualism. There are others whose experience of life includes participation both in domains where French dominates and in ones dominated by English. Official (i.e. adult-run) school activities are normatively conducted in French (on the same grounds that Louise insists on being served in French at the grocery store); peer-group activities are normatively conducted in English. In order to participate in both, one has to be bilingual. The first set of children, the French- or English-dominant ones, either withdraw into monolingual networks (to the point of being marginalised either by peers or by the school) or learn enough of the other language to get by. The second use both, codeswitching at points where the different frames of reference of their experience collapse. For these bilingual children, what is important is to retain access to both French and English worlds, and so it is necessary to neutralise tension between the languages when they clash, but otherwise keep them separate. Sometimes, this has to be dealt with explicitly. In one instance, a group of 7-year-olds were playing in English in the schoolyard. A classmate, who, for the purposes of our research, was wearing a tape-recorder, wanted to join in. One of the boys told her she couldn't; his problem was that, since she was wearing a tape-recorder for adults who were clearly associated with the French-only explicit norm of the school, she would be expected to speak French, but the game was taking place in English ('on joue en anglais (we're playing in English)'). Other times, the collapse of frames can be dealt with implicitly through codeswitching or otherwise embedding different cultural and linguistic elements within a single frame. These are usually moments when students must perform in front of an audience which consists both of their teachers and of their peers (these are mainly teacher-controlled situations, like oral reports or dramatic skits), or groupwork involving student collaboration in the accomplishment of a teacher-defined task. In the following example, a Grade 6 student interrupted a Grade 7/8 class in the middle of a lesson, in order to make a request of the teacher. Example 9 (Toronto 1983; French-language elementary school, Grades 7-8): Student: Uh Monsieur je m'excuse de vous déranger I know I better be mais est-ce que je pourrais avoir le poids rond? (uh Sir I'm sorry to disturb you . . . but could I have the round weight?) The student's sarcastic side comment is clearly intended not for the teacher but for the other students looking on; he may have to be overtly polite to the teacher, but he can show the others that he is not a 'suck'. By using English for this purpose, he also accomplishes a certain degree of deniability:

< previous page

page_136

next page >

< previous page

page_137

next page > Page 137

his official exchange with the teacher is in French, and anything else is off the record. In examples 10 and 11, students are engaged in small-group work in the classroom. In example 10, three students are in the process of putting together a poster, glueing on pictures they have cut out of magazines. Example 10 (Toronto 1986; French-language elementary school, Grade 3): Marie: eh colle bien (hey glue well). Walter: oui (yes). Sophia: It's good enough. Marie: It's all crooked. Walter: Crooked. Marie: Oh gosh oh j'ai presque mis à l'envers (oh I almost put upside down). Here, the codeswitching moves are not necessarily significant in and of themselves; rather, the relatively free combination of the languages allows the students to conduct (French-associated) classroom tasks without overly contradicting their (Englishassociated) peer relations. In example 11, two older students are trying to put together, and, later, practice, a skit. The skit is in the form of a television advertisement for an Oriental sauce; one element revolves around a waitress bringing the sauce to customers in a restaurant, and tripping over a chair. Example 11 (Toronto 1987; French-language elementary school, Grade 8): Irene: Wait, then what happens? okay ça vole dans les airs . . . non éa vole dans les airs and then what happens? falls? breaks? ( . . . okay it flies through the air . . . no it flies through the air . . .) ...... Anna: Okay on pratique on pratique? on move les chairs (okay we practice we practice? we move the chairs). While Anna's use of codeswitching resembles in many ways those of the younger students, Irene (here as elsewhere in the transcript) tends to use codeswitching not only to neutralise the tension flowing from collapsed frames of reference, but, further, to organise her discourse. In particular, she offers candidate pieces of script in French (the language in which the script will have to be written in any case), while her comments on it are in English. Elsewhere, I have argued that these differences among students can be attributed to the different forms of their individual repertoires, shaped by their socially-constrained experience of French and English (Heller, 1989b). Irene indeed has had much opportunity to use both French and English in her life, and has a broader set of resources in French to draw on than does Anna. The other side of the coin, however, is that the social situation at hand provides opportunities for using these resources (and not others) in different ways. It is significant in this regard that both Irene and Anna speak a third language, Irene Italian and Anna Farsi. Irene, however, never has an opportunity at school to make use of her (in any case

< previous page

page_137

next page >

< previous page

page_138

next page > Page 138

limited) Italian, while the presence of several other Iranian students makes it possible for Anna to use Farsi. The last example shows how codeswitching can frequently be bound up in wordplay and punning. While humour, of course, is an excellent resource for neutralising tension or creating role distancing, it also points to the benefits of bilingualism shared among those on the language border (I use the term in a sociological, not strictly geographical, sense). It can also, of course, become a pleasurable end in itself. Example 12 (Toronto 1983; French-language elementary school, Kindergarten): Student: Est-ce que tu sais qu'est-ce que 'je m'en fiche' veut dire? Teacher: Quoi? Student: Je m'en poissonne! Teacher: ??? Student: Fiche, fish! In this example, a 5-year-old boy demonstrates considerably more metalinguistic bilingual sophistication than his hapless teacher. While she does in fact speak English, she thinks of herself primarily as a francophone, and, in keeping with school policy and the general ideology of mobilised francophones, she carefully protects the French-only domain of school. Her separation of the languages, however, makes it impossible for her to understand the boy's joke, and he is reduced to practically spelling it out for her (fiche sounds like 'fish', in French 'fish' is poisson, so fich-er leads to fish-er, and finally to poisson-ner). In this section, I have tried to show how the distribution and use of French and English, and within that, of French-English codeswitching is part of a process of francophone ethnic mobilisation in Quebec and Ontario. That process directly involves some (but not all) individuals, whether they are engaged in it, resist it, or adapt to it (that is, try to position themselves in such a way as to take advantage of its consequences). That involvement is made possible (or at least facilitated) by virtue of their social position, and they go about it, in part, by drawing on and exploiting in specific ways the resources which form part of their verbal repertoire. Others, usually because they were too far from the boundary to begin with, are either insulated from the process altogether, or are left behind unless they can organise themselves (collectively or individually) either to fight for the value of the symbolic capital they do possess, or to acquire new forms of capital whose value is increasing. Conclusion In the particular case discussed here, codeswitching is part of a process of ethnic mobilisation which is characterised less by social transformation than by a realignment of the relations of power between ethnic groups. The

< previous page

page_138

next page >

< previous page

page_139

next page > Page 139

mobilisation of francophones has mainly been about substituting francophone (or in some cases bilingual) control of resources hitherto controlled by anglophones. It has not been about the creation of alternative markets where different kinds of resources acquire value and are exchanged. The success of the mobilisation has been brought about in part through the establishment of new conventions of language use, that is, by the ability of francophones to impose French as the key to gaining access to situations in which the resources they now control are distributed. Codeswitching used to be about participation in anglophone-controlled networks where crucial economic goods circulated and participation in francophone-controlled net-works where members of a subordinate group provided each other with the means to live with their subordination. Now it is about participation in circles where the same kinds of goods circulate, but are differentially con-trolled. This process is not without its ironies. English has not totally lost its value in the marketplace, not only because of the power which anglophones retain in Canada, but because of the role of English in the global economy. Francophones have mobilised in order to gain access, as francophones, to that economy, in order to control the way its benefits are distributed among them. However, in order to participate successfully there, they must learn and use English. The way to the top may have been monolingual, but bilingualism has come to meet the new francophone élite at the summit. Second, competition for bilingualism between francophone and anglophone élites has emerged. As the francophone élite manoeuvres to acquire English, the anglophone élite rushes to learn French, in both cases in order to gain or retain privileged access to precisely the same kinds of jobs and class positions in both the private and the public sector. Finally, francophone mobilisation has been built not only on the simple re-evaluation of French as a symbolic resource, but on the development of a particular variety of highly-valued French. This variety is a Canadian standard which distances its speakers from (and establishes their legitimacy with respect to) both European French and the still stigmatised vernacular Canadian varieties. Finally, French is now set up, parallel to English, as a crucial linguistic resource to which not only powerful anglophones may aspire, but so must relatively powerless groups such as members of native groups and immigrants. What this does, of course, is create the kinds of social inequalities within francophone society against which mobilisation was directed in terms of the broader frame of Canadian society. However, this only serves to reinforce my major point. In order to understand the role and significance of codeswitching in the kinds of political processes described here, it is essential to understand not only its distribution in the community, but, more importantly, how that distribution is tied to the way groups control both the distribution of access to valued resources and the way in which that value is assigned. Further, the study of language choice and codeswitching can shed light on the ways in which groups

< previous page

page_139

next page >

< previous page

page_140

next page > Page 140

struggle over resources, and on the ways in which individual members of a community contribute to that struggle by creatively and strategically exploiting their linguistic resources in key interactions. Notes 1. An earlier version of this paper was presented at the Vrije Universiteit Brussel and at the Third Workshop on Codeswitching and Language Contact of the European Science Foundation, Brussels, November 1990. I am grateful to all the participants in that workshop for their extremely helpful comments and suggestions, and, in particular to Georges Lüdi, Peter Nelde and Piet Van de Craen. 2. Most of this material has been published or presented elsewhere. In this paper I draw most heavily on material discussed in greater detail in Heller, 1982a,b, 1984, 1985, 1987, 1988, 1989a,b; Heller et al., 1982; Heller & Lévy, 1992. The research on which this paper draws was funded by the Social Sciences and Humanities Research Council of Canada, the Quebec Ministry of Cultural Development, the Multiculturalism Directorate of the Secretary of State (Government of Canada), the Ontario Ministry of Education Transfer Grant to the Ontario Institute for Studies in Education, and a school board. Their support is gratefully acknowledged. 3. Of course, Gumperz, drawing on ethnomethodology, has always focused on communication as constitutive of social reality. He has, however, less frequently examined the processes which link communication in face-to-face interaction to broader questions of social relations of power and solidarity. 4. The purpose of this paper is to concentrate specifically on the relationship between speakers of French and speakers of English in Canada. In adopting this stance, I do not mean to imply that these were, or are now, the only groups involved in Canadian society. Certainly, it is important to keep in mind that even that dynamic is embedded in important ways in the relationship of both groups to others, whether to the indigenous population or to more recently arrived immigrants from other parts of the world, or even, beyond Canada's borders, to people elsewhere. Exploration of these relationships is, however, beyond the scope of this paper. 5. Of course, intermarriage with Irish members of the working class was not the only way in which francophones could lose their language, but it was one element in a more complex process. 6. Indeed, political scientists and historians continue to debate the nature and significance of those processes, notably the extent to which class interests (and the emergency of new class formations within francophone society) were central to the mobilisation process (Anctil, 1990). 7. The French original is given in italics, followed by an English translation in regular typeface. Where examples contain utterances in English, these are given in boldface. The interviewer's comments or back channel responses are inserted into the quotation in square brackets. Features of the discourse are in curly brackets; unintelligible speech is indicated as {xx}. 8. In a recent article, Cartwright (1987) has examined aggregate manifestations of shifts in patterns of bilingualism between 1971 and 1986, the period of most intense francophone mobilisation. He points out that it is certainly possible to discern a shift among Quebec anglophones, particularly younger ones (between the ages of 5 and 25), towards a greater degree of bilingualism; nonetheless, this may be due in part to an exodus of unilingual anglophones from the province, leaving behind those who either were already bilingual, or who were for some reason more open to the idea of becoming bilingual.

< previous page

page_140

next page >

< previous page

page_141

next page > Page 141

References Anctil, P. (ed.) (1990) La Révolution tranquille: vers une réinterprétation. Colloque (octobre 1990). Êtudes sur le Québec à McGill, Bulletin no. 8, Montréal: Programme d'études canadiennes-françaises, Université McGill. Bourdieu, P. (1977) L'économie des échanges linguistiques. Langue française 34, 17-34. (1982) Ce que parler veut dire. Paris: Fayard. Cartwright, D. (1987) Accommodation among the anglophone minority in Quebec to official language policy: a shift in traditional patterns of contact. Journal of Multilingual and Multicultural Development 8(1,2), 187-212. Cicourel, A. (1975) Discourse and text: cognitive and linguistic processes in studies of social structure. Versus 12(2), 33-84. (1978) Language and society: cognitive, cultural and linguistic aspects of language use. Sozialwissenschaftliche Annalen. Band 2, Seite B25-B58. Vienna: Physica-Verlag. (1980) Three models of discourse analysis: the role of social structure. Discourse Processes 3, 101-32. Collins, J. (1987) Conversation and knowledge in bureaucratic settings. Discourse Processes 10(4), 303-19. (1988) Language and class in minority education. Anthropology and Education Quarterly 19(4), 299-326. Davis, K. (1988) Power Under the Microscope. Amsterdam: Foris. Erickson, F. and Schultz, J. (1982) The. as Gatekeeper. N.Y.: Academic Press. Fennario, D. (1980) Balconville. Vancouver, Los Angeles: Talonbooks. Gal, S. (1988) The political economy of code choice. In M. Heller (ed.) Codeswitching: Anthropological and Sociolinguistic Perspectives (pp. 245-64). Berlin: Mouton de Gruyter. (1989) Language and political economy. Annual Review of Anthropology 18, 345-67. Giddens, A. (1984) The Constitution of Society. Berkeley, Los Angeles: University of California Press. Gumperz, J. (1982) Discourse Strategies. Cambridge: Cambridge University Press. Heath, S. B. (1983) Ways With Words. Cambridge: Cambridge University Press. Heller, M. (1982a) Negotiations of language choice in Montreal. In J. Gumperz (ed.) Language and Social Identity (pp. 10818). Cambridge: Cambridge University Press. (1982b) Language, Ethnicity and Politics in Quebec. Unpublished Ph.D. Thesis, Dept. of Linguistics, University of California, Berkeley. (1984) Language and ethnic identity in a Toronto French-language school. Canadian Ethnic Studies 16(2), 1-14. (1985) Ethnic relations and language use in Montreal. In N. Wolfson and J. Manes (eds) Language of Inequality (pp. 75-90). Berlin: Mouton de Gruyter. (1987) The role of language in the formation of ethnic identity. In J. Phinney and M. Rotheram (eds) Children's Ethnic Socialization (pp. 180-200). Newbury Park: Sage. (ed.) (1988) Codeswitching: Anthropological and Sociolinguistic Perspectives. Berlin: Mouton de Gruyter. (1989a) Aspects sociolinguistiques de la francisation d'une entreprise privée. Sociologie et sociétés 21(2), 115-27. (1989b) Communicative resources and local configurations: an exploration of language contact processes. Multilingua 8(4), 35796. Heller, M., Bartholomot, J.-P., Lévy, L. and Ostiguy, L. (1982) Le processus de francisation dans une entreprise montréalaise: une analyse sociolinguistique. Québec: l Éditeur officiel. Heller, M. and Lévy, L. (1992) Mixed marriages: life on the linguistic frontier. Multilingua 11(1): 11-43. Hill, J. (1985) The grammar of consciousness and the consciousness of grammar. American Ethnologist 12(4), 725-37. Hughes, E. (1943) French Canada in Transition: The Effects of Anglo-American Industrialization upon a French-Canadian Small Town. Chicago: University of Chicago Press.

< previous page

page_141

next page >

< previous page

page_142

next page > Page 142

Lachapelle, R. and Henripin J. (1980) La situation démolinguistique au Canada. Montreal: Institute for Research on Public Policy. Lieberson, S. (1965) Bilingualism in Montreal: a demographic analysis. American Journal of Sociology 71, 10-25. MacLennan, H. (1945) Two Solitudes. Toronto: Collins. Marcus, G. (1986) Contemporary problems of ethnography in the modern world system. In J. Clifford and G. Marcus (eds) Writing Culture. Berkeley, Los Angeles: University of California Press, pp. 165-93 · Mehan, H. (1983) The role of language and the language of role in institutional decision-making. Language in Society 12(2), 187-212. (1987) Language and power in organizational process. Discourse Processes 10(4), 291-302. Mehan, H., Hertweck, A. and Meihls, J. (1986) Handicapping the Handicapped: Decision-making in Students' Educational Careers. Stanford: Stanford University Press. Michaels, S. (1981) 'Sharing time': childrens' narrative styles and differential access to literacy. Language in Society 10(3), 42342. Shirley, D. (1986) A critical review and appropriation of Pierre Bourdieu's analysis of social and cultural reproduction. Journal of Education 168, 96-112. Williams, R. (1976) Keywords: A Vocabulary of Culture and Society. London: Fontana. Woolard, K. (1985) Language variation and cultural hegemony: toward an integration of sociolinguistic and social theory. American Ethnologist 12(4), 738-48. Woolard, K. (1989) Double Talk: Bilingualism and the Politics of Ethnicity in Catalonia. Stanford: Stanford University Press.

< previous page

page_142

next page >

< previous page

page_143

next page > Page 143

French-Dutch Codeswitching in Brussels: Social Factors Explaining its Disappearance Jeanine Treffers-Daller 1 University of Amsterdam, Department of General Linguistics, Spuistraat 210, NL-1012 VT Amsterdam, The Netherlands Abstract The focus of this paper is on intrasentential codeswitching among locally-born inhabitants of Brussels. The older Brusselers have the reputation of switching frequently between the local varieties of Dutch and French, but the younger generations switch much less often. The disappearance of intrasentential codeswitching is shown to be related to knowledge of standard Belgian Dutch, that is the southern variant of standard Dutch, among other factors. Informants who have a good knowledge of standard Dutch (for example because they have attended Dutch-speaking schools) switch less often than those who do not master this code very well. Apart from this, intrasentential codeswitching is probably no longer a generalised practice in Brussels because 'the codes involved symbolise social groups in conflict/competition with each other', as Myers-Scotton (1991) formulated it. Introduction Codeswitching, the alternation of two languages in discourse, has often been studied in immigrant communities, such as the Puerto Rican community in New York (Poplack, 1980) or the Moroccan-Arabic community in Utrecht, the Netherlands (Nortier, 1989). In these communities, code-switching is a generalised practice, especially among adolescents. In this paper, the focus is not on immigrants but on the indigenous inhabitants of Brussels, who are reputed for switching between the local varieties of French and Dutch (Baetens Beardsmore, 1983). According to Louckx & Ter Hoeven (1975), codeswitching is no longer practiced by younger inhabitants of the capital. Although there are no other explicit statements in the literature with respect to the disappearance of codeswitching in Brussels, most observers seem to agree with these authors. The object of this study is to investigate whether codeswitching is indeed disappearing and, if this can be demon-

< previous page

page_143

next page >

< previous page

page_144

next page > Page 144

strated to be the case, which factors contribute to its disappearance. The data analysed are drawn from my thesis (TreffersDaller, forthcoming). Since different definitions of codeswitching are used in the literature, I will give my definitions of several types of codeswitching first. The concept 'intersentential codeswitching' refers to switching between sentences. For switching within sentences, I will use the term 'intrasentential codeswitching'. An example of intrasentential codeswitching is given in (1). The French part of the sentence is in italics. (1) Je suis au balcon op mijn gemakske zo en train de regarder les étoiles (I am watching the stars at my ease on the balcony) Switches between two co-ordinated sentences (cf. (2)) are counted as inter-sentential codeswitches. I chose not to treat them as intrasentential switches because the two clauses are only semantically related and not syntactically. (2) Nadine est née au mois d'avril en dan in de maand oktober heb ik een winkel opengedaan in de Birminghamstraat (Nadine was born in April and then in October I opened a shop in Birmingham Street) A switch between a main clause and a subordinate clause, however, is treated as an intrasentential codeswitch (cf. (3)). (3) 't is dat que j'ai dit à madame (That is what I said to the lady) All single French items, such as bureau, that are used in Brussels Dutch sentences, as well as all single Dutch items in Brussels French, such as zeg (you know, lit. say) are excluded from the present analysis, since they may very well be borrowings. Although some authors treat single elements as codeswitches (Poplack, 1980; Berk-Seligson, 1986; Nortier, 1989), the distinction between borrowings and single word switches is rather arbitrary in my view. I excluded the single word switches from the analysis, in order to make sure that my calculations would be based on a corpus of unambiguous codeswitches. Poplack (1980) was the first to do a systematic analysis of the relation between the occurrence of inter- and intrasentential codeswitching on the one hand and the background of the speakers on the other hand. She found that reported language proficiency was one of the best predictors of inter-and intrasentential codeswitching among Puerto Ricans in New York. The bilinguals among the informants switched significantly more often within sentences than the Spanish-dominant speakers. Poplack's results were largely confirmed by Nortier (1989), in her study of Dutch-Moroccan Arabic code-switching. On the basis of Poplack (1980) and Nortier (1989), I hypothesised that bilingual language proficiency, that is active proficiency in the local varieties of French and Dutch, would be an important predictor of codeswitching

< previous page

page_144

next page >

< previous page

page_145

next page > Page 145

behaviour. Knowledge of standard Dutch 2 was assumed to correlate negatively to codeswitching, for reasons that will be explained below in detail. For the same reasons, informants who had gone to Dutch-speaking schools were supposed to switch significantly less than informants who had attended French-speaking schools. Finally, positive attitudes towards language mixing, and linguistically diverse social networks were supposed to have a positive influence on codeswitching. I will come back to these hypotheses in later sections. First, I will briefly present the Brussels linguistic situation. The Linguistic Situation Brussels Brussels is situated on the north of the language frontier, which divides Belgium into Dutch-speaking Flanders in the north and French-speaking Wallonia in the south. The urban area consists of nineteen municipalities. Inhabitants of two of these municipalities have taken part in this study: the central municipality of Brussels and Anderlecht. The inhabitants of the central municipality, and especially of the popular quarter called 'the Marolles', are supposed to be the most frequent codeswitchers. Other informants come from Anderlecht, which is situated in the west of the urban area, where one can find more speakers of Dutch than in the south or the east of the city. Since 1963 the capital of Belgium is officially bilingual, which means that both Dutch and French are used in administration, schools and public services. Not all of the inhabitants of the city are bilingual, however. Estimates of the Dutch-speaking population vary from 15% to 30%. The majority of the inhabitants speak French only. Though French is tradition-ally the more prestigious language in Belgium, Dutch has gained importance since, from the 1960s on, Flanders has become the stronger economic region (Van Velthoven, 1989). Since the 1960s, the socio-political climate in the city has been severely affected by the ongoing conflict between the Dutch-speaking and the French-speaking population groups concerning the federalisation of the state, in which the struggle over the status of Brussels was one of the most difficult issues. Though I cannot go into the details of this conflict, it may very well bear influence upon codeswitching behaviour. According to Myers-Scotton (1991:6), codeswitching as an unmarked choice (i.e. rapid unintentional codeswitching) does not occur when 'the codes involved symbolize social groups in conflict/competition with each other'. Louckx & Ter Hoeven (1975), assume that only the older Brusselers, from 60 years onwards, still identify with both French and Dutch, whereas the younger Brusselers identify with either French or Dutch: the polarisation between the two major linguistic groups in the countrythe speakers of French in the south and the speakers of Dutch in the north makes it difficult not to Abe on one or the other side. In such a situation the Brusselers may no longer consider a mixed code to be an appropriate expression of their identity.

< previous page

page_145

next page >

< previous page

page_146

next page > Page 146

In a metropolis like Brussels, the local dialects, Brussels Dutch and Brussels French, are quickly losing ground and are being replaced by the standard varieties of French and Dutch. Brussels Dutch belongs to the group of Brabantic dialects spoken in the central part of the Dutch-speaking territory of Belgium. According to most authors, it is rapidly disappearing, and it turned out to be relatively difficult to find younger speakers of this dialect. Nowadays hardly anyone speaks solely Brussels Dutch (De Vriendt & Goyvaerts, 1989). Most Brusselers also speak (Brussels) French and some speak Belgian Dutch as well. De Vriendt & Willemyns (1987:204) define Belgian Dutch as a 'supra-regional language which is more or less standardized and may, for those who speak it, function as a standard language'. In the Dutch-speaking schools in Brussels and elsewhere in the country, much attention is paid to teaching 'correct standard Dutch' without archaisms, gallicisms, etc. Only this code is assumed to be a serious alternative for French in the Belgian context. The growing importance of standard Dutch in Brussels would seem to have serious implications for codeswitching behaviour. Brusselers who speak and hear standard Dutch relatively often may very well be influenced by its normative and puristic characteristics, and may therefore be inclined to mix Dutch and French less often in their daily life than those who have less contact with standard Dutch and speak most often (Brussels) French. Brussels French cannot be delimited as clearly as Brussels Dutch. According to De Vriendt & Willemyns (1987: 211) there is a continuum 'going from people whose speech displays (almost) all characteristics of the code to, at the other extreme, people who speak it with very few, more or less occasional, remnants of Brussels French (typical words, maybe a particular way of stressing phrases, perhaps the pronounciation of a vowel)' (p. 211). Though Belgian French is very similar to standard French, the lexicon of both varieties varies slightly. The Present Study Methodology Recordings were made with 34 locally-born Brusselers from Anderlecht and the central municipality of Brussels who could speak Brussels Dutch and Brussels French. These informants were contacted through a social network approach. The informants were either skilled workers or self-employed. Ages ranged from 21 to 80, though dialect speakers under 50 could be recorded only in Anderlecht. The choice of inhabitants of Anderlecht allowed us to test our hypothesis concerning the supposed negative influence of standard Dutch on codeswitching, since standard Dutch is spoken by more people in Anderlecht than in the central municipality, according to the language censuses from 1947 3 and later surveys (Logie, 1981). Data concerning language proficiency in Brussels Dutch, Brussels

< previous page

page_146

next page >

< previous page

page_147

next page > Page 147

French, standard Dutch and standard French were obtained with the help of five point self-rating scales. Because of the highly sensitive character of the language issue in Brussels, language tests were considered inappropriate for measuring language proficiency, instead an idiom test was used to get an impression of overall language proficiency. The general procedure was as follows: Informants received a list of ten Dutch and ten French idioms, which were offered in two formulations: the correct one and an invented, but almost similar one. The informants had to say which version was generally used in Brussels. Thus, no explicit reference was made to language competence, though a try-out session showed that only persons who had a good active command of Dutch and French were able to recognise the correct idioms. Furthermore, a general questionnaire was used to obtain information concerning their attendance at a French-speaking or a Dutch-speaking school, personal background, age, profession, language choice in different situations and attitudes towards language mixing. Finally, the informants were requested to fill out a questionnaire concerning their social network. This questionnaire asked the informants to think of six persons with whom they spoke most often and to indicate which language(s) they spoke with these persons. This allowed us to obtain an impression of the linguistic composition of the informant's primary network, and made it possible to test the hypothesis that informants with diversified networks would be most likely to switch. All the calculations presented below are based on the relative frequency of language mixture in the speech of the informants, and not on absolute numbers of these phenomena, because the amount of speech recorded from each informant differs considerably. The relative frequency of intersentential and intrasentential codeswitching has been calculated in the following way. First, I calculated for each informant the mean sentence length for his/her French and Dutch utterances (on the basis of 20 French and 20 Dutch sentences). Then the total number of French and Dutch sentences realised by this informant was calculated: the total number of French words was divided by the mean sentence length, which gave the (estimated) number of French sentences and the same procedure was followed for the Dutch sentences. The numbers of French and Dutch sentences were then totalled. This gave the total number of possible inter-sentential switches, as well as the total number of sentences in which theoretically a codeswitch could have taken place. 4 Finally, I calculated the permillage of intersentential and intrasentential switches, with respect to the total number of sentences realised by each informant. The mean permillage of intrasentential codeswitches in the speech of all 34 inform-ants is 4.82 (SD 7.98), and the mean permillage of intersentential code-switches is 26.74 (SD 42.57).

< previous page

page_147

next page >

< previous page

page_148

next page > Page 148

Table 1 Pearson correlations between age and codeswitching Variable N r p 34 0.25 0.160 intrasent. switches ( ) 34 0.09 0.627 intersent. switches ( ) N = number of informants; r = correlation coefficient Results The corpus consists of approximately 150,000 Brussels Dutch words and 40,000 Brussels French words, transcribed from 31 hours of recordings of spontaneous conversations with 34 locally born Brusselers, from Anderlecht and the central municipality of Brussels. In total, 909 intersentential switches and 168 intrasentential switches were found in the data. The number of intrasentential codeswitches is relatively low in comparison with other studies, because all single word switches and borrowings (together almost 4,000) are excluded from the analysis for reasons explained above. Contrary to my expectations, there are no significant correlations between age and codeswitching (cf. Table 1). The permillage of codeswitches, however, seems to show a weak correlation with the age factor. Therefore the relationship was tested again, in a different way: the group of informants was divided into a group of persons of 60 years and younger, and a group of persons of 61 years and older. I chose 60 years as the cutting off point, because Louckx & Ter Hoeven (1975) assume that Brusselers older than 60 still identify with both languages, whereas the younger Brussels identify with either Dutch or French. The older inhabitants of the city might therefore be inclined to switch more often than younger ones (cf. above). The results of the t-test are more in line with the expectations than the Pearson correlations. Although the differences are still not significant, Table 2 shows that there is a trend that older informants switch more within sentences than younger informants. Table 2 Permillage of intrasentential codeswitches among younger (60-) and older informants (60+) Age group N M SD £60 years 16 2.38 5.48 >60 years 18 7.00 9.31 t-test; t = 1.737; df = 32; p = 0.092; N = number of informants; M = mean; SD = standard deviation

< previous page

page_148

next page >

< previous page

page_149

next page > Page 149

There are probably technical reasons for the absence of significant correlations between age and intrasentential codeswitching. With more inform-ants, especially more younger dialect speakers, the correlations would probably have been significant. In the discussion, I will come back to this result, and discuss the fact that, in contrast to intrasentential codeswitching, intersentential codeswitching turned out to be practiced by informants from all age groups, and not only by the oldest informants. Since there are only clear differences between speakers of different age groups with respect to intrasentential codeswitching, below most attention will be paid this type of switching and less to intersentential codeswitching. In order to test my hypotheses concerning the factors that might contribute negatively or positively to the frequency of intrasentential codeswitching, I studied the correlations between codeswitching on the one hand, and local background, schooling (French-speaking vs. Dutch-speaking), language proficiency, attitudes, and social networks on the other hand. With the exception of the factor 'social network', all factors showed a significant correlation to intrasentential codeswitching, and some to intersentential codeswitching as well. With respect to the factor 'local background', I obtained the following results. Informants from Brussels (i.e. the central municipality) mix significantly more often than the informants from Anderlecht (t-test; t = -2.669; df = 32, p = 0.012). Above I hypothesised that the difference in codeswitching behaviour between informants from both municipalities might be related to differences in knowledge of standard Dutch. As a matter of fact, the informants from Anderlecht have a better active knowledge of standard Dutch than the informants from the centre of Brussels, as measured by the self-rating scales (Mann Whitney U-test, U = 88.5; p = <0.001) and the idiom test (t-test, t = 2.601, p = 0.013). Knowledge of standard Dutch turns out to be an important predictor of intrasentential codeswitching. Those informants who have a good active command of standard Dutch do not mix codes very often, whereas those who do not speak standard Dutch very well switch more often (Kendall's tau, N = 30, r = 0.268; p = 0.038). On the other hand, switching within sentences requires a good knowledge of Brussels French. Informants who claimed to speak Brussels French 'very well' were found to switch significantly more than informants who had a less good active command of this code (Kendall's tau, N = 31; r = 0.351; p = 0.006). The test of the correlation between schooling and intrasentential code-switching reconfirms that informants who have a good knowledge of standard Dutch switch less often. Informants who had gone to a Dutch-speaking school switch significantly less often within sentences than those who had attended a French-speaking school (Mann Whitney U-test, U = 92.0, p (one-tailed) = 0.010). The frequency of intrasentential codeswitching is slightly higher for

< previous page

page_149

next page >

< previous page

page_150

next page > Page 150

informants who express positive attitudes towards switching (Kendall's tau; N = 32; r = 0.190; p = 0.127), even though the results are not significant. Contrary to my expectations, informants who have a negative attitude towards borrowing (and are in favour of the use of purisms) are more inclined to switch within sentences, instead of less (Kendall's tau, N = 30, r = -0.334; p = 0.009). The total network score 5 seems to have no clear influence on the permillage of intrasentential codeswitches (Pearson, N = 31, r = 0.28; p = 0.124), though with more informants, the correlations could become significant. Only one background factor, active proficiency in Brussels French, correlates significantly, though weakly, to intersentential codeswitching (Kendall's tau; N = 31; r = 0.252; p = 0.047). Interestingly, there is no relationship at all between proficiency in standard Dutch and intersentential codeswitching. Brusselers tend to switch a bit more between sentences than informants from Anderlecht, and informants with predominantly French networks tend to switch slightly more than those with Dutch networks, but the results are not significant. The most interesting question still to be answered is: which of the background variables is the best predictor of the dependent variable 'relative frequency of intrasentential codeswitching'? This question can best be treated through a multiple regression analysis. The analysis has been performed on scores of 26 informants, that is, all informants who score on all linguistic and all background variables. Since this type of analysis has been developed for variables measured on an interval scale and not for variables measured on an ordinal or a nominal scale, which is the case in this study, and because of the relatively small size of the sample, the results presented here have an explorative character. Table 3 presents the first and Table 4 the last step of the backward multiple regression analysis and shows that the five variables explain together almost 40% of the variance. Even if this means that there are probably other factors that contribute to the variance as well, the factors chosen here are certainly relevant predictors of intrasentential codeswitching. Intrasentential codeswitching correlates most strongly with the variable 'schooling'. As a matter of fact, this variable explains alone almost 30% of the variance (cf. last step in the multiple regression analysis). This means that the added contribution of the other four variables to the variance is no more than 10%. Especially the variable 'local background' does not contribute very much to the prediction of intrasentential codeswitching. The individual contributions of the factor 'active proficiency in Brussels French' and of the attitudinal factor are relatively important in comparison to 'local background' and 'proficiency in standard Dutch', even though their influence is not as important as the school factor.

< previous page

page_150

next page >

< previous page

page_151

next page > Page 151

Table 3 The standardised weights of background variables in predicting the frequency of occurrence of intrasentential codeswitching (backward multiple regression analysis) Pearson r Beta First step Criterion intrasentential codeswitching predictors Var LB -0.430 -0.068 Var SCH 0.538 0.397 Var BF -0.288 -0.246 Var DT 0.407 0.180 Var PUR -0.235 -0.302 Explained variance: 0.399 Multiple R: 0.632 F-value = 2.658 df: 5 20 p-value = 0.053 Last step (all other variables have been excluded) Criterion intrasentential codeswitching Var SCH 0.538 0.538 Explained variance: 0.289 Multiple R: 0.538 F-value = 9.759 df = 1 24 p-value = 0.005. Var LB: local background; 1 = Brussel, 2 = Anderlecht Var SCH: language of school; 1 = Dutch, 2 = French, 3 = both Var BF: active proficiency in Brussels French according to self-rating; highest score = 1; lowest score = 5 Var DT: active proficiency in standard Dutch according to self-rating; highest score = 1; lowest score = 5 Var PUR: attitude towards purisms; low scores represent puristic attitudes, high scores represent a negative attitude towards purisms Discussion The statistical treatment presented above shows that the frequency of intrasentential codeswitching is best predicted by the variable 'schooling'. This variable is not independent from the variable 'local background', because the informants from Anderlecht went more often to Dutch-speaking schools than the informants from Brussels. In order to exclude local differences, a correlation test between permillage of intrasentential codeswitching and schooling has been done for informants from Brussels and for informants from Anderlecht separately. Within both groups, there is no significant correlation between schooling and frequency of intrasentential codeswitch-

< previous page

page_151

next page >

< previous page

page_152

next page > Page 152

ing. This means that the local differences cannot be reduced entirely to differences in schooling. Apparently, the frequency of intrasentential code-switching is dependent on both schooling (French-speaking or Dutch-speaking) and the local background of the informant. The factor 'active proficiency in standard Dutch' is clearly dependent on the factor 'schooling' as well. Of course, it is not surprising that informants who went to Dutch-speaking schools speak Dutch better than those who went to French-speaking schools. The common variance of these two variables amounts to 0.37 (Pearson; r = 0.61, p = 0.001), which is relatively important. Three of the five factors that correlate significantly to intrasentential codeswitchingactive proficiency in standard Dutch, local back-ground and schoolingpoint to knowledge of standard Dutch as having a negative influence on intrasentential codeswitching. Since teaching 'correct Dutch' (that is free from archaisms, gallicisms, etc.) is an important aim in Dutchspeaking schools in Belgium and is highly valued in other institutions as well, mixing of French and Dutch becomes most probably less popular among Brusselers who speak standard Dutch relatively often. Puristic attitudes are not popular among the informants, however. Only 20% of the informants in this study are in favour of replacing French borrowings by Dutch words in Brussels Dutch. Though more informants are against mixing French and Dutch (33%), the Brusselers do not reject intrasentential codeswitching as often as could be expected on the basis of the literature. Contrary to my expectations, however, the informants who mix codes frequently express negative feelings about the use of French words, and are in favour of purisms like regenscherm (the puristic equivalent of paraplu, umbrella). Whereas the majority of the informants do not consider the use of French borrowings to be problematic, the ones who switch within sentences claim to disapprove of French loan words more often. Maybe linguistic insecurity among the codeswitchers should be held responsible for the fact that they defend more puristic norms than the other informants. Since the codeswitchers have a poorer command of standard Dutch, they are less familiar with the norms this language represents. They may, for that reason, be tempted to demonstrate special attachment to generally known stereotypical characteristics of standard Dutch, in the hope of concealing their lesser command of this code. However, a significant correlation between positive attitudes towards purisms and low active proficiency in standard Dutch could not be demonstrated on the basis of my data. Apart from attitudes, knowledge of the local variety of French, Brussels French, is an important predictor of intrasentential codeswitching. I assume that persons who have gone to a French-speaking school switch more often than those who have attended a Dutch-speaking school, because they are used to switching between French and Dutch since their youth. As it turns out, most informants indicated that they speak or used to speak Brussels Dutch with their parents (51.4%) or both French and Dutch (31.0%). A

< previous page

page_152

next page >

< previous page

page_153

next page > Page 153

minority of the informants said to speak only (Brussels) French with their parents (11%) or only standard Dutch (7%). If the home language of the informants is predominantly Dutch, and the school language is French, switching becomes a necessity. Of course a combination of a French-speaking home and a Dutch-speaking school would produce the same mixture, but this combination is exceedingly rare in Brussels. As a matter of fact, the Dutch-speaking Brusselers choose French-speaking schools much more often than the French-speaking Brusselers choose Dutch-speaking schools (cf. for more details Deprez & Wynants, 1989). The results of the correlation test concerning intersentential switching and active proficiency in Brussels French confirm earlier findings of Poplack (1980), who stated that intersentential codeswitching requires a good active command of both languages. 6 There is, however, a much stronger relation-ship between knowledge of Brussels French and intrasentential codeswitching than between knowledge of Brussels French and intersentential code-switching. We may conclude from this that a real mixture of both languages requires apparently more proficiency in Brussels French than switching between sentences. This, again, confirms Poplack's findings. There were no other significant correlations between background factors and intersentential codeswitching. In my view this is due to the fact that almost all informants switched between sentences, whether they were young or old, from Anderlecht or from Brussels, from Dutch-speaking or from French-speaking schools. Knowledge of standard Dutch does not influence this type of codeswitching at all. Actually, there is no reason why someone who speaks Dutch very often and very well should not use French in a quote, for example. In that case, the contact between the two languages is very limited, as both languages are kept separate. Thus, puristic norms associated with standard Dutch should not have much influence on code-switching between sentences. Standard Dutch has gained importance in Brussels, as Flanders has become the stronger economic region since the 1960s (cf. above). Just like in other parts of Belgium, the local dialects are disappearing slowly in Brussels, because they can no longer compete with the standard languages. The general tendency to use standard languages instead of dialects contributes probably to the disappearance of intrasentential codeswitching. Moreover, the puristic norms associated with one of these languages, standard Dutch, exerts a negative influence on intrasentential codeswitching. Apart from this, the tension between the two major linguistic groups in society, the Francophones and the Dutch-speaking populations, may very well have created a negative climate for mixing languages that are associated with the two groups in conflict. This climate makes it more difficult for younger informants to identify with both language groups. Whereas older informants tend to stress the fact that Brusselers are neither Flemings nor Walloons, a 25 year old informant from Anderlecht said he considered himself to be a Fleming, although not a fanatic one. He used French relatively often for

< previous page

page_153

next page >

< previous page

page_154

next page > Page 154

quotes, when telling stories in the local dialect, but he did not mix languages within one sentence. Additional support for this interpretation comes from Myers-Scotton (1991:6) and from Poplack (1987). Myers-Scotton assumes that codeswitching as an unmarked choice (i.e. rapid unintentional codeswitching) does not occur when 'the codes involved symbolize social groups in conflict/competition with each other'. This description seems to apply not only to Brussels, but also to the linguistic situation of Ottawa-Hull, described by Poplack (1987). In her comparative study of codeswitching among French-speakers in Ottawa-Hull and Puerto Ricans in New York, Poplack (1987) points to the fact that in Ottawa-Hull hardly any smooth intrasentential French-English codeswitching was found, whereas Spanish-English bilingual Puerto Ricans in New York produced many smooth intrasentential switches. Poplack explains the divergent codeswitching patterns with reference to differences in attitudes: The different social, historical and political factors have led to differences in attitudes towards the use of English in the two situations, which themselves may be partly determinant of the contrasting codeswitching patterns. These attitudes may reflect the fact that bilingualism is seen to be emblematic of New York Puerto Rican identity (as compared both with Island Puerto Ricans and non-Puerto Rican anglophones), whereas in the Ottawa-Hull situation, knowledge of English does not appear to be associated with any emergent ethnic grouping (Poplack, 1987: 67). Though Poplack does not explain why bilingualism does not form the basis of any emergent ethnic grouping in Ottawa-Hull, the tensions between the Francophones and the Anglophones probably prevent this identification. Brussels and Ottawa-Hull have much in common, because these cities are the capitals of countries in which the antagonism between the major linguistic groups has had a tremendous political impact (cf. Deprez, 1981 for a comparative analysis of the Belgian and Canadian language situation). Even though maintaining a bilingual identity may not be easy for Puerto Ricans in New York either, it is probably easier for them than for the inhabitants of Brussels and of Ottawa-Hull, because identity conflicts are less politicised. Conclusion Codeswitching between local varieties of French and Dutch is no longer a generalised practice among locally born Brusselers. Though older inhabitants (from 60 years onwards) of the Belgian capital may switch regularly within sentences and between sentences, younger Brusselers can only be heard switching between sentences, for example when quoting another person. In this study, it was shown that the disappearance of intrasentential codeswitching should be attributed to the puristic influence of standard Dutch, since informants who master this code well, and/or have attended a Dutch-

< previous page

page_154

next page >

< previous page

page_155

next page > Page 155

speaking school, and/or live in Dutch-speaking areas of the city, switch significantly less than others. Apart from the influence of the standard languages, which are replacing the local variants of Dutch and French, the tension between the linguistic groups in Belgium has probably created an unfavourable climate for code-switching. A similar situation exists in Ottawa-Hull, where intrasentential codeswitching is rare as well, according to Poplack (1987), because bilingual-ism is not considered to be emblematic of the local identity. Notes 1. Thanks are due to Hans van Kemenade (vof PCA, Utrecht, The Netherlands), who carried out the statistical treatment of the data. I have profited greatly from discussions with Pieter Muysken, Sera de Vriendt, Helmut Daller and Rob Schoonen. I am particularly grateful to Frank Desmet, Inge Callebaut and Patrick de Kreyger for their help in collecting the data and transcribing them. 2. From now on, the concept 'standard Dutch' refers to the southern variant of standard Dutch as spoken in Belgium. 3. These are the most recent official data concerning language knowledge and use in the city. Language censuses were abolished later on due to political tensions. 4. In theory, it is possible that one sentence contains more than one intrasentential codeswitch. Generally, however, one sentence contains only one codeswitch. 5. Informants indicated which code they used with six persons they used to see often. The results of this test were reduced as follows: if the informant claimed to speak standard Dutch with any person mentioned in the network test, this was coded as (1). For Brussels Dutch a (2) was assigned, for a combination of (Brussels) Dutch and (Brussels) French a (3) and for the exclusive use of (Brussels) French a (4). Thus, the higher a score, the more French was used in contact with the person in the network. The results were added up per informant. The total network scores per person can therefore vary from 6 to 24. 6. All informants indicated to have a good knowledge of the Brussels Dutch. There was more variation with respect to knowledge of (Brussels) French. References Baetens Beardsmore, H. (1983) Substratum, adstratum and residual bilingualism in Brussels. Journal of Multilingual and Multicultural Development 4, 1, 1-14. Berk-Seligson, S. (1986) Linguistic constraints on intrasentential codeswitching: A study of Spanish/Hebrew bilingualism. Language in Society 15, 313-48. Deprez, K. (1981) Belgisch Nederlands en Canadees Frans: een sociolinguistische vergelijking. In De Nieuwe Taalgids 74, 51031. Deprez, K. and Wynants, A. (1989) Minority problems: on the progress of Netherlandic primary education in Brussels. In Kas Deprez (ed.) Language and Intergroup Relations in Flanders and in The Netherlands. Dordrecht: Foris Publications, 6, 29-44. De Vriendt, S. and Willemyns, R. (1987) Linguistic Research on Brussels. In Els Witte and Hugo Baetens Beardsmore (eds) The Interdisciplinary Study of Urban Bilingualism in Brussels. 195-231. Clevedon: Multilingual Matters. De Vriendt, S. and Goyvaerts, D. (1989) Assimilation and sandi in Brussels. Leuvense Bijdragen Vol. 78, 1, 1-93. Logie, F. (1981) Ruimtelijke spreiding van de Nederlandstalige bevolking in Brussel-Hoofdstad. Taal en Sociale Integratie 3, 87-109. Louckx, F. and Ter Hoeven, P. J. A. (1975) Wetenschappelijk onderzoek van de taaltoestanden in de Brusselse agglomeratie, Autochtone Brusselaars op de tweesprong. Brussel: Nederlandse Commissie voor de Cultuur van de Brusselse agglomeratie.

< previous page

page_155

next page >

< previous page

page_156

next page > Page 156

Myers-Scotton, C. (1991) Intersections between social motivations and structural processing in codeswitching. In Papers for the Workshop on Constraints, Conditions and Models, London, 27-29 September 1990. Nortier, J. M. (1989) Dutch and Moroccan Arabic in Contact: Code Switching among Moroccans in the Netherlands. PhD thesis, University of Amsterdam. Poplack, S. (1980) Sometimes I'll start a sentence in Spanish Y TERMINO EN ESPANOL: toward a typology of codeswitching. Linguistics 18, 581-618. (1987) Contrasting patterns of code-switching in two communities. In E. Wande et al. (eds) Aspects of Multilingualism. Proceedings from the Fourth Nordic Symposium on Bilingualism, 1984, 51-77. Treffers-Daller, Jeanine (forthcoming) French-Dutch Language Mixture in Brussels, PhD thesis, University of Amsterdam. Van Velthoven, H. (1989) The relationship between Flanders and Brussels from 1830 to 1980. Mechanisms of power in a historical context. In Kas Deprez (ed.) Language and Intergroup Relations in Flanders and The Netherlands. Topics in Sociolinguistics 6. Dordrecht: Foris Publications, 11-28.

< previous page

page_156

next page >

< previous page

page_157

next page > Page 157

Codeswitching: Black English and Standard English in the African-American Linguistic Repertoire Charles E. DeBose Department of English, California State University, Hayward, California, USA Abstract In this paper, Black English (BE) and standard English (SE) are treated as two different closely-related linguistic systems which coexist in the African-American linguistic repertoire. Each system is defined as an autonomous grammar, and the interaction between them is considered to be governed by the same principles as those that govern languages in contact generally. In the process of collecting data for the purpose of describing the BE grammar the author observed some of the informants apparently codeswitching. This paper focuses on a middle-class female informant who appears to be a balanced bilingual. In the first few minutes of her performance she could be mistaken for a monolingual SE speaker. As the session progresses, however, she makes several notable switches to BE. Her last performance, in contrast to the earlier ones, is frequently in BE. The evidence considered in this paper is striking counter evidence to the claim that BE is spoken mainly by poor and uneducated persons. Overview There is a gap in the existing body of literature on Black English which this paper attempts to fill by making a modest contribution to the subject of codeswitching by African-Americans. Labov's collected works on the Black English Vernacular (1972) does not contain a single mention of codeswitchingnor does Wolfram's (1969) . . . Sociolinguistic Description of Detroit Negro Speech. Dillard, in the chapter on 'Who Speaks Black English' acknowledges that 'There are nowand were in the pastBlack bidialectal speakers' (1972:230). Further along in the same discussion Dillard makes reference to 'two nearly bidialectal boys . . . who served as informants.' Dillard attests to the occasional codeswitching of these youngsters:

< previous page

page_157

next page >

< previous page

page_158

next page > Page 158

There was a way of talking that they used with their friends and a way they spoke around their parents (Black English and Standard English approximately), and they had kept them apart without devoting any special effort to doing so. He does not provide any direct evidence of their linguistic behaviour, however. Taylor surveyed black ninth graders from southern and border states from whom he elicited a 70% affirmative response to the question: 'Can you switch dialects easily?' (1975:36). In reporting his findings Taylor comments that: Dialect switching is a commonly observed phenomenon, especially among older and better educated blacks (1975:37). Although he provides no direct evidence of 'dialect switching' by black persons, due to the focus of his studies upon language attitudes, Taylor's status as an African-American scholar lends credibility to his observation regarding codeswitching. The scant attention paid to BE/SE codeswitching in previous studies seems to be a consequence of prevalent assumptions about the language situation in black America, the nature of the African-American linguistic repertoire, and the methods of data collection and analysis that have been employed in those studies. Labov's theoretical claims regarding the inherent variability of American English presuppose a monolingual language situation in the African-American speech community in which: The gears and axles of English grammatical machinery are available to speakers of all dialects, whether or not they use all of them in their everyday speech (1972:64). Dillard, while acknowledging a 'bidialectal' linguistic repertoire for African-Americans, seems to under-estimate the prevalence of bilingual speakers. He seems to believe that most middle-class blacks are monolingual speakers of SE, whereas most poor blacks are monolingual BE speakers; and that middle class African-Americans rely upon 'ethnic slang' for the purpose of maintaining linguistic identity with their group (1972:239). Wolfram conceptualises the African-American language situation for the purpose of his Detroit study as follows: Ideally, one may think of the speech of Negroes in Detroit in terms of two polar dialectsa socially superordinate standard variety of English (referred to by Stewart . . . as 'acrolect') and a socially subordinate nonstandard variety (referred to as 'basilect') (1969:2,3). BE speakers are located at different points on the continuum between the idealised polar lects, in a manner that might be amenable to analysis using a (post-) creole continuum model. The method actually chosen for Wolfram's study, however, that of 'correlating social with linguistic variables', is not

< previous page

page_158

next page >

< previous page

page_159

next page > Page 159

well-suited for studies of codeswitching. That method, pioneered by Labov, relies upon statistical analysis of the frequency of occurrence of selected phonological or grammatical variants in data from groups of individuals, spanning a wide range of times and locations; whereas the locus of code-switching is individual speakers in particular conversational episodes. By focusing upon group rather than individual data, scholars such as Wolfram and Labov have allowed the belief that most middle-class African-Americans do not speak BE to go unchallenged. Taylor (1983) is rightly critical of the tendency of researchers such as Dillard to associate BE with poor uneducated AfricanAmericans and associate SE with middle-class blacks. To avoid the consequence of an excessively narrow definition of BE, i.e. of excluding middle-class blacks from the community of BE speakers, Taylor proposes a 'comprehensive' definition of BE as consisting of 'the totality of language used in the black community'. In Taylor's definition, standard and nonstandard forms of language are components of the 'totality' which is Black English. Without resorting to what Dillard calls 'ethnic slang', an educated, middle-class black person may express his or her identification with African-American culture, free of the stigma attached to nonstandard speech, through the use of what Taylor calls 'Standard Black English'. While Taylor's definition nicely accommodates the diversity of language phenomena that exist within the rubric of African-American culture, it is not clear how it would handle the particular phenomenon of codeswitching. In this paper, BE and SE are treated as two different closely-related linguistic systems which coexist in the African-American linguistic repertoire. Each system is defined as an autonomous grammar, and the interaction between them is considered to be governed by the same principles as those that govern languages in contact (Weinreich, 1953) generally. Such a model does not rule out Taylor's claim of standard/vernacular variation within BE, but it allows for the separate existence of an ethnicallyunmarked standard English as a superposed variety. In the process of collecting tape-recorded data, primarily for the purpose of describing the autonomous BE grammar, I observed some of the inform-ants engaging in what seemed to be codeswitching behaviour. Some examples of that behaviour are examined below following a brief overview of current approaches to the study of codeswitching. The analysis is admittedly tentative, and subject to the benefits and limitations of my insights and intuitions as a bilingual member of the AfricanAmerican speech community. Codeswitching Codeswitching, defined as 'the use of more than one language in the course of a single communicative episode' (Heller, 1988), is a widespread if not universal phenomenon of multilingual speech communities. Within the discipline of sociolinguistics, a considerable body of literature has developed

< previous page

page_159

next page >

< previous page

page_160

next page > Page 160

in pursuit of an adequate answer to the question 'Why do people code-switch?' Some of the first attempts to answer that question were characterised by classification of instances of codeswitching into opposing categories such as 'situational' vs. 'metaphorical' (Blom & Gumperz, 1972). Other studies have focused upon the problem of attempting to determine whether or not there are any linguistic constraints on when a bilingual speaker might switch from one code to another (Pfaff, 1979). One shortcoming of the situational vs. metaphorical dichotomy is that it fails to account for data in which switching occurs frequently without any clear change in the social parameters of the situation and without any apparent social meaning encoded in the switches. Scotton elaborates a comprehensive model based on the premise that: speakers have tacit knowledge of the indexical quality of code choices and their markedness as part of their communicative competence . . . (1988:64). Within such a perspective, the seemingly unrestrained kind of codeswitching in which intraword, as well as intrasentential, switching occurs might be considered codeswitching as an unmarked choice, whereas instances in which speakers appear to switch languages in order to 'change the balance of rights and obligations between participants' (1988:62) are accounted for by the notion of codeswitching as a marked choice. Methodology The methodology adopted for the present study was first developed in Nigeria by Faraclas (1987) for the sociolinguistic study of Nigerian Pidgin. Its successful application to the study of Black English in Oakland, California (DeBose & Faraclas, in press) resulted in the corpus of data from which the instances of codeswitching discussed in this paper were extracted. Our general aim was to create a taping situation in which, despite the presence of microphones and researchers, participants would speak BE, rather than (or as well as) SE. To achieve that objective, we needed persons who could speak BE, and I sought to identify and recruit such individuals from my existing network of friends and family members. The persons selected as participants understood the purpose of the study, and the unbiased attitude of the researchers toward BE. In particular, they under-stood that the researchers did not particularly want them to talk 'proper' English. To minimise the 'observer' effect, i.e. the tendency for linguistic informants to suppress ethnically-marked, or socially-stigmatised language use in the presence of a microphone, they were encouraged to talk to one another, instead of to the researchers, and to talk as they would naturally. Each participant was provided a modest honorarium as a token of appreciation for his or her effort. A tacit working hypothesis that guided our data collection efforts and is

< previous page

page_160

next page >

< previous page

page_161

next page > Page 161

supported somewhat by the analysis below is that BE speakers consider SE the appropriate code for use with outsiders or in mainstream settings, and consider BE appropriate for ingroup use among African-Americans. While the presence of the researchers and their recording equipment was known to be a strong situational cue for the choice of SE by bilingual informants, it was hoped that the group format of the recording sessions, and the participants' motivation to fulfil the research objectives would eventually cause them to define the situation as one in which BE was appropriate and acceptable. Analysis Each of our informants exhibits some degree of codeswitching/mixing in his or her linguistic performance, and there is more data than could possibly be covered in the space allotted to this paper. I have chosen to focus on one informant, referred to by her first initial, P., who is by my estimation a balanced bilingual speaker of BE and SE. P. switches effortlessly from one variety to the other according to cues that become apparent in the course of the analysis. P.'s codeswitching was recorded in two different sessions in her home, with the researchers C. and N., her husband F., and her daughter M. present. P. is a college graduate who works as a supervisor for a government agency. She was born in South Carolina and raised, from an early age, in the San Francisco/Oakland Bay area. F. is a high school graduate with some college who is employed on the custodial staff of a local educational institution. He was born and raised in a rural community in northern Louisiana, and has resided in Oakland for most of his adult life. M., who was born and raised in Oakland, is a college student. From listening to the first few minutes of the recording, one might get the impression that P. is a monolingual SE speaker, incapable of speaking BE. As the sessions progress, however, she makes several notable switches to BE. Her last performance, in contrast to the earlier ones, is frequently in BE. For purposes of this analysis, a sentence is considered to be in BE to the extent that it is grammatically compatible with an autonomous description of BE (DeBose & Faraclas in press; forthcoming), contains markers or elements that would be ethnically-marked in SE usage, and does not contain any marked SE features. The following comment made by P. about the subject of one of F.'s stories is an example of a BE sentence: He done lost his poor mind out there, huh? Grammatically, it is an instance of the done-VERB subconstruction of the BE subject-predicate construction. It is incompatible with SE grammar, and has forced Anglocentric scholars such as Labov to postulate ad hoc devices such as subsystems of SE in order to maintain the position that BE is not a separate system from SE. The form done would be ethnically-marked in SE usage. The forms he, his and out are common to BE and SE. The forms lost and mind are realised variably in BE and SE without the final alveolar

< previous page

page_161

next page >

< previous page

page_162

next page > Page 162

stop, and mind is realised in both varieties with and without the /y/ glide following /a/ in the vocalic nucleus. The forms poor and there occur variably in BE and SE without the final /r/. P. pronounces the final /t/ of lost and the final /r/ of poor, but she pronounces mind without a final /d/, and only a trace of the /y/ glide following the vowel /a/, and she pronounces there without final /r/. In the examples cited below, instances of BE are italicised. The beginning of the first tape contains casual conversation between P. and N., picked up while the latter was setting up the recording equipment, and P. did not realise that the recorder was on. Her use of standard English, therefore, cannot be attributed to the 'observer effect'. N.'s status as an outsider to the ethnic community is apparently the cue for P.'s choice of SE here. The classification of her first utterance as SE is supported by the occurrence of can't, pronounced with the low front (nasalised) vowel /ae/, rather than the ethnically-marked cain't, pronounced with the mid front tense (nasalised) vowel /ey/. P: I just can't stay in the bed late. I can't do it! P. pronounces I without the final /y/ glide characteristic of the highest prestige variant of this pronoun, but that does not negate the classification of the utterance as SE, since the monophthongal pronounciation of I is common to BE and SE. P. continues to speak SE with N. after introducing him to M. The topic of conversation is the experience of being a university student. P: I mean there's no talking to them, you know, when, I mean, you're just part of the statistics. That's what, her cousin D went there. He said I'm just a number up here, boy. P. continues to speak to N. and M. in SE until the group gathers at the table around the microphones to officially begin the session. C., at N.'s urging, extemporises about the purpose of the research, and then invites F., whom he knows to be an excellent storyteller, to favour the group with the story of his Uncle Zeke's first trip to California. At the conclusion of F.'s narrative, P. makes the above-mentioned comment in BE. F: . . . a guy came by there from the, comin' down the interstate, and stopped. He asked him where was he goin', he say 'I'm goin' up to Mr Billy Bob's to get me a box of snuff'. (laughter). P: He done lost his poor mind out there, huh? The choice of BE in this instance may be accounted for by the focus of the conversation, and P.'s particular comment, upon family members: F. as the narrator, and Uncle Zeke as the main character of the narrative. P. continues to use BE in commenting upon F.'s subsequent storytelling, and urging him to continue: What about some more of them stories? . . .

< previous page

page_162

next page >

< previous page

page_163

next page > Page 163

Is that y'all church? . . . What about the time they was in the church prayin'? . . . Oh! I done forgot it . . . After F. finishes his stories the topic of conversation shifts to an incident in which M. claims that her mother 'got on her case' about a telephone bill. M: My mother got on my case one time about the phone bills that I did not cause. I was away at school at . . . and somebody had stole my credit card, ah, calling card number. C: How did you feel? What was your reaction, P. . .? In narrating her recollection of that incident, P. opts to speak SE but switches to BE several times. P: When I got the bill I was really mad, and I'm known to be a cusser. Even all the cousins they say 'Yo mother, yo mother kin cuss. Oooo! Yo mother kin cuss!' Cause I can cuss! (laughter). Then I say 'M, what the so and so is goin' on here? What the hell goin' on?' You know, with this phone bill. 'I didn't make no calls, and blah, blah, blah.' And I say 'Well, how in the hell did these calls git on here'. So after I checked it out and, uh, found out that her roommate, (had made the calls) you know, I'd called the phone company and gave 'em the number, and all that stuff. Most, but not all, of the BE in P's response consists of direct quotes of in-group conversations in which the choice of BE might have been expected. What seems to account for the choice of SE in the rest of the above conversational segment is the role of the researcher in eliciting remarks from the informant, who responds primarily to the researcher, with the other informants in the role of audience. P. narrates another incident in which she 'fusses' at another daughter L. who is late coming home from a music rehearsal, and admits having spent time at a local shopping mall 'looking around'. P. again switches to BE to describe her conversation with L. in direct quotation. Her choice of SE for introducing the narrative appears tentative, however, as she alternates several times between wasn't/weren't, was/were, and seen/saw: P: So at about, uh, two o'clock they wasn't back, you know, one o'clock they weren't back. They was supposed to take the bus. Two o'clock they weren't back. Three o'clock they weren't back, you know, till around four or five, I was sitting here at this counter, and I seen, saw 'em, seen 'em come, you know, slippin' by here. I said 'L! Bring yo ass (laughter)! She slip . . ., she's goin' on over to my sister's house. I said 'Where have you been all day! Where have you been!' 'We just went up to the Mall. We was, we just walkin' around. We just lookin' at the Mall, I said 'Lookin' at the Mall? Thugs and hoods hang out at the Mall! I ain't raise no thug and hood, here!' You know. So then she 'Well, we didn't, we wasn't, we just lookin' around and we got us sumpm to eat and stuff'.

< previous page

page_163

next page >

< previous page

page_164

next page > Page 164

Oh I fussed and cussed. I said 'You on punishment now for six months. You cain't look at no TV You cain't do nothin!' So then, uh, a few nights ago me and F went up to the Mall (laughter) . . . P. makes several other code switches during the first session, all of which conform to the pattern we have seen in which she uses SE mainly to narrate family experiences to N. and C. The switches to BE are almost invariably direct quotations. In the second session, P. joins N., C., and F. who have been discussing various topics while waiting for P. to return from taking her youngest daugher K. to a Saturday computer class. Although my intuitive feeling is that she is consistently speaking BE in much of this session, some segments which I intuitively identify as in BE are devoid of marked BE features and could, therefore, be interpreted as switches to SE. P's rather consistent use of BE suggests that she now defines the situation as one in which BE (or frequent switching between BE and SE) is appropriate. F. is speaking to their two-year-old niece D. when P. enters. F: Another doughnut? What you gon do wit another doughnut? P: Y'all still talkin? (laughter) Oh, God! I say, aw they still goin! F: You kin, you kin git back on now. P: I gotta run K up to her school. They have that computer, um, class set from ten to twelve up there to Howard today. They gon have computer science . . . F: Uh, You kin git back on in there. P: . . . math clubs. They sick of me! (laughter) I wear 'em out. Uh, They let the water boil on down, huh? F: We made more coffee. N: That's the second one. P: Oh. That's the second one? (laughter) F: I guess I left (unintelligible) P: Naw, Its not too (unintelligible). F: Oh, yeah. You kin go. P: Okay, let's go. (unin.) Let's run up here real quick cause (unin.) They gon be on time. They ain't gon be on time. C: Dispense the honorarium. P: Say what? C: Dispense the honorarium. P: I was tellin 'em that after. I say I been interviewed this morning . . . N: Mother and daughter and (unin.). We gotta give a little extra for, uh, this premiere performance. P: Who is this? (laughter) N: It's nothin actually (muted comments). P: Who did a performance? F. . .? N: Yep. (laughter) F: We been all over the place.

< previous page

page_164

next page >

< previous page

page_165

next page > Page 165

N: Yeah. We tired. But I think we tired you out, sorry. (laughter) P: I think he like that. F: Naw. I been all over the place. P: Did I do all right? (laughter) N: It's nothin much. It's not gonna get you to Louisiana for, ah Christmas. That's for sure. F: Guess P gon have to pay my way down there. P: Guess I have to do 'im, huh? (laughter) I didn't know M was so . . . She miss N (her brother) so (laughter). Get all their family feelings out. You guys can charge us for counselling (laughter). She hadn't heard her father's stories before . . . That was fun though. F: Yeah. You shouldn't a left, there. P: I better not say that, they take my money back (laughter). I'm'on run L up here for this computer class. Not L, K. They suppose to start the Dad's club. I think it's starting now so I'm'on run up here and see what they're doin, it's about fifteen after, they suppose to have from ten to twelve . . . They gon try to start some Saturday activities at the school. F supposed to be in the Dad's club F: Cain't keep me runnin too much, now . . . Keep me runnin too much, on the go all the time. P: All the time, huh? I was gon take D up there let her run around get all her energy out. But I bet she don't have no clothes on. M!, M! Are you on the phone? . . . Whether P's behaviour at this point is seen as speaking BE consistently, or switching frequently between BE and SE, she displays a degree of BE proficiency that is largely absent from her earlier behaviour. Her behaviour in this session is apparently the way that she normally acts when she is 'at home' with her family. Conclusion The evidence considered in this paper is striking counter evidence to the claim that BE is spoken mainly by poor and uneducated persons. We must await the results of future research for a definitive assessment of the prevalence of BE/SE codeswitching among African-Americans, but my impression as a member of the speech community is that BE is frequently spoken by middleclass persons. That impression is consistent with the fact that many educated, middle-class African-Americans begin their lives in predominantly-black urban ghettoes or rural southern communities where BE is the normal medium of everyday communication. The subsequent upward mobility of such persons usually entails becoming bilingual speakers of BE and SE rather than replacement of BE by SE. The linguistic behaviour of P., therefore, may be typical of middle-aged, middle-class black females like herself with southern roots, but raised outside the south.

< previous page

page_165

next page >

< previous page

page_166

next page > Page 166

The working hypothesis is that members of the African-American speech community consider SE appropriate for communicating with outsiders is supported by P.'s choice of SE in the first session to address the researchers. Her switching to BE to quote herself and report the speech of relatives in in-group situations is consistent with the hypothesis that BE is considered appropriate for communication within the ethnic group. Her choice of BE during the second session to act out her present self suggests that participants had by then achieved a sufficient degree of familiarity with the researchers to be able to act naturally in their presence, oblivious to the fact that the tape recorder was turned on. To the extent that frequent switching between BE and SE occurs in in-group situations, the notion of situational switching does not adequately account for all instances of codeswitching by African-Americans. It may be the case that among certain AfricanAmericans it is normal and acceptable to speak either BE, SE, or both in certain situations, and that frequent codeswitching in such situations is the unmarked choice (Scotton, 1988). One methodological obstacle to pursuing such an hypothesis is posed by the extreme similarity of BE to SE, and the practical difficulty of establishing that one code or the other is the matrix system for a given instance. The experience of analysing the data for this paper indicates that speech community members have accurate intuitions of codeswitching which can be relied upon for preliminary identification of the code(s) spoken within given segments of speech. Systematic analysis of the lexical, grammatical, and phonological details of code tokens which were initially identified intuitively can result in a set of rigorous criteria for code identification to be used in future studies of codeswitching between BE and SE. Continued study of the case of BE from a perspective which considers the language situation in black America analogous to that of a bilingual community can lead to significant advances in the general study of closely related languages in contact. Work in progress on the BE tense-mood-aspect system is showing that the vernacular language of African Americans is best accounted for by a different autonomous grammar than that of SE (DeBose & Faraclas, in press; forthcoming). The present findings of codeswitching indicate that contrary to popular belief the two codes are not maintained by separate segments of the speech community, but are often used alternatively by the same speakers. References Blom, Jan-Petter and Gumperz, John J. (1972) Social meaning in linguistic structures: code-switching in Norway. In John J. Gumperz and Dell Hymes (eds) Directions in Sociolinguistics: The Ethnography of Communication. New York: Holt-Rinehart. DeBose, C. and Faraclas, N. (in press) Getting to the roots of the tense/aspect/and copula systems in Afro-American. In Salikoko Mufwene (ed.) Africanisms in Afro-American Language Varieties. Athens: University of Georgia Press. (forthcoming) The role of the feature stative in Black English Predicates. Annual Meeting of the Linguistic Society of America, New Orleans, December 1988.

< previous page

page_166

next page >

< previous page

page_167

next page > Page 167

Dillard, J. L. (1972) Black English: Its History and Usage in the United States. New York: Random House. Faraclas, N. (1987) Creolization and the tense-aspect-modality system of Nigerian Pidgin. Journal of African Languages and Linguistics 9, 45-59. Heller, Monica (ed.) (1988) Codeswitching: Anthropological and Sociolinguistic Perspectives. Berlin: Mouton de Gruyter. Labov, William A. (1972) Language in the Inner City: Studies in the Black English Vernacular. Philadelphia: University of Pennsylvania Press. Pfaff, C. (1979) Constraints on language mixing: intrasentential codeswitching and borrowing in Spanish/English. Language 55, 291-318. Scotton, Carol Myers (1988) Codeswitching and types of multilingual communities. In P. Lowenburg (ed.) Georgetown University Round Table on Languages and Linguistics. Washington: Georgetown University Press (pp. 61-82). Taylor, Orlando L. (1975) Black Language and what to do about it: some Black community perspectives. In Robert L. Williams (ed.) Ebonics: The True Language of Black Folks. St. Louis: Institute of Black Studies (pp. 29-39). (1983) Black English: an agenda for the 1980's. In John Chambers Jr. (ed.) Black English, Educational Equity and the Law. Ann Arbor: Karoma Publishers (pp. 133-43). Weinreich, Uriel (1953) Languages in Contact. The Hague: Mouton. Wolfram, Walter A. (1969) A Sociolinguistic Description of Detroit Negro Speech. Washington, D.C: Center for Applied Linguistics.

< previous page

page_167

next page >

< previous page

page_168

next page > Page 168

< previous page

page_168

next page >

< previous page

page_169

next page > Page 169

Codeswitching, Convergence and Compliance: The Development of Micro-Community Speech Norms Susan Meredith Burt English Department, Illinois State University, Illinois, USA Abstract In conversations between bilinguals, each of whom is a learner of the other's language, two different local patterns of codeswitching may emerge. In the pattern of mutual convergence or complementary schismogenesis, each speaker attempts to speak the interlocutor's native language. Conversations characterised by this pattern reveal that the interlocutors are having difficulty agreeing on a shared norm for conversational code choice. On the other hand, in conversations characterised by compliance, each speaker agrees with and complies with the interlocutor's code choice, even if the interlocutor chooses to change the code during the conversation. In conversations which exhibit compliance, the speakers are successful in establishing a shared code or a mode of code choice for future conversations. It is argued that a pattern of compliance is ultimately more accommodating than convergence, contrary to the claims of Speech Accommodation Theory. Background and Introduction Within the past two decades, research on the social meaning of codeswitching has developed along three main lines: (a) research within the framework of interpretive sociolinguistics as in Blom & Gumperz (1972), Gumperz (1976, 1982) and Heller (1982), (b) work on code choice within the frame-work of Speech Accommodation Theory (SAT) as in Bourhis (1984; 1985), Bourhis, Giles & Lambert (1975), Genesee & Bourhis (1982; 1988), Giles & Powesland (1975), Giles & Smith (1979), Giles, Taylor & Bourhis (1973), and (c) work within the markedness model of C. M. Scotton. Authors within the interpretive sociolinguistics framework (a) see code-switching as a part of the speech repertoire available to bilingual speakers, along with other features, such as pause, emphasis and variations of discourse structure. The social meaning of such features is learned by speakers through participation in the speech network of the community which shares norms

< previous page

page_169

next page >

< previous page

page_170

next page > Page 170

of interpretation for the choice of feature variants. Of significant concern to interpretive sociolinguistics are the misunderstandings produced when speakers attempt to interpret the speech of members of different communities, whose norms they do not share, or even know. However, with the exception of Heller (1982), many studies of codeswitching within this frame-work have been confined to speakers of homogeneouseven if bilingualcommunities. The framework of speech accommodation theory (SAT) (b) is specifically concerned with the problems of interaction between members of different groups. Genesee & Bourhis (1982) have presented groups of judges with constructed conversations in which codeswitching occurs, and asked the judges for evaluations of the personality characteristics of the speakers (who were bilingual actors). Using the matched-guise technique, they have discovered that speakers tend to be more positively evaluated when they accommodate to the language or linguistic characteristics of the interlocutor. For example, English Canadians are rated more positively when they at least attempt to speak French to French Canadian interlocutors. Indeed, within SAT in general, it is expected that interlocutors will welcome speakers' attempts at convergence or, making one's own speech more like that of one's interlocutor (with some exceptions: convergence in some cases may be seen as patronising (see Platt & Weber, 1984)). Both interpretive sociolinguistics and SAT assume that, in general, listeners within a speech community will share norms for the interpretation of discourse phenomena such as code choice and codeswitching, but listeners from different groups may not necessarily do so. A third approach (c) is represented by Scotton (1983, 1988). Scotton uses an explicitly Gricean approach, using maxims, and the interpretation of speakers' following or flouting them, to describe the markedness of certain code choices for certain situations and configurations of speakers. Codes are not marked a priori, but acquire markednessand with markedness, their social meaningas speakers mutually develop norms for the interpretation of their future conversational choices. It is within Scotton's general framework that I would like to place this paper, although I will also discuss the SAT notion of convergence. This paper deals with the process of developing norms for the speech micro-community which consists of two speakers with different native languages, who nevertheless share two languages, because each is a learner of the other's native language. Within such a dyad, absent any third person or other external factor favouring one language, there is no clearly unmarked code. Although the larger setting may favour the choice of one language over the other (e.g. English is the expected, favoured language in Illinois), committed language learners look for occasions to use their second language, so setting alone may not be enough to determine code choice, particularly if the interlocutor is known to be a native speaker of the learner's second language. Furthermore, if speakers are matched for sex, age, and status,

< previous page

page_170

next page >

< previous page

page_171

next page > Page 171

these sociolinguistic factors will also play no role in the determination of code choice. Consequently, an initial conversational exchange between two such speakers will be 'non-conventionalised' (Scotton, 1988: 157); that is, it is not clear what the code of choice should be. In non-conventionalised exchanges, such as these conversations, code-switching indeed plays an exploratory role, as Scotton (1983: 125) claims. But in the absence of shared norms for code choice and for the interpretation of choices (the absence of'parochial pragmatics,' as Green (1990) puts it), speakers are thrown back on universal pragmatic principles for the interpretation of their exploratory moves. However, candidates for universal pragmatic principles, such as those proposed by Brown & Levinson (1987), or by Scotton (1983), offer more than one interpretation of any individual code choice. Given the pragmatic ambiguity of code choices in this non-conventionalised situation, the road to the establishing of shared norms may be a rocky one. The pragmatic dilemma this causes speakers is described in the following section. However, I will suggest that codeswitching plays a role in the development of shared norms, in the event that these speakers continue their acquaintance. The phenomenon of speech accommodation through the mechanism of convergence might be expected to play a role in the establishment of a speech micro-community. However, I hope to show in the third section that the SAT notion of convergence describes a phenomenon that is not universally positively evaluated by speakers. The fourth section will introduce the term compliance to describe sequences of codeswitching which ultimately show greater accommodation of speakers to interlocutors than sequences characterised by convergence. I shall suggest that conversations in which speakers tend towards compliance may prove to be those in which speakers are most successful in establishing shared norms for future conversations. The data in this paper come from the investigation reported in Butt (1990), in which pairs of previously unacquainted studentsone American student of German, and one German student learning English were introduced to each other, left alone in a university office (at Illinois State University) with a tape recorder, and told to get to know each other. The resulting half-hour conversations were transcribed, and participants were later individually debriefed. The Pragmatic Dilemma for Bilingual Speakers Burt (1990) has described the dilemma of a native speaker of language A, learner of B, who meets a native speaker of B, learner of A. In the initial conversation between this pair, either language might seem to be a candidate for use as the 'unmarked' code for conversation. However, in reality, a number of factors play a role in choosing the code: the proficiency of each speaker in her second language, speakers' assessments of relative proficiency, and the desires of both speakers with regard to language choice (these factors

< previous page

page_171

next page >

< previous page

page_172

next page > Page 172

probably outweigh the additional factor of setting, as mentioned above, as long as only a pair of speakers is involved). At the outset of the conversation, all of these factors are unknowns. In the light of SAT, we could expect speakers who are proficient enough to do so to attempt to speak their second language, in an attempt to converge. But, while convergence attempts can be interpreted as polite, each speaker must eventually face the possibility that the other speaker may want to use her second language rather than her first. Burt (1990) describes this dilemma in terms of Brown & Levinson's (1987) concept of 'face wants'. According to Brown & Levinson, people have two basic kinds of wants: to be unimpeded in their actions, and to receive approval from others. These wants are satisfiable only by acts (including speech acts) by others; those acts which satisfy the first want constitute 'negative politeness,' whereas those which satisfy the second are 'positive politeness'. Enabling an individual to speak her own language can be seen as an act of negative politeness in that it removes some of the possible impediments to her expressing herself as she wants. On the other hand, giving a person the opportunity to speak her second language with native speakers may constitute positive politeness, in that she may feel approval for her accomplishment in being able to do so. Of course, it is not possible for a speaker to know immediately whether the negative face want or the positive face want is uppermost in a given interlocutor's mind. Thus, the individual who wants to be polite cannot know whether convergence to the point of speaking the interlocutor's native language is polite or not. The interlocutor faces a parallel dilemma: should she accede to the speaker's putative negative face want to have her own language spoken, or should she accede to the speaker's possible positive face want of being accepted as a competent speaker of her second language? The situation is not only non-conventionalised with a vengeance; it is, in fact, a double bind. As a speaker decides whether to pay negative politeness to the interlocutor by converging, or whether to pay positive politeness to the interlocutor by speaking her own first language, codeswitching as an exploratory choice (Scotton, 1983: 125) may well emerge. The dilemma faced by speakers can be similarly described in terms of the maxims within the model proposed by Scotton (1983) that govern code choice, the deference maxim and the virtuosity maxim. The deference maxim, 'show deference in your code choice to those from whom you desire something' (Scotton, 1983: 123), guides speakers to converge to the native language of the interlocutor. On the other hand, the virtuosity maxim (Scotton, 1983: 125) instructs speakers to 'make an otherwise marked choice whenever the linguistic ability of either S or A makes the unmarked choice for the unmarked RO set in a conventionalized exchange infelicitous.' In a situation in which one speaker discovers the limited proficiency of the other, the virtuosity maxim licenses choice of the less-fluent speaker's native language.

< previous page

page_172

next page >

< previous page

page_173

next page > Page 173

The explicitly pragmatic approach of Scotton's model, however, highlights the dilemma for the speaker. If speakers are aware (in some sense of universal pragmatics) of both maxims, two possible interpretations are available for any individual code choice. If the native speaker of A converges to B, the native speaker of B may interpret this as deference towards her, or she may interpret it as action taken because of her inadequate virtuosity in language A. If the native speaker of A fails to converge to B, the native speaker of B may interpret this as a lack of deference to her or as a compliment to her abilities in A. Thus, each choice of code is open to a systematic ambiguity of pragmatic interpretation. How a speaker will choose between the available interpretations probably depends on her own intentions and desires for the conversation. The interpretations and reactions of actual bilingual speakers placed in this situation are given in Burt (1990). Again, to the extent that speakers are aware that any choice of the conversational code is open to two interpretations, speakers may try to hedge their implicative bets by codeswitching. Scotton's model further illuminates the dilemma faced by language learners in this situation, with the notion that codes function as metaphors for sets of rights and obligations (RO sets) that hold between the participants. What are the relevant RO sets here? Regardless of whether language A or language B is chosen, one of the dyad will be cast in the role of native speaker, while the other will take on the role of non-native speaker. The role of native speaker entails the right of acting as authority in deciding the right way to speak the chosen language, and the obligation of offering help in the form of instruction to the non-native speaker. The role of non-native speaker entails the right to help, patience, simplified speech or perhaps overt instruction from the native speaker, and the obligation to submit to correction from the native speaker if this is forthcoming. Again, members of a pair of speakers of the type described here may conflict in terms of the RO set each wants to assume. If one speaker wants to take on the RO set of the non-native, and the other wants to 'play teacher,' it may be easy to agree on a conversational code. But if both want to play teacher, or if both want to lay claim to the other's competence to instruct, the conflict in wants will be reflected in a conflict in code choice. Codeswitching, under these circumstances, may devolve from an exploratory choice to a vehicle for speaker conflict, as described in the next section. Mutual Convergence or Complementary Schismogenesis? Scotton (1983: 122) argues that if codeswitching is an unmarked code for interaction, each individual switch may carry little social meaning. In the double-bind situation described here, this is obviously not the case: a switch is likely to carry a great deal of meaning, i.e. to be pragmatically ambiguous as described above. Thus, this section and the next will show that the local patterning of switches within a conversation is revealing, and that different

< previous page

page_173

next page >

< previous page

page_174

next page > Page 174

patterns of switchcs allow inferences as to the participants' language preference, and their evaluations of interlocutor proficiency. Scotton (1988: 156) argues that 'codeswitching' may include the introduction of individual lexical items such as in (1): (1) Lynn: ich hab' ein Jahr studiert, meine Junior Year Abroad, und . . . dann bin ich zuruckgekommen und habe meine Masters gemacht und dann wieder nach uh Osterreich mit einer Fulbright, eine Assistentin zu werden, auf ein Gymnasium. (I studied one year, my Junior year abroad, and I came back and did my Master's, and then [went] to Austria again, with a Fulbright, [in order to] become an assistant at a high school.) However, in this article, we are interested in code changes that last over major stretches of the utterance. Thus, there are two major types of switches that concern us: (a) the extended use (over most or all of an utterance) of a language other than that which the interlocutor last spoke, and (b) the extended use of a language other than that which the speaker herself last spoke. Sequences of switches of these different types, it will be argued, are crucially different in their pragmatic significance. First, we are concerned with sequences of type (a) switches, such as in (2): (2) Speaker 1: A Speaker 2: A Speaker 1: B (switch 1) Speaker 2: A (switch 2) In such a sequence, speaker 1 attempts to change the conversational code from language A to language B, but speaker 2 does not follow her lead. Both switch 1 and switch 2 are type (a) switches, in that each speaker switches to the language other than that which the interlocutor last spoke. A series of such type (a) switches can reveal conflict between speakers as to preference for conversational code, as in the passage in (3), from one of the four half-hour conversations recorded (in the investigation reported in Burt (1990)) between German and American university students matched for age and sex. In one of these conversations, conflict as to code choice arose between Stacy, an American university student majoring in German, and Clara, a German university student spending a semester in the U.S., in part to improve her English (parentheses enclose English translations of German utterances). (3) Clara: So my first minor is economics and the second minor is law. Stacy: So, sehr viel (laughs) (So, a great deal) Clara: What are you studying now? Stacy: So habe ich dich gefragt umm welches Jahr Sie studieren. Umm, zweiten oder dritten? (I asked you which year you are studying [in]. Second or third?)

< previous page

page_174

next page >

< previous page

page_175

next page > Page 175

Clara: No, I am in the fourth year. Stacy: Ich auch (me too). Umm. I studied German, Education . . . Clara: Ah so. Stacy: Next year, I'll be seeing what I can do with it. In this conversation, Stacy had already tried earlier to switch the conversational code to German, but without success. Passage (3) was particularly revealing in that a number of these attempts follow closely on one another. Stacy's interlocutor, Clara, was firm in her commitment to speak English (except for her German back-channel utterance 'Ah so'). Thus, each speaker attempted to speak the other's language, in a series of type (a) switches, that is, each speaker switched the code away from that chosen by the interlocutor in the immediately preceding utterance. 1 In the light of SAT, such attempts might be seen as mutual convergence. But in the light of the conflicts described in section 2, this pattern looks instead like a case of complementary schismogenesis (Bateson, 1958: 176). In complementary schismogenesis, each speaker (or actor) models the type of behaviour which she would like to see her interlocutor display, but since the two individuals differ on the values assigned to the behaviours in question, each continues to behave in precisely the opposite fashion to that which her interlocutor wishes. Evidence that this type of mechanism was at work in this situation comes from the debriefings of two American students (Stacy and Terry), both of whom noted that hearing German-accented English motivated them to try to speak German. In the passage in (3), Stacy eventually gives up trying to change the code of conversation, probably because to continue to do so would escalate the conflict between the two speakers. What is important here is that, SAT to the contrary, attempts at mutual convergence are not necessarily accommodating, if the speakers in question are both learners of each other's language. Indeed, in an individual debriefing after this conversation, Clara reported that she found that Stacy's attempts to speak German with her 'remind[ed] me of a school situation;' she also described how previous conversations in German with Americans lacked depth and interest, and how her own 'willingness to speak English here is very great.' Clara found speaking German in this case to be an impediment to her goals in the conversation, which were to get to know another person and to speak English. Stacy, on the other hand, during her debriefing, stated that her main goal for the conversation had been to practise her German. She reported that she had felt somewhat disappointed in herself in not speaking German more during the conversation with Clara. Thus, the pattern of complementary schismogenesis in (3) reflects the conflicting wants of these two speakers for the conversation, rather than revealing a mutual convergence towards each other's preferred language. Indeed, a pattern of non-convergence may occasionally be more likely than a pattern of mutual convergence to evolve into an unmarked code

< previous page

page_175

next page >

< previous page

page_176

next page > Page 176

within a speech micro-community of learners of each other's languages. One linguist, Gregory Richter (personal communication), has told me of his experience learning Chinese. Richter understood Chinese but did not feel comfortable speaking it; his Chinese friends felt more comfortable speaking Chinese than English, which they none the less understood. In conversations, each of these speakers spoke his own language, confident that the interlocutors would understand. Although speakers failed to converge, they agreed upon a mutually acceptable unmarked code consisting of type (a) switches, that is, switches away from the interlocutor's code of choice. Thus, type (a) switches are not necessarily conflictual, but in a pattern of complementary schismogenesis as in (3), they may become so. Richter and his Chinese friends developed micro-community speech norms in which type (a) switches were characteristic of the unmarked code. What about Stacy and Clara? When Stacy was asked what language she thought she would use with Clara if she got another chance to talk with her, Stacy reported that she would greet Clara in German; although she could sympathise with Clara's wish to speak English, Stacy hoped that if they became friends, Clara would accede at some point to Stacy's wish to speak German. Clara, on the other hand, reported that only if she met Stacy at the university's German club, where German is supposed to be spoken, would she speak German with her. Otherwise, she would greet her in English and converse in English. Stacy's and Clara's projections reflect the lack of any developing consensus between them about an unmarked code, a lack of consensus which is encapsulated in example (3). These two speakers have a long way to go towards developing a mutually acceptable micro-community speech norm. Compliance This section will be concerned with a contrasting pattern of codeswitching, a pattern in which the speaker complies with the interlocutor's language choice, in that she chooses to speak the same language that the interlocutor initiates conversation in, or the language the interlocutor switches to. A pattern of compliance is shown in (4), in contrast with the pattern of (2), above, which characterises complementary schismogenesis: (4) Speaker 1: A Speaker 2: A Speaker 1: B (switch 1) Speaker 2: B (switch 2) In this case, speaker 1 signals, by switching, her wish to speak language B. Speaker 2 follows her lead, and in also switching to language B, complies. A pattern of codeswitching which is characterised by compliance may contain many of these type (b) switches, but relatively few sequences of (a) switches; that is, while a speaker may switch from the language she herself just spoke,

< previous page

page_176

next page >

< previous page

page_177

next page > Page 177

she does not switch from the language her interlocutor just spoke. The long series of (a) switches (see (2) above) which characterise complementary schismogenesis are crucially absent in conversations characterised by compliance. In another conversation recorded between university students, the German student Brigitta shows instant compliance when her American interlocutor Lynn initiates a change in the conversational code: (5) Brigitta: How long did you take German? Just, just now? Lynn: No, I'm, I'm, um . . . I took it all through high school, and I'm now a graduate student, and I teach the first level, so ich kann schon etwas. Ein bibchen. (I can [speak] some already. A little.) Brigitta: Warst du schon mal in Deutschland? (Have you ever been to Germany?) Brigitta complied with Lynn's initial switch to German by switching to German herself, and indeed, the rest of this conversation was in German. When Lynn switched from English to German mid-utterance, her switch was both a type (a) and a type (b) switch, in that she switched from the language that both she and her interlocutor had just used. What is salient in defining this as a case of compliance is that Brigitta's subsequent utterance is a type (b) switch, that is, she uses a different language than she had used before, but it is not a type (a) switch, in that she does not switch away from her interlocutor's most recent language choice. Brigitta accommodates to her interlocutor, not by speaking the interlocutor's native language, but by choosing the code that the interlocutor seems to prefer. Lynn and Brigitta switched codes only once in their half-hour conversation. For this particular dyad, it was easy to find an unmarked codeGerman. Debriefings confirmed the likelihood that this state of affairs would contine: Lynn stated that she would speak German with Brigitta; Brigitta stated that she might say 'Hello,' but would not be surprised if Lynn spoke German. While both speakers were open to the possibility of speaking English in later conversations, both clearly felt that German would work well as a shared micro-community (in this case, dyadic) code. This dyad was well on the way to development of a shared microcommunity speech norm, and the early incidence of compliance in the conversation indicates this. But not all bilingual learners will find their way to a shared code so easily; Stacy and Clara are a case in point. Nevertheless, over the stretch of a conversation, we may note whether speakers comply with each other's exploratory choices, or with choices made because of expression or comprehension difficulties, by noting whether speakers tend eventually towards compliance in their codeswitches as the conversation progresses. In the conversation between the pair Todd and Axel, for example, the two speakers switched a total of 12 times in the half-hour conversation (though some of these switches are single backchannel words, such as an isolated ja, or borrowed particles such as OK). The first, exploratory

< previous page

page_177

next page >

< previous page

page_178

next page > Page 178

switches seem to threaten to develop into complementary schismogenesis, but these two speakers tackle and deal with the code choice overtly: (6) Todd (a): Guten Tag. Ich bin der Todd. (Hello. I'm Todd.) Axel (b): Ja? I'm Axel. Todd (c): Und . . . wo kommst du her? (And . . . where are you from?) Axel (d): Umm . . . shall we speak German? OK. Todd (e): Wie du willst. Or English. It's either one. (As you like.) Axel (f): Ja. Can youI would in German . . . or Todd (g)): Ja. Sprechen wir ja Deutsch. (Yes. Let's speak German.) Axel (h): OK. Ja, Ich komme also aus Deutschland . . . (Yes. I'm from Germany . . .) Axel switches to English in turn (b), a switch which Todd does not comply with in (c). Axel follows with another type (a) switch (to English) in (d), and Todd similarly does a type (a) switch in (e). If one discounts the significance of Axel's ja's in (b) and (f), the lack of compliance in this sequence is apparent. The exchange gives the impression of a verbal Alphonse and Gaston routine, until Axel complies with Todd's switch (g) with a German utterance in (h). However, later in the conversation, a different picture emerges, in that initial switches of the conversational code are complied with by the interlocutor. In this conversation, Todd took the lead (for example, Todd initiated 22 topic changes, Axel only 9), including choosing the conversational code. Thirteen minutes after both speakers had established German as the language for the conversation, Todd changed the code to English because he thought this was part of the 'task' of the experiment: (7) Axel: Joa, ich meine . . . umm, so für ein Semester macht's einem nicht so viel aus, aber wenn's länger, weiss ich nicht. (Yeah, I mean . . . umm, for a semester it doesn't bother me, but if it [were] longer, I don't know.) Todd: Also, das ist mir interessant. [inaudible passage] Umm, we probably should speak English for a while, too, just . . . (Well, that's interesting.) Axel: OK. Todd: Umm, let's see, what should we talk about? What kind of music do you listen to? What kind of music do you like? Axel: Oh, uh . . . umm, Renaissance and Baroque? Six minutes later, Todd changed the code back to German, because a German word came to mind before its English equivalent: (8) Todd: I thright now I only have two dollars, but, maybe some day. Axel (laughs): Two dollars? Todd: Umm. Axel: In your pocket?

< previous page

page_178

next page >

< previous page

page_179

next page > Page 179

Todd: Uhuberhaupt. Nur noch zwei Dollar. (At all. Only two dollars left). Todd then began a long sequence in German detailing his impecuniousness, during which Axel contributed only back-channel responses. But when Axel took a full turn again, he spoke German, complying again with Todd's switch. In both these sequences Todd changed the code, and Axel complied with his change. As the conversation progressed, any tendency towards complementary schismogenesis disappeared, and Axel accommodated to Todd, not by converging to English, but by complying with whichever code Todd chose. In mutually creating this conversational modus operandi, these speakers began moving towards a micro-community speech norm characterised by codeswitching and compliance. A similar trend appeared in a fourth conversation, between the American student Terry and her German conversation partner Daniella. The beginning of the conversation (9) was characterised by some exploratory switches: (9) Daniella: So you speak German? Terry: Yeah, I'm studying it. I Daniella: Ah, so. Terry: I'm kind of underconfident in my ability. Daniella: Mmm (both laugh). Terry: Um, the whole summer I didn't use it. Daniella: Mm. Terry: at all, sodo you want to try it in German? Daniella: Uhit's egal. Es ist egal, also . . . (it's OK either way. It is OK either way, so . . .) Terry: OK. Daniella: Wenn du es möchtest? (If you want to?) Terry: yeah. Daniella: OK, um. Ja, was liest du alsona in der Schule, also in der Universitäit, zu Lektüre? (Yeah, what are you reading, then, well, in school, so at the university, in lectures?) Terry: Um, du musst langsam sprechen! (You have to speak slowly!) (both laugh) But these speakers quickly progressed to the point of switching codes when the switch aided comprehensibility: (10) Daniella: Und habt Ihr da Einzelzimmer oder müsst Ihr dort auch ein Zimmer teilen? (Do you have single rooms there or do you have to share a room?) Terry: Wie, bitte? (Excuse me?) Daniella: Do you have a room of your own there or do you have to share a room with uh

< previous page

page_179

next page >

< previous page

page_180

next page > Page 180

Terry: I share a room. Wie sagt man roommate auf Deutsch? (How do you say 'roommate' in German?) Daniella: Uh, uh, we don't have roommates. Shortly after this stretch of conversation, Daniella switched because she suspected that her interlocutor wanted her to do so: (11) Daniella: I was uh not so umwe are only two persons in one room, and last year I was into Champaign, and they had to shareum um my host family's daughtershe had to share a room with three other girls and Terry: Uh huh Daniella:it was quitemust be making you nervous sometimes or so. Terry: Umhm. Daniella: Und warst du schon mal in Deutschland? (And have you ever been in Germany?) Terry: Um, ja. Um, in vierundachtzig war ich Austauschschülerin in Bremen. (in '84 I was an exchange student in Bremen.) Later in the conversation, when one speaker switched to aid self-expression, the interlocutor complied without hesitation. (12) Daniella: Und wo kommst du her? (And where are you from?) Terry: Von Quincy. Quincy. (From Quincy.) Daniella: Kenne ich auch nicht, so. (I don't know [the place] either.) Terry: Quincy istuh, liegt west yon hier, um . . . right on the Mississippi River. (Quincy isuh, lies [to the] west of here,) Daniella: Is it in Illinois or another state? When the same speaker, Terry, initiated a switch back to German, Daniella again complied (despite Terry's brief switch back to English at the end of her utterance): (13) Daniella: Uh . . . And, uh, do you often uh travel through America, to the west coast or so? Terry: I have never been uh west, anywhere. Meine Grosseltern haben in Florida gewohnt, und im März sind wir nach Florida gefahren, um sie zu besuchen. Aber jetzt wohnen siejetzt ist mein Grossvater tot, und mein Grossmutter wohnt irgendwo sonst so. Wir waren nach Florida on the beach. (My grandparents lived in Florida, and in March we went to Florida to visit them. But now they livenow my grandfather is dead, and my grandmother lives someplace else. We were in Florida on the beach.) Daniella: Ja, wir waren letztes Jahr auchumm in Miami, und Key Westdas war eine schöne Fahrt . . . (Yeah, we were in Miami and Key West last year. That was a nice trip).

< previous page

page_180

next page >

< previous page

page_181

next page > Page 181

Indeed, most of this conversation, after the first 30 seconds, is characterised by compliance by the interlocutor when a speaker switches codes. That Terry and Daniella were quite successful in beginning to develop a micro-community speech norm was also demonstrated in their debriefings. When asked what language she thought they would speak if they met again, Daniella reported that she and Terry had met again, and had spoken English. She speculated that future conversations would be 'a mixture'. Terry speculated that a 'fleeting conversation' would probably be in English, but that she would like to speak German. Since Daniella's debriefing showed that she was quite aware of Terry's eagerness to speak German, and since Daniella switched once to German specifically to accommodate this (example 11), it seems that these speakers have already reached a mutually acceptable micro-community speech norm, one characterised by switching followed by compliance. It is possible to determine whether a conversation trends towards compliance or not, by examining the series of switches and subsequent utterances. In Lynn and Brigitta's conversation of course, the only initiating switch is followed by a complying switch. In Todd and Axel's conversation, if we discount the significance of back-channel utterances, switches internal to an utterance (so that the interlocutor has no chance to respond) and borrowed OK's, the series looks like this (Y indicates a switch which is complied with; N indicates one which is not complied with): (14) N N N Y N Y Y Y It is significant that the first six switches in the series occurred within the initial 30 seconds of the exchange. Thereafter, all other switches were complied with: this looks like a clear trend over time towards compliance. Similarly, in the conversation between Daniella and Terry, the series looks like this: (15) N N N N Y Y Y Y Y

< previous page

page_181

next page >

< previous page

page_182

next page > Page 182

Y N Y Y Y Y The first four switches, which were not complied with, all took place within the first 30 seconds of the conversation. Only once after the initial exchanges, at seven minutes into the half-hour conversation, did a speaker in this dyad fail to comply with an interlocutor's codeswitch. What then of the series produced by the conversation between Stacy and Clara, the speakers who seemed unable to find a mode of code choice that both were happy with? The sequence of switches, and whether or not those switches were complied with, looks like this: (16) N N N Y N N N N N/Y Y N/Y Y N/Y What makes the sequences of switches between Stacy and Clara different is a set of three switches marked N/Y: these were utterances in which Stacy tried to switch to German, but then switched back to English, because she was having trouble expressing herself, before Clara had a chance to reply. Clara invariably answered in English, thus failing to comply with Stacy's obvious wish to speak German, but complying instead with her default to English (contrast with Daniella's compliance in example (13)). It is significant that these failed attempts on Stacy's part to switch the code occur throughout the conversation: although Clara converged to English, she did not really accommodate Stacy's obvious wish to speak German. Thus, this conversation does not show a trend towards compliance. This lack of a trend towards compliance is reflected in the lack of consensus on the part of these two speakers (described above) on what language any future conversations would be in. In summary, in conversations characterised by compliance, speakers tend to follow their interlocutors' choice of code: codeswitching is characterised by a mutual give and take which is ultimately more accommodating than

< previous page

page_182

next page >

< previous page

page_183

next page > Page 183

simple convergence. Indeed, the willingness to switch codes within the conversation, rather than to stick with a simple convergence strategy seems essential to real speech accommodation, at least for the category of speakers described here. In contrast, in conversations where both speakers insist on converging to the interlocutor's native language, a conflictual pattern of codeswitching, complementary schismogenesis, may evolve. Conclusions Pragmatic principles, such as those proposed by Brown & Levinson (1987) and Scotton (1983), when applied to the speaking situation in which each speaker is a learner of the other's native language, lead to the realisation that code-choice in this situation is potentially pragmatically ambiguous, and consequently problematic. In terms of Brown & Levinson (1987), choosing one's native language as the conversational code may constitute payment of positive face to the interlocutor, or it may signal a lack of deference (negative face) to her. Choosing the interlocutor's native language as code may constitute an act of deference, or it may be interpreted as an indirect indication of her inadequate proficiency in her second language. Scotton's (1983) deference maxim and virtuosity maxim allow for these same inferences. In terms of the sets of rights and obligations each code symbolises, choosing to speak one's native language involves taking on a different RO set than choosing to speak the language one is learning. Given the systematic pragmatic ambiguity of any code choice, and the potential conflict between speakers for the RO set of their preference, it is not surprising that the first minutes of conversations between learners are characterised by codeswitching. It should also be clear, given this maelstrom of pragmatic considerations, that, contrary to the claims of Speech Accommodation Theory, convergence by a speaker to her interlocutor's language will not necessarily be interpreted positively. In one conversation reported on here, a pattern of what might have been seen as mutual accommodation (example 3), in which each speaker spoke the language of the other, was actually better interpreted as complementary schismogenesis, as each speaker modelledin vainthe type of language choice she hoped her interlocutor would make. The conflictual nature of this pattern of codeswitching was substantiated by the lack of congruence between these speakers' projections for the likely code for future conversations. These speakers, whose conversation contains complementary schismogenesis, failed to progress towards development of a dyadic or micro-community speech norm. This seems like good evidence that convergence is not necessarily a successful act of speech accommodation. In contrast, those speakers who demonstrated compliance, that is, a willingness to speak whatever the interlocutor wanted to speak, did succeed in being accommodating. Furthermore, the speakers whose conversations showed trends over time towards compliance were also those whose projec-

< previous page

page_183

next page >

< previous page

page_184

next page > Page 184

tions for code choice for future conversations were more likely to be congruent. These speakers were beginning to establish micro-community speech norms for future conversations. It is the establishment of these norms that links the investigation here to codeswitching issues in other contexts. While bilinguals in a bilingual community learn the norms for code choice applicable to their community, incipient bilinguals in conversation with each other have no such community norms to fall back onor even to acquire. What I wish to claim here is that successful language learners will manage to create such norms in the process of establishing relationships with speakers of the language they are learning. For speakers and learners of 'world' languages such as English, German, Russian, Japanese and so on, the pragmatic problem is a significant one. Finding a solution acceptable to both interlocutors is a significant step on the road to becoming a successful speaker of the second language. The hypotheses presented here are based on data from a small number of short conversations between a small number of speakers. These hypotheses would be strengthened by confirmation by other bilingual learners of the interpretations taken here, of the compliance pattern as accommodating and the complementary schismogenesis pattern as non-accommodating. It is to be hoped that the tools of Speech Accommodation Theory and the insights of pragmatics can be combined in research that tests the claims of this paper. Note 1. Given that Clara chooses one language and sticks to it, it might be asked whether the conversation here contains actual codeswitching. It should be noted that even learners, who by definition are unbalanced bilinguals, have more than one code in their repertoires. A single speaker may choose to employ only one language of her repertoire, as Clara does in (3), but if her interlocutor tries out a different code, the resulting conversation as a whole may be seen as characterised by codeswitching. Certainly, if either learner changes codes, as happens in all four conversations discussed here, the phenomenon is clearly codeswitching. Thus, even if a given speaker never switches codes, she may find herself in a codeswitched conversation. The phenomena described here do seem to fit the definition of codeswitching of Heller (1988: 1) 'the use of more than one language in the course of a single communicative episode.' References Bateson, G. (1958) Naven. Stanford: Stanford University Press. Blom, J.-P., and Gumperz, J. J. (1972) Social meaning in linguistic structure: Code-switching in Norway. In J. Gumperz and D. Hymes (eds) Directions in Sociolinguistics (pp. 407-34). New York: Holt. Bourhis, R. Y. (1984) Cross-cultural communication in Montreal: Two field studies since Bill 101. International Journal of the Sociology of Language 46, 33-47. .(1985) The sequential nature of language choice in cross-cultural communication. In R. L. Street and J. N. Cappella (eds) Sequence and Pattern in Communicative Behavior (pp. 120-41). London: Edward Arnold. Bourhis, R. Y., Giles, H., and Lambert, W. E. (1975) Social consequences of accommodating

< previous page

page_184

next page >

< previous page

page_185

next page > Page 185

one's style of speech: a cross-national investigation. International Journal of the Sociology of Language 6, 53-71. Brown, P., and Levinson, S. C. (1987) Politeness. Cambridge: Cambridge University Press. Burr, S. M. (1990) External and internal conflict: Conversational codeswitching and the theory of politeness. Sociolinguistics 19, 21-35. Genesee, F. and Bourhis, R. Y. (1982) The social psychological significance of codeswitching in cross-cultural communication. Journal of Language and Social Psychology 1, 1-27. (1988) Evaluative reactions to language choice strategies: The role of socio-structural factors. Language and Communication 8, 3-4, 22-50. Giles, H. and Powlesland, P. F. (1975) Speech Style and Social Evaluation. London: Academic Press. Giles, H. and Smith, P. (1979) Accommodation theory: Optimal levels of convergence. In H. Giles and R. N. St. Clair (eds) Language and Social Psychology (pp. 45-65). Oxford: Basil Blackwell. Giles, H., Taylor, D. M. and Bourhis, R. Y. (1973) Towards a theory of interpersonal accommodation through language: Some Canadian data. Language in Society 2, 177-92. Green, G. (1990) The role of pragmatics in language understanding. In L. F. Bouton and Y. Kachru (eds) Pragmatics and Language Learning 1, 21-41. Gumperz, J. J. (1976) The sociolinguistic significance of conversational codeswitching. In J. Cook-Gumperz and J. J. Gumperz (eds) Papers on Language and Context (University of California Language Behavior Research Laboratory Working Paper 46). (1982) Discourse Strategies. Cambridge: Cambridge University Press. Heller, M. (1982) Negotiations of language choice in Montreal. In J. J. Gumperz (ed.) Language and Social Identity (pp. 10818). Cambridge: Cambridge University Press. (1988) Introduction. In Heller (ed.) Codeswitching: Anthropological and Sociolinguistic Perspectives (pp. 1-24). Berlin: Mouton de Gruyter. Platt, J. and Weber, H. (1984) Speech convergence miscarried: An investigation into inappropriate accommodation strategies. International Journal of the Sociology of Language 46, 131-46. Scotton, C. M. (1983) The negotiation of identities in conversation: A theory of markedness and code-choice. International Journal of the Sociology of Language 44, 115-36. (1988) Code-switching as indexical of social negotiations. In M. Heller (ed.) Codeswitching: Anthropological and Sociolinguistic Perspectives (pp. 151-86). Berlin: Mouton de Gruyter.

< previous page

page_185

next page >

< previous page

page_186

next page > Page 186

< previous page

page_186

next page >

< previous page

page_187

next page > Page 187

Chinese Preschool Codeswitching: Mandarin Babytalk and the Voice of Authority 1 Catherine S. Farris Department of Sociology & Anthropology, University of Northern Iowa, Cedar Falls, IA 50614-0513, USA. Abstract Using Carol Scotton's (1983, 1988) model of codeswitching based on markedness theory I examine a type of register variation which is referred to as 'babytalk', ambiguously talk of babies (or young children) and talk to babies or young children. Examples are drawn from videotapes of naturally occurring conversation among Mandarin Chinese speaking children (ages 2 1/2-6 years) and teachers in several preschools in Taiwan. Considering babytalk as one type of codeswitching adds a new perspective to studies of child language acquisition and socialisation, while highlighting the need for a developmental perspective and more sensitivity to local constructions of person and society in the codeswitching literature. The claim that babytalk is a universal speech register is examined in light of cross-cultural evidence, including my Chinese data. A style related to babytalk, a 'language of socialisation' (Gleason, 1973) or what I refer to as 'the voice of authority', is used by teachers to preschool students, while babytalk is retained on a diminishing scale as children become older, as though to ease the transition to a clearly didactic style. In talking to children, Chinese teachers codeswitch between this unmarked voice of authority and a babytalk register. I argue that these two speech styles, babytalk and the voice of authority, are motivated by Chinese cultural assumptions about children and childhood. A metaphorical or secondary use of babytalk, referred to in Chinese as sajiao, is intimately intertwined with notions of gender and the expression of masculinity and femininity in Taiwan's society. Introduction Scholars of language and speech increasingly recognise that knowledge of and use of varieties or codes (i.e. different languages, dialects, or registers [Hudson, 1980]) are part of the sociolinguistic resources of a speech community, and that their description must be included in a grammar of a

< previous page

page_187

next page >

< previous page

page_188

next page > Page 188

native speaker's knowledge. Moreover, control over varieties is not spread uniformly throughout a speech community, but may cross-cut regional, class, ethnic, gender, and generational lines. Carol Scotton (1983, 1988) has developed a model of code switching based on markedness theory and emphasising the interactive and negotiated nature of face to face interaction. Monica Heller (1988:2) calls for an examination of 'ways in which code switching is used to signal social, discourse and referential meaning' and a 'model which unifies macro-level and micro-level approaches to the study of language change and social change'. The codeswitching phenomenon that I will examine here is a type of register variation (i.e. variation according to use [Hudson, 1980]) which is referred to as 'babytalk', ambiguously talk of babies (or young children ) and talk to babies or young children. My examples are drawn from videotapes of naturally occurring conversation among Mandarin Chinese speaking children (ages 2 1/2-6 years) and teachers in several preschools in Taiwan. Considering babytalk as one type of codeswitching adds a new perspective to studies of child language acquisition and socialisation, while highlighting the need for a developmental perspective and more sensitivity to local constructions of person and society in the codeswitching literature. Many argue that some type of babytalk is a universal speech register which adults and older children direct at babies and young children. This claim will be examined in light of cross-cultural evidence, including my Chinese data. A style related to babytalk, a 'language of socialisation' (Gleason, 1973) or what I refer to as 'the voice of authority', is used by teachers to preschool students, while babytalk is retained on a diminishing scale as children become older, as though to ease the transition to a clearly didactic style. In talking to children, Chinese teachers codeswitch between this unmarked voice of authority and a babytalk register. In this research, the first to consider babytalk in a codeswitching framework, I will show that these two speech styles, babytalk and the voice of authority, are motivated by Chinese cultural assumptions about children and childhood. A metaphorical or secondary use of babytalk, referred to in Chinese as sajiao, is intimately intertwined with notions of gender and the expression of masculinity and femininity in Taiwan's society. Babytalk as a Language Universal According to Charles Ferguson (1978), the study of babytalk registers should be seen as part of the general question of language varieties, and is part of a speaker's 'communicative competence' (Hymes, 1972). In examining published reports of babytalk in 27 languages (ten of them Indo-European languages) Ferguson argues that in every society, people modify their speech in talking to young children and that these modifications occur in at least one of the following areas: prosody (e.g. higher pitch), grammar (e.g. shorter sentences), lexicon (e.g. special vocabulary'doggie', 'pee-pee'), phonology (e.g. simplification of consonant clusters), and discourse

< previous page

page_188

next page >

< previous page

page_189

next page > Page 189

(e.g. more questions). Ferguson notes that the most notable characteristic of babytalk is prosody. That is, there is a 'tone of voice' thought appropriate for speaking to young children; in addition to raised pitch, this may include exaggerated intonation contours, lento or slow speech, careful enunciation, fewer disfluencies and longer pauses than in adult-directed speech. There is also a range of intraregister variability, or, what Ferguson refers to as 'degree of babyishness' (1978: 212). For instance, a 'heavier' babytalk register might be used with younger as opposed to older children. In some societies babytalk has secondary uses, e.g. in talk to pets, between lovers, to the aged or infirm, in sarcasm or in irony (Ferguson, 1977:229). In these secondary usages, certain features of babytalk, especially prosody and special lexicon appear. Such secondary uses can be considered as metaphorical code switching (Blom & Gumperz, 1972), i.e. something about the parent-child relationship is evoked through the use of babytalk. Although conventional-ised and variable among speech communities, the modifications constitute a coherent set of features that should be regarded as a register. Ferguson asserts that babytalk registers have 'an innate basis in pan-human child-care behaviors' and 'in part arise directly from interactional needs and limitations of children's behavior'. Babytalk is thus an important factor in child socialisation, 'apparently assisting in the acquisition of linguistic structure, the development of interactional patterns, the transmission of cultural values, and the expression of the user's affective relationship with the addressee' (1978:214-15). With the advent of Chomskian linguistics, and its strong argument for an innate, species-specific capacity to acquire a languageany languagegiven that a child is exposed to a minimal amount of speech, the role of caregivers in the child's acquisition of language seemingly was reduced to a minimum. However, working largely in white, middle-class, American English speech communities with mothers and children who lived in nuclear family households, and taking this model as the unexamined norm, researchers of child language acquisition began to report that mothers (and other caregivers) regularly modified their speech when speaking to very young children. Important to these scholars were reported modifications in grammar, specifically, shorter sentences, fewer subordinate clauses, fewer grammatical relations, and more repetitions than in adult-to-adult speech (Ferguson, 1978). 2 An orgy of Chomsky-bashing began. Mothers, it was argued, seemed to be providing 'language lessons' for these novices, without which lessons it was further assumed, children could not learn a language. After several decades of research however, this argument began to crumble. Newport et al. (1977) and Gleitman et al. (1984) among others, pointed out the great difficulty in the cause and effect argument. Do speech modifications of caregivers affect the child's acquisition of certain linguistic structures, or, does the child's immature speech invoke speech modifications in the speech of the caregivers? Newport et al. found that in mothers' speech to their children, they seemed to be concerned with processing simplicity and

< previous page

page_189

next page >

< previous page

page_190

next page > Page 190

semantic cohesion rather than syntactic simplicity. They did find that the growth of certain language-specific aspects of the child's speech was correlated with aspects of the mothers' usage. Whereas, measures of child language growth which are indices of universal aspects of language structure appeared to be insensitive to maternal variation in speech styles (1977:133). Certain aspects of some mothers' speech, such as greater use of verbal and nonverbal deictics (as in pointing at and naming an object), interacted with children's learning strategies and promoted vocabulary growth. Ben Blount (1981:396) notes that caregiver speech embedded in concrete, referentially explicit situations, rather than the linguistic structure of the speech itself, appears to foster language acquisition. It is not clear from the literature whether this special maternal speech or 'motherese' (sic) is a subset of babytalk or the basis for a separate speech style. Ferguson includes such grammatical modifications in his specification of babytalk. If we consider this 'motherese' not as a language teaching technique but as a socialisation tool, its functional rather than formal independence from babytalk will become evident. I argue that 'motherese' is the precursor or ucleus through which a language of socialisation or voice of authority emerges. What is the language of socialisation? For an American-English setting, Gleason (1973) describes different speech styles of adults to children of varying ages. Speech to babies and toddlers was characterised by a clear babytalk style, including raised pitch, simple, short sentences, concrete nouns, diminutives, hypocorisms (i.e. terms of endearment) and expansions of the child's utterances. As the child's speech became more comprehensible, raised pitch and expansions and grammatical simplifications dropped out, while expressive features such as hypocorisms and word play remained. In other words as Ferguson noted, the degree of babyishness in the register diminishes with increasing age of the child. Gleason found that adults addressed children aged 4-8 in a somewhat different style, which she calls a language of socialisation. It is 'a very controlling language' and one 'filled with social rules', telling the child 'what to do, what to think and how to feel', in the process, teaching the child a specific worldview. This style can be described functionally and formally in terms of speech acts (many directive types, including imperatives, questions, suggestions, warnings, etc.), questions which contain their own answers, thus teaching the child how to make conversation, and exaggerated responses (e.g. 'Heh, that's really something, isn't it?', which includes special lexical items [interjections], exaggerated contours, and tag questions) which Gleason argues, teaches children how they ought to feel. However, the latter example, exaggerated responses, clearly contains features identified with babytalk. Thus, the language of socialisation is not as well defined formally as is babytalk. Instead, it appears to emerge from babytalk, but to include prominently directive speech acts and explicit articulation of social rules and mores (e.g. 'Eat with your mouth closed.'). In the cross-cultural literature on language socialisation, caregivers in a number of communities from

< previous page

page_190

next page >

< previous page

page_191

next page > Page 191

Polynesia to Africa and across classes and ethnic groups in the United States have been observed to provide 'explicit instruction in what to say and how to speak in a range of recurrent activities and events. . .' (Ochs, 1986: 5). For instance, Watson-Gegeo & Gegeo (1986) describe calling-out and repeating routines which caregivers use with Kwara'ae (Solomon Islands) children to teach them politeness routines as well as for explicit corrections of linguistic forms. Demuth (1986) reports on caregivers' use of prompting routines ('say x') to Basotho (South African) children as a didactic device for instructions in how to speak and act appropriately. Pye (1985) also describes, for a Quiche Mayan community, a special verbal routine (again, using the verb 'to say') which functions to direct the child's activities. I will define this language of socialisation or the voice of authority in functional or speech act terms, as the speech of caregivers to children which has the intention to direct or control the child's behaviour and attitude and to socialise the child into the social roles and cultural meanings of the local community. The voice of authority should be described in terms of its discourse functions, rather than formally in terms of a feature- or item-based register such as babytalk. In a developmental perspective, caregivers use a heavier babytalk register in conveying their voice of authority to younger children and a lighter babytalk register to older children, until, by school age use of babytalk has dropped out (except in intimate situations) while the voice of authority remains. What are some implications of babytalk and the voice of authority as types of codeswitching phenomena? Carol Scotton (1983) argues that codeswitching should be seen as a process of the negotiation of social identities in conversation, and that code choices are interpreted in terms of relative markedness. 'The social purpose of conversation is negotiation of a set of rights and obligations (RO set) between speaker and addressee. . .' (1983:117). In conventionalised talk exchanges, social norms specify an unmarked RO set for certain pairs of social identities. The unmarked code choice for caregivers in speech to young children is the voice of authority. The relationship set which is evoked by this choice is a state of immaturity of the child and an obligation by the adult to nurture and to guide. The caregiver has the right to command and the child is expected to be obedient and compliant. When the caregiver switches to a babytalk register, this situational switching is, in Scotton's framework, a movement from one unmarked choice to another (1988:160). A new RO set is implicated. In Chinese preschools, teachers of the younger children in particular often mix the voice of authority with a babytalk register, thereby 'implicating multiple identities' (Scotton, 1983:136). That is, they wish to imply that the RO set established and maintained through the use of the voice of authority is still in effect, but that a new RO set has temporarily come into play as well (i.e. an effective relation between the teacher and student). As the child grows older the degree of babyishness decreases; that is, speech with fewer babytalk features appears as part of the unmarked choice.

< previous page

page_191

next page >

< previous page

page_192

next page > Page 192

Recently, many scholars have argued for attention to patterns of child socialisation in societies other than the US, and for a shift in emphasis from a narrow concern with language acquisition to broader questions about the social and cognitive environment of children (e.g. Schieffelin & Ochs 1986). In a cross-cultural perspective, the form and function of babytalk as a register found in all speech communities has been questioned. For example, studies by Harkness (1976) in a Spanish-speaking Guatemalan community and by Pye (1985) in a Quiche Mayan community indicate that recognisable babytalk registers none the less differ formally from the American-English pattern and that 'the whole phenomenon of speech to children depends crucially upon cultural concepts about children and conventions for interacting with them' (Pye, 1985:98). The very notion that in all societies people have a special register for talking to children has been disputed by evidence from Western Samoa. Elinor Ochs (1982:101) reports that she was 'stunned by the relative absence of many features that have been described as caregiver register' in her Samoan sample. She found no use of babytalk lexicon, special morphology, simpler or reduced syntax, expansions, and only 'a minimum of cooperative utterance-building between caregiver and charge'. In particular, Ochs details cultural assumptions underlying the use of expansions in American mothers' speech to children, and its absence in Samoan caregivers' speech. American caregivers adopt the child's perspective (they 'decenter'), and adjust to the child's egocentrism, in expanding the child's utterance (e.g. 'ball'; 'yes, that's a ball'). However, in Samoan society, it is high status people who have the right to act egocentric and low status persons who must strive to understand and to serve them (1982:95). Thus, an older child or adult would not try to make sense out of a child's partially unintelligible utterance, as an American caregiver probably would, but instead dismisses it as non-speech. In other words, Samoan and American caregivers' interactive styles with children are predicated upon very different cultural assumptions about personhood and interactive requirements. Samoan assumptions appear to result in the almost complete absence of anything approaching a true baby-talk register. Ochs does report that 'infants are sung to and cooed over to distract them from their hunger or to put them to sleep or simply to amuse them' (1982:89). By the time the child becomes a toddler and one word utterances begin, 'the process of conscious social instruction begins in earnest' (1982:95). Children are encouraged, via a verbal routine, to notice and to name family members, and to ask questions about other people: where are they?, what are they doing?, how are they feeling?, etc. 'The caregiver prompts the child in the production of socially appropriate conversational contributions' (1982:98). In other words, caregivers use a 'language of socialisation', or voice of authority, in interacting with children. Ochs (1988) describes Samoan children's acquisition of affect-laden lexical forms which characteristically have prosodic marking, prior to acquisition of neutral ones.

< previous page

page_192

next page >

< previous page

page_193

next page > Page 193

Blount (1990:556) points out that this indicates Samoan children do receive 'special prosody' (as Ferguson suggested, a babytalk 'tone of voice'), 'but in a different form from the exaggerated pitch amplitude seen in simplified registers. . .' in other languages. The accompanying package of babytalk features described for many other languages does appear to be absent in the Samoan example. Ochs' work highlights the need for attention to local sociocultural systems in describing linguistic socialisation. Rather than acting as language instructions per se, modifications in caregiver speech to children appear to function to bring children into the social world of the people around them and to socialise them to cultural assumptions and culturally appropriate ways of interacting. It may be that a voice of authority, and not babytalk, is a universal strategy of caregiver-child interaction. Babytalk and the Voice of Authority in Chinese Introduction Elinor Ochs was initially stunned by the absence of a babytalk register in Western Samoa; I was similarly surprised to discover the overwhelming prevalence of varying degrees of babytalk in the speech of Chinese preschool teachers to children as young as 2 1/2 and as old as 6 years. Many of the features which Ferguson describes as a universal set which all speech communities draw on to construct a babytalk register appear to be present in these caregivers' speech. As my sample is from speech addressed to children who are generally older than those in many of the published reports of babytalk, these data actually underrepresent the extent of a babytalk register in this Chinese speech community. In a study of the development of Mandarin syntax in children ages 1 1/2 to 3 1/2 years, Mary Erbaugh (1982) reports that certain features Ferguson notes, but which I did not find in the older sample, are present in interaction between very young children and caregivers in the intimate setting of the home. The research reported here is part of a larger study of language socialisation for gender in Taiwan, and my discovery of a childish, pleading speech style, encoded in Mandarin as sajiao, and associated with the speech of very young children of both sexes and with the speech of young (usually unmarried) women, led to my critical examination of the forms and functions of babytalk in a preschool setting (Farris, 1988, forthcoming). The data discussed below was collected in 1984 and 1988 at two private, three-year preschools in Taibei, the Republic of China on Taiwan. In the former study, conducted at a small (30 children and two teachers), Catholic Church-run preschool in a lowermiddle class neighbourhood, an eight month period of participant observation was followed by six hours of VHS videotaping over a three day period. 3 In the latter study, conducted in a large (100 children, eight teachers and four student-teachers), upper class neighbourhood, six weeks of observation was followed by 4 hours of VHS

< previous page

page_193

next page >

< previous page

page_194

next page > Page 194

videotaping over a two day period. Data collection also includes journal notes, and interactions with teachers, students, parents, and older siblings. My analysis of these data is also informed by a longitudinal preschool study carried out in 1989-90. Although preschool teachers in Taiwan are predominantly women, at the larger preschool a male art teacher taught the oldest class once a week, and I had the opportunity to videotape one of his lessons. Chinese preschools are typically divided into three classes: xiaoban ('little class', i.e. 3-4 year olds), zhongban ('middle class', 45 year olds), and daban ('big class', 5-6 year olds), the oldest class comparble to an American kindergarten level. As both these periods of videotaping occurred at the end of summer, when a new school year has just begun, the youngest children were barely three years old (and a few not yet three), while the oldest children were between 5 and 5 1/2. In the smaller preschool, the daban students constituted about half of all students; the younger two classes were combined under one teacher that year. In the larger preschool (and the common pattern), students were divided into the three grade levels. The older two classes (30-35 students each) were divided into two 'small groups', each with its own teacher. The entire age grade met together for some activities and the daban and zhongban also did song and dance routines and small group activities together (at a variety of work stations scattered around the large, main room). The xiaoban class was segregated from the other two, and was attended by four teachers and two student teachers. Native speakers transcribed the verbatim text of the videotapes for me; I then transcribed prosodic and paralinguistic cues as well as aspects of kinesic and proxemic behaviours. 4 I videotaped students playing inside the classroom under the supervision of teachers and outside in the playground, usually in self-selected play. By the time I began videotaping, the children were quite familiar with me and called me Fu Ayi, 'Aunt Fu' (my Chinese surname), or, simple 'Ayi'. For many people in Taiwan's growing middle class, as well as for those who aspire to it, the three-year preschool or youzhiyuan is seen as the ideal place for young children to gain the social skills for later success in modern life. Thus, many working and non-working mothers send their children to preschool. Taiwan's preschool population has increased more than ten-fold since 1950 (ROC Reference Book 1987) and is the fastest growing segment of the student population. Preschools undoubtedly play an increasingly important role in the development of a national culture in Taiwan, which is based on the Mandarin language (rather than Taiwanese, a variety of southern Chinese spoken natively by 85% of the population), stands in symbolic opposition to communist Chinese culture and society, and consciously strives to preserve elements of 'traditional' Chinese culture while selectively incorporating elements from the West.5 Chinese is a language family, part of the Sino-Tibetan branch of languages. Standard Chinese, known in the West as Mandarin, is the most widely

< previous page

page_194

next page >

< previous page

page_195

next page > Page 195

spoken of the seven major 'dialects' or languages of China. Some variety of Mandarin is spoken by 70% of the mainland Chinese population, and it is the official language of both the People's Republic of China and of the ROC on Taiwan (Charles Li & Sandra Thompson, 1981). The native language of the majority population on Taiwan is Minnan ('southern Min'), a variety of Min, which is one of the seven major Chinese languages, all of which are mutually unintelligible. However, Mandarin Chinese has been the official language of government, military and education on Taiwan since the Nationalist government took over administration of Taiwan in 1947. Today 85% of the population (half of whom were born after World War II) is effectively bilingual (see Tse, 1982 for further details on the sociolinguistic situation). Taiwanese speaking families in Taibei, the seat of government and stronghold of the Nationalists, typically expose their preschool age children to Mandarin language television, radio, videotapes, etc., and often use Mandarin when speaking to them, in order to acclimatise them to the classroom language and ensure success in school. The children in this study, from a middle class or upper class background, were all either monolingual Mandarin speakers or bilingual Mandarin and Taiwanese speakers. Most of the preschool children here spoke standard Taiwan Mandarin, although a few had heavy accents. Mandarin is the language of instruction and also is used on the playground. I heard only a few of the youngest children speak any Taiwanese and teachers occasionally spoke to visiting parents in Taiwanese. A Chinese Babytalk Register In a study of children's acquisition of Mandarin syntax, Erbaugh (1982) reports that Chinese people generally do not consider that children can talk until they are over one year old, and can say a number of words clearly. Adults 'do talk to babies in a way quite similar to American middle class whites' (1982: 77). As Ferguson (1978) noted, prosody is important in Chinese babytalk. When talking to children, Chinese adults exaggerate tones in pitch contour; tones are pronounced on all syllables rather than neutralised in unstressed ones (Erbaugh, 1982:117). Erbaugh also reports noun reduplication, producing a special babytalk lexicon (reduplication is one of the few morphological devices available in Chinese), such as gou-gou ('dog-dog''doggie'). 6 Sound imitation or onomatopoeia, particularly animal sounds, are also evident and can be considered part of the reduplication style producing a babytalk lexicon. In what is functionally part of the voice of authority, syntax and discourse features of adult Chinese speech to children include a 'quiz style of conversation'; central to this style are questions (e.g. 'Zhei shi shemme?', 'What's this?'). Scaffolding is provided through frequent question frames. The 'unanimous Chinese opinion is that children must be taught to speak properly and intelligently'. Thus, two year olds are drilled on names of people, animals, shapes, colours, numbers, address,

< previous page

page_195

next page >

< previous page

page_196

next page > Page 196

phone number, children's poems, songs, and standard stories (1982: 121). Culturally important domains, such as kin terms, are repeatedly modelled for young children by adults and older children. Although little time is spent on modelling politeness routines, in contrast to the American middle class pattern, children are coached in the appropriate kin term, or professional term, surname, and given name. 'In Chinese society it is absolutely necessary to know who someone is by name and rank in the family and society' (Erbaugh, 1982:385). Erbaugh concludes that there is much uniformity in adult speech patterns to children and that it would be difficult to find control groups of children who learned Chinese 'without also receiving the sociolinguistic training which is central to Chinese culture' (1982:708). Preschools present a different setting from the home, and the social identities are shifted also. Teachers have an explicitly pedagogic role and the child is the novice, the student. One could therefore expect from teachers a more 'authoritarian' voice and lighter babytalk register than from parents and other relatives in the home. In my preschool data, babytalk from the teachers clearly decreases as students become older; the voice of authority, although obvious even in the 'little class', is softened with many babytalk features, while for older children, usually it is not. Adult babytalk in the classroom draws on most of the features in Ferguson (1978), as listed in Table 1. Teachers codeswitch between their authoritarian speech style, the voice of authority, which is the unmarked code choice for this social identity and Table 1 Babytalk features in Chinese preschools Prosodic slow or lento speech, careful enunciation, frequent tempo shifts, special voice quality ('soft', or 'warm') Grammatical shorter sentences, self repetitions, self paraphrasing, many affective sentence final particles* Lexical words in restricted semantic domains (body parts and functions, kin terms, food, animals, infant games), hypocoristic forms, phonological simplification (reduplication*), onomatopoeia, interjections Discourse many questions, tag questions, pronoun avoidance* (name + title for first person) Secondary uses Adult intimacy: sajiao* pleading style (subordinate or intimate); to pets * language-specific feature (adapted from Ferguson 1978)

< previous page

page_196

next page >

< previous page

page_197

next page > Page 197

setting, and the babytalk register, which they apparently use to ease the child's transition from the intimate and babyish milieu of the home to the regimented, public institution of the school. The students, especially the younger ones (> four years) often interact with the teachers via the pleading, manipulative sajiao speech style, which demonstrates their growing control over rulegoverned variation in speech (see also Anderson, 1986; Gleason, 1971; Sachs & Devin 1975; Weeks 1971 on children's control of different speech styles). A childish pleading style of communication may be a developmental universal. In Chinese, this style, sajiao, is culturally salient, being labelled with a monolexemic term, and is easily available for metacomment. That is, children not only learn to 'do sajiao', but also learn that this pleading behaviour has a specific label. Adults label instances of sajiao behaviour of children and other adults and are able to discuss features which comprise the style (Farris, forthcoming). The sajiao communication style, which comprises both verbal and nonverbal features, draws on a subset of the typical Chinese babytalk register: lento (slow) and piano (soft) speech, often a nasal quality, frequent use of particles (particularly ma a 'softener' and la, an impatient, childish usage), a pronoun shift using renjia ('someone, others', used in this style as a deferential, first person reference), and kinesically, pouting or smiling, and shoulder twists. Native speakers associated this style with the speech of young children of both sexes and with young women; it is considered a 'soft' and indirect strategy of manipulation which is appropriate for these social identities. Teachers themselves occasionally use this style, rather than their usual voice of authority, especially to younger children to direct them via 'soft' manipulation. The Preschool Studies The head teacher at the small preschool was an unmarried, Catholic Chinese woman. She spoke very warmly to all the students, but especially to the younger ones. That is, her authority voice is usually overlaid with or realised through a babytalk register. The students reciprocated her friendliness and some, younger children of both sexes frequently evoked their dependent status by using the sajiao pleading style in interacting with her. In Example 1 below, 'Aunt' has brought her videocamera to class and the head teacher is telling the entire class about it. Several 'big class' (daban) boys hurry over to ask me about the camera, but teacher, in a voice of authority without any babytalk features, tells them to sit down. She then switches back into her usual babytalk register, suitable for a mixed age audience. In the dialogues below, a sentence in Chinese is followed by my colloquial English translations. 7 (For transcription and grammatical conventions, see Appendix I.) In the examples below, babytalk features are in italics.8

< previous page

page_197

next page >

< previous page

page_198

next page > Page 198

Example 1 Participants: 'Li Laoshi' Teacher Li [TC]; Class [C1] (responding en masse); Boy 1 [B1], 4 years old (in zhongban) (1) TC: Ai::ya! Huanying xin pengyou huanying, Ayi gen women ((rallentando)) yiqi wan. ((lento, nasal)) Dui bu dui? Shi bu shi, a?! (A-ya! Welcome new friends and welcome Aunt, come to play together with us. Isn't that correct? Isn't that so, a?!) (2) C1: Dui! (Correct!) (3) TC: ((lento)) Ayi jintian dailai yige shemme lao gaosu wo:: ((allegro)) shei zhidao?// (Today, Aunt has brought what with her? Tell me, who knows?) (4) B1: [Laoshi, laoshi!] (Teacher, teacher!) (5) C1: [Sheyingji.] (A movie camera.) [B1 jumps up from his seat, pointing across the room:] (6) B1: Laoshi, lashi// [ta na tade. . .] (Teacher, teacher, he* took his. . .) (7) TC: [Dui:: la.] Sheyingji. (That's right, a. A movie camera.) [B1 fails to get teacher's attention so runs over to where she is sitting.] (8) B1: Laoshi, laoshi. . . (Teacher, teacher. . .) (9) TC: E? (Eh?) [Several daban boys run over behind teacher to where I am standing, and begin asking me about the camera.] (10) B1: Laoshi, laoshi. You yige ren ba renjia dongxi nalai wan. (Teacher, teacher. There is someone who is taking others' things to play with.) (11) TC: Hao! Ba ta fangqilai hao ma?! Bu yao wan renjia de dongxi. (Ok! Put it away, ok?! Don't play with others' things.) [Meanwhile, the boys are talking to me, but teacher's louder voice is recorded.] (12) TC: Ayi shuo, 'xiao pengyou hao keai'. Ye yao paipai xiao pengyou huodong de qingxing. (Aunt says 'little friends are so adorable'. And (she) wants to just film a bit, (her) little friends' activities.) [Another boy runs over behind teacher and joins the crowd of boys looking at and questioning me about the camera. Teacher now turns around and directs her attention to these boys.] (13) TC: Hao, bu keyi mo. Mole ta jiu ((claps hands once)) meiyou paidao

< previous page

page_198

next page >

< previous page

page_199

next page > Page 199

ni le. (Ok (you) can't touch. If (it) is touched, then (it) won't film you.) [Here she holds a 'camera' up to her face and 'snaps' a picture.] (14) TC: ((lento, nasal)) Dui bu dui? Hao guai; qing zuoxia:: a. (Isn't that correct? Really obedient; please sit down, a.) [She gently shoves them back toward their seats.] * The third person pronoun in Chinese ta, is not marked for gender. Since context usually disambiguates it, I am translating it with gender here. This teacher is addressing en masse, students from 2 1/2 years old to 5 1/2 years. She typically speaks gently, using many babytalk features, here including tempo shifts and marked voice quality (nasal), many interjections and affective particles, special vocabulary ('little friends'), and tag questions. The example also shows how this teacher moves from an authoritative voice with many babytalk features (lines, 7, 11) to an authoritative one with no babytalk features, i.e. 'the gloves are off' (lines 12, 13) and back again, in succession (lines 11-12, 13-14), a virtuosity of 'situational' codeswitching (Blom & Gumperz, 1972). The teacher's authority voice is at normal adult tempo and tone, contrasting with the slow speech and warm voice quality of her babytalk register; it is also full of the language of socialisation, explicitly citing rules (line 13: 'You can't touch'.) and using the rhetorical tag question, 'hao ma?' (line 11: Ok, then?) which assumes agreement. 9 In line 12, the teacher puts words in my mouth ('Aunt says, ''little friends are so adorable"'), in effect telling the students (and me!) how to feel about themselves ('adorable') and how adults should feel about them. Finally, notice the verbal and kinesic pleading behaviour ('laoshi, laoshi') of the four year old boy (lines, 6 11) who succeeds in directing teacher's attention to a rule infraction (line 10) and its explicit pronouncement, (line 11: 'Don't play with it'.). In the larger preschool, the youngest group, the xiaoban students, are segregated from the older two classes and occupy the second floor rooms. Here about 35 students spend most of their time in small, self-selected groups, working with clay, lego toys, punch boards, etc., or in 'hands-on' demonstrations (e.g. making fruit juice). With this age group, the teachers speak very gently and the voice of authority is always tempered with some babytalk features. In Example 2, eight xiaoban students are about to play with clay. First teacher kneads the clay, adding water to make it soft, then invites some of the students to help her knead. Example 2 Participants: Teacher [TC], girl 1, girl 2, both three years old (15) TC: Lai bang wo a! Bang wo cuo yixia. Kan shei lai bang wo cuo. (come help me a! Help me knead a while. See who can come help me knead.) (16) G1: ((lento, nasal)) Ni zuo shemme a? (What are you doing a?)

< previous page

page_199

next page >

< previous page

page_200

next page > Page 200

[She comes up and blows on the clay, possibly confused by the word cuo ('to knead') and chui, ('to blow').] (17) TC: Bu shi chui la. Shi cuo de la. Shei yao bang wo cuo? (It's not to blow on la. It's to be kneaded la. Who wants to help me knead?) [To a girl sitting next to her:] (18) TC: Lai shi yixia. Ni lai shishi kan. Lai shishi kan! (Come try a while. You come try it and see. Come try it and see!) (19) G2: ((creaky, nasal)) Shi shemme? (Try what?) (20) TC: Shishi kan! (Try and see!) [The girl begins to knead the clay] (21) TC: Wa: ((lento)) hao bang o! Dui, hao bang! (Wow! Really outstanding o! Correct, really outstanding!) [She calls another student over and who begins to knead] (22) TC: (Name) hao bang o! Hee:yo hee:yo! ((Name) that's really outstanding o! Heh-yo hey -yo!) [Teacher encourages her with 'effort' onomatopoeia, beginning to knead the clay herself.] Using her voice of authority, this teacher is instructing students in artistic and manual skills and simultaneously facilitating affection between herself and the students via babytalk and accompanying kinesics, such as a smiling face. Her babytalk register here includes a soft or warm voice quality, expressive sentential particles (a, la, o), interjections (wa!) and onomatopoeia, which is modelled extensively by all the teachers. The teacher's repetition and paraphrasing of her own words (e.g. lines 15, 18) is both babyish (adults don't need such repetition) and didactic (every sentence is functionally a directive, and is in an imperative form; e.g. line 15, 'Come help me a?) The exaggerated comments on the student's kneading ability (e.g. line 21, 'Wow! Really outstanding oh!') in babytalk suggests to the student, as Gleason noted, how children should feel about their own accomplishments, i.e. very pleased. The students in turn produce a very soft, babyish tone of voice (i.e. nasal, creaky, lento). Thus, babytalk is at least in part invoked from adults by the speech of babies and small children. The children's speech here can also be seen as part of the sajiao style, evoking dependency and help from the teacher. Most of these three year olds are away from home and in a preschool environment for the first time and their speech and mannerisms are very tender. A generalised baby tone of voice (creaky and often nasal) is apparent in all their speech; the voice is often soft (piano) and the speech may be accompanied by lip-rounding or pouting. 10 They also used many sentence final particles, which express speaker affect and solicit a response. Teachers in turn responded gently, in both verbal and nonverbal behaviour. The

< previous page

page_200

next page >

< previous page

page_201

next page > Page 201

youngest students also used special vocabulary, modelled by teachers, such as noun reduplication and onomatopoeia. In Example 3, a xiaoban girl defies a teacher, using the sajiao style. Example 3 Participants: Teacher [TC]; three year old girl [G1] [G1 is directed by teacher to go upstairs when the teacher sees that she has brought a book from home, which is forbidden.] (23) TC:: E! Xiao Lan e! ((raised pitch)) Shang qu o! (E! Little Orchid e! Go on up o!) [She starts upstairs but the teacher puts out her hand to stop her.] (24) TC: Nei ge shu shi yong lai gan shemme de? (That book, what is (it) brought here for?) (25) G1: ((piano)) Naxia lai kan. ((It) is brought here to look at.) (26) TC: O! Naxialai kan de. Na ni xian shouzhe. ((lento)) Hao bu hao. (O! Brought here to look at. Now, you just put (it) away, OK?) [On 'you', the teacher pats her on the cheek.] (27) G1: ((very lento)) Bu:: yao ma. (Don't want to, ma.) [Her mouth remains open and she tilts her head back and to the side, pouting slightly.] (28) TC: Na ni, na ni yubei jiali kan jiu hao. ((warm voice)) Bu yao naxialai ma! (Now you, now you just prepare '(it) to read at home, then it's ok. Don't bring (it) here, ma!) The teacher is speaking in the voice of authority, directing the girl (lines 23, 26) to comply with rules and citing one (line 28), in the formula 'Bu yao Verb' ('Don't bring it here!'). In line 28 she softens her speech, using a voice quality shift (warm) and the softener particle ma in tone with the girl's extremely charming pleading style (line 27). This girl has intensified the babyishness of her speech in a bid to deny the authoritarian RO set which the teacher has invoked. The child's behaviourher strategic use of the sajiao pleading stylefalls under Scotton's 'deference maxim', which enjoins the speaker to 'show deference in your code choice to those from whom you desire something' (1983:123). The teacher is not swayed to bend the rules however, and has the last word. Notice also the affectionate gesture (after line 26) by teacher. The verbal and nonverbal interaction between teachers and three year old students, in which the voice of authority is realised through babytalk along with kinesic and proxemic displays of affection by the teachers, produces a warm and dependent childish ethos and a certain adult-child complex, in which further socialisation into Chinese culture and society will now take place. Erbaugh notes that children learning Chinese have, by the age of 3 1/2 years, achieved mastery over grammatical features of Mandarin, as well as

< previous page

page_201

next page >

< previous page

page_202

next page > Page 202

discourse sensitive reordering. Their speech is fluent and long, narratives are increasingly coherent, and there is increasing hearer and register sensitivity, while some grammatical and discourse mistakes continue into the school years (1982: 291). The zhongban or 'middle class' students in the preschool, ages four to five years, are also addressed by the teachers in a babytalk register, although somewhat modified or 'lighter' than the style directed at the xiaoban students. In particular, as Ferguson's model would predict, prosody is less central; the warm voice quality and tempo shifts in speech to the youngest class are less frequent. Discursively, the voice of authority is more apparent; the question and answer format which Erbaugh (1982) discusses is a frequent pedagogic tool, as well as much drilling in names of people and things, and group recitation of children's songs. Special vocabulary, noun reduplication and onomatopoeia, affective particles, pronoun shift ('teacher' for first person, e.g. 'Teacher says, don't do it!'), self repetition and paraphrasing, and tag questions are also part of this style. In Example 4 below, a zhongban teacher asks a group of students sitting in a circle with her, if they remember going to the wholesale store the day before, and what they saw there. They volunteer a few answers, but she prods them for more. This example, from a long interaction sequence between the teacher and students which included a discussion about the field trip and teacher-led songs with gestures, is a connected monologue. I have broken it into lines for easier reading. Example 4 Participants: Teacher Zhou [TC]; class as audience (29) TC: Xiao-pengyou kandao shemme dian? (Name) Zuo xialai! Zuo xialai! (Little-friends saw what kind of shop? (Name) sit down! Sit down!) (30) TC: Xiao-pengyou xiang-yi-xiang, hai you shemme dian? (Little-friends, think a bit, what other kind of shops?) (31) TC: Wo gaosu nimen hao le. Nimen dou jide de dou mei you Zhou laoshi duo. (I'll tell you then. None of you remember as much as teacher Zhou does.) (32) TC: Zhou laoshi yi jinqu de shihou, jiu kandao mai yifu mai yigui de shangdian. (Teacher Zhou, as soon as (she) went in, (she) saw a shop that sells clothes and clothes cabinets.) (33) TC: Haiyou kandao huazhuangpin de o! Haiyou mai huazhuangpin de! Dui bu dui? You mei you? (And (she) also saw one with cosmetics, o! Also one that sells cosmetics! Isn't that correct? Isn't that so?) (34) TC: Hai you zuo zuo zuo. Hai you kandao mai nansheng yifu de dian o! You mei you? (And then (she/we) continued walking a bit. And then (she/we) saw a shop which sells boy-students' clothing, o! Isn't that so?)

< previous page

page_202

next page >

< previous page

page_203

next page > Page 203

(35) TC: Hai you xie dian you mei you? (Also there was a shoe store, isn't that so?) In this example, the function of the teacher's discourse is to teach the children about the social and material world. Gone is a special tone of voice which xiaoban teachers often use, although other babytalk features remain, such as special lexicon ('little friends': lines 29, 30), tag questions, sentential particles (o), pronoun shift (lines 31, 32), and repetition and paraphrasing of her own speech. The didactic function of the teacher's speech is realised through the question frames and enumeration of types of stores. The children evidently respond favourably to this level of babytalk register, and continue using babytalk themselves, in particular noun reduplication. For example, earlier in this conversation, one boy volunteered the answer xiong-xiong dian ('bearbear store', 'bear-ie store'), which is strictly a childish or babytalk usage. Children at this age are also very interested in onomatopoeia. Teachers model it for them and they also appear to produce noval forms. Earlier in this sequence, the teacher led students in several songs with accompanying gestures. In the first song a wolf is shot; 'plow! plow!' the teacher sings, holding out both index fingers to 'shoot', as the students imitate her. In the second song, a cricket sound and gestures are modelled as the teacher, with students following her, holds two fingers up to each side of her head as feelers, saying 'xiou, xiou!'. Children of all ages at the preschool produce imitations of animals, machines and other natural and artificial sounds. Gender differences in the use of onomatopoeia come into play by late preschool years, when boys continue to produce all types of sounds, but especially machines, fighting, and explosions, while girls imitate animal sounds but rarely machines, etc. (Farris, 1991). In the daban group, teachers treat the students more maturely, especially by the end of the school year, when they are being prepared to enter primary school. In both of the preschools examined here, students are at the beginning of their school year, thus, members of the oldest class are 5-5 1/2 years old. The voice of authority is less intertwined with a babytalk register. In particular, prosodic features such as special voice quality and tempo shifts (but not lento speech), and grammatical redundancy features such as self-repetitions and paraphrasing are all but gone. Some babytalk features remain, such as hypocorisms, exaggerated use of particles, onomatopoeia, and pronoun shift ('teacher'for first person). Noun reduplication is no longer addressed to this age group, and by the end of the year students have stopped producing such forms. Aspects of the teachers' speech more clearly associated with the voice of authority didactic style are evident, such as frequent question frames, paraphrasing of students' speech, and tag questions. In Example 5 below, a male art teacher has come for his weekly lesson with the daban. Students of both sexes appear to be quite fond of him and clamour for his attention. He is demonstrating how to cut out figures and paste them on a piece of construction paper, soliciting suggestions

< previous page

page_203

next page >

< previous page

page_204

next page > Page 204

from the class as he does so. Although individual student voices are recorded, it was usually not possible to identify who spoke. This teacher's normal speaking voice to students is slightly lento (i.e. in comparison to his speech to adults) with careful enunciation throughout. Example 5 Participants: Teacher Liu [TC]; daban students individually [St] [Teacher has cut out a fish and pasted it on the paper, and is now drawing friends for the fish.] (36) St 1: Liu laoshi ni zai shui li youyong. (Teacher Liu, you're in the water swimming!) (37) TC: A! Hai you Liu laoshi zai shui limian youyong! (A! And still teacher Liu is in the water swimming.) [He continues drawing.] (38) TC: Na yu: hai you mei you biede pengyou a? (And does the fish have other friends a?) (39) St 2: You. Ta you caihong-yu, meiren-yu. (Yes. It has a rainbow fish (friend), (and) a mermaid (friend).) (40) TC: Caihong-yu, meiren-yu hui youguolai gen ta jianghua. (Rainbow fish and mermaid can swim over and speak with it.) (41) TC: E! Ni kan zai zheili you yitiao xiaoxiao de yu. Xiao yu ye hui you guolai. (E! Look, here is an itty-bitty fish. The little fish can also swim over.) (42) TC: Hai you dachang-yu. Hai you qianshuiting. 'Bu-le bu-le' ye lai zhao ta. (Also an octopus. Also a submarine. 'Blub blub' (it) also comes looking for it (the first fish).) ['bu-le bu-le' is onomatopoeia, presumably the sound of the submarine in the water.] [The teacher continues to solicit from the students 'friends of the fish' and 'things on the sea', while students raise their hands and also shout out items.] This art teacher typically speaks with students in a gentle manner and his tempo is slower than in speech to adults. He often uses interjections and particles (a, o, ma) to express exaggerated emotion and to solicit the students' attention and response. In this conversation, as with the zhongban teacher's use above (lines 31, 32) the teacher uses a pronoun shift, taking the student's perspective (line 37). This is common in teachers' speech to students of all ages at the preschool. Indeed, Y. R. Chao (1956: 239-40) notes that this is a general sociolinguistic rule in Chinese. 'In speaking to a child or a person of lower class, one always takes the hearer's point of view. . .'. 11 In Example 5, notice how teacher enumerates 'things in the sea' (lines 40, 41, 42), through question frames and answers (lines 38, 39). Finally, modelling of machine onomatopoeia occurs in line 42. Although the art teacher usually tempers his voice of authority with light babytalk

< previous page

page_204

next page >

< previous page

page_205

next page > Page 205

register to these older students, he does resort to a cold and punitive authoritarian voice when the situation calls for it. In Example 6 below, he switches from speech with babytalk features to an angry voice when a girl daban student rudely rejects his offer of help in drawing. Example 6 Participants: Teacher Liu [TC]; 5 year old girl student [G1]. [The teacher is sitting at a table with students, giving them suggestions on how to draw. A girl comes up to him and asks for help.] (43) G1: Laoshi, wo bu hui hua hudie. (Teacher, I can't draw a butterfly.) [She hands him her paper and he begins to draw on it, saying:] (44) TC: Ni you mei you kan guo hudie a? Aiya, hao kelian a. Jiao mama dai ni dao huayuan qu kan hudie a. (Have you ever seen a butterfly a? Good grief, how sad a. Ask mama to take you to a garden to see a butterfly a.) (45) TC: Hudie you yige shenti, dui bu dui? You mei you? You yige shenti, yizhi da hudie ( ). Ni zhao zheige jian. Kan a! (A butterfly has a body, isn't that right? Isn't that so? Has a body, one big butterfly ( ). You find these scissors. Look a!) [He holds the paper up, showing her what he is drawing. But she grabs it away.] (46) TC: ((allegro)) Zhemme mei you limao. Yong ((extra aspiration)) [q]iang de. Yihou bu bang ni zuo le. (This is lacking in politeness. Just steal it away. Next time (I) won't help you do it.) [He grimaces and waves an index finger at her, then turns away. She walks off.] This is a good example of teachers' codeswitching from his voice of authority with gentle babytalk features to an angry authoritarian one (with quickened tempo and extra aspiration). Unlike Example 1 above, where the teacher merely deleted her normal babytalk features in explicitly ordering students to comply with rules (lines 11, 13), in this example the teacher goes further and displays his extreme displeasure with the student via prosodic and kinesic shifts. At first, the teacher's speech is light and pleasant, gently mocking the student when she indicates she has never seen a butterfly (line 44). His change of demeanour when the girl jerks the paper out of his hand then is audible and visible (line 46). It should be noted that his reaction to what he labels impolite behaviour is typical for Chinese teachers and is not a personal quirk. This student's chastisement is a lesson not only for her, but also for the students around her.

< previous page

page_205

next page >

< previous page

page_206

next page > Page 206

Discussion A few comments about the issue of individual differences in teachers' uses of babytalk in the preschool are in order. Superimposed over the degree of babyishness in teachers' speech due to developmental differences in children of different ages as detailed above, there are of course individual differences in style. The head teacher in the small preschool, for example, had a much warmer, and generally more babyish style than did the second teacher. Children certainly reacted to them differently, the younger children in particular seeking the teacher out for comfort or assistance. 12 In the large preschool, although all of the 'little class' teachers spoke gently to their charges, individual differences appear especially in terms of voice quality, with the softest styles being those with a warm and often nasal quality, and also much use of expressive particles (as in the speech style of the teacher in Example 2). Other features of the Chinese babytalk register, such as special vocabulary (e.g. 'little friends', noun reduplication, onomatopoeia, games and songs), pronoun shift, tag questions, and self-repetition and paraphrasing, regularly appear in all of the 'little class' teachers' speech to some degree. Although children generally do not have control over all features that the adult babytalk register contains, they are learning it as an abstraction (Sachs & Devin, 1975), a script for interacting with others in certain situations. In addition to learning a special register for talking to younger children (e.g. Gelman & Shatz, 1977), older children can learn to use babytalk strategically with peers and adults, to evoke a subordinate or vulnerable position. As adult speakers, knowledge of a babytalk register is then available not only for its 'primary' use of addressing babies and young children, but also in secondary uses, as metaphor. This is what happens in Chinese, where a babyish speech style, encoded in the language as sajiao is first learned by children as a pleading style in interaction with powerful others (e.g. Example 3). When a babytalk register is used by one adult to another (e.g. when a young woman uses the sajiao style in courtship behaviour) the unmarked choice (normal adult speech) is 'flouted' (Scotton, 1983) and the babytalk usage, now marked, can be seen as a negotiation to establish a different RO set, one that implicates intimacy and dependency. I have been told by Chinese women that men may also use this style in intimate situations, such as talk between lovers or spouses, or in subordinate situations, such as when a male student pleads with a female professor for a deadline extension. However, some Chinese men deny that men sajiao, suggesting that this metaphorical use of a babytalk register is a contested site for the construction and transformation of a Chinese gender ideology (Farris, forthcoming). The cross-cultural evidence suggests babytalk from adults to very young children is not necessary for the child to learn a language, and indeed, babytalk as a speech register does not appear universally in all societies. A 'language of socialisation' or 'voice of authority' however, may be a universal

< previous page

page_206

next page >

< previous page

page_207

next page > Page 207

of childcare behaviour. In the Mandarin preschool example the evidence is overwhelming that Chinese speakers on Taiwan combine a voice of authority with rather heavy babytalk register and this register continues, although on a diminishing scale, through the preschool years. What is the sociocultural milieu in which caretakers of young Chinese children, both in the home and preschool settings, are motivated to use this simplified register in interactions with their charges? Chinese notions of childhood and family come into play here. Scholars agree that, 'Chinese parents tend to be highly lenient or even indulgent in their attitudes toward the infant and young child, in sharp contrast to the strict discipline they impose on older children' (David Ho, 1986:35). Parental attitudes toward and treatment of Chinese children change when children reach what is considered the age of reason, between four and six years of age. Ho & Kang (1984: 1008) studied Chinese parents and grandparents in Hong Kong and found that fathers, particularly well educated ones, have shown the most change in attitude toward child rearing. While grandfathers believed that the age at which children begin to 'understand things' (dong-shi) to be 61/2, for fathers, the age is 3 1/2, although both generations believe that children need to be disciplined by the age of three. There is also much continuity with traditional attitudes about children and human nature. Children are born 'good', and the proper nurturing of the child is thought to be far more important in personality development than are 'natural' attributes. For both generations in the Hong Kong study, competence and achievement are the most desired attributes of grown children. However, as Ho (1986:36) points out, 'Achievement motivation [in Chinese society]. . . is more firmly rooted in the collectivist than in the individualistic orientation'. The success of a child brings honour on the family and indeed is due to the parents and grandparents who have raised the child. Children owe their parents xiao, the Confucian concept of filial piety and although obligations of xiao (to produce male heirs to continue the family line, and to succour the parents in old age, as they succoured their children in youth) are not as intense as in traditional China, it remains a motivating force in the Chinese world view. 13 Returning to the linguistic socialisation of Chinese children, we have seen that parents, teachers and other caregivers, relying on a babytalk register, discursively create a warm and loving ambience in interactions with young children, within which more explicit instruction in Chinese culture can then take place (such as drilling in kin terms and professional titles, which socialises children into knowledge of proper roles in the family and in society). This protective, dependent atmosphere and the necessity to train children to be socially competent adults are apparent in the babytalk registers and the voice of authority that the preschool teachers use. Erbaugh (1982: 707) notes that Chinese believe children need help in talking properly and that the adult is giving them the means to express themselves. The quiz style of speaking with a child emphasises strictly structured question and answer routines; this style in US society is associated with a 'cold and punitive' attitude. However, in

< previous page

page_207

next page >

< previous page

page_208

next page > Page 208

Taiwan 'the emotional tone is usually loving, warm, and exuberant in both the child and adult'. Thus, cultural attitudes and sociolinguistic practices in Taiwan are seen to underlie the form and function of babytalk and other speech modifications directed at children. Scotton's codeswitching framework, which uses a natural theory of markedness, brings order and coherence to the literature on speech modifications directed at small children and variously referred to as 'babytalk', 'motherese', and 'the language of socialisation'. It is increasingly clear that such speech modifications are not 'language lessons'. Children appear to develop language-specific grammars based on a cognitively programmed universal grammar through exposure to 'diffuse environmental conditions' (Newport et al., 1977:135). Rather, the speech modifications of adults and older children to young children and babies appear to be sociolinguistic and cultural lessons, instructing the child in appropriate conversational styles and socialising the child into cultural meaning systems and social roles. The caregiver assumes certain identities in interacting with the child, i.e. that of the nurturer and protector (using a babytalk register) or a teacher and guide in an authority role (using the language of socialisation or voice of authority). As children increasingly become able to control speech registers themselves, they can negotiate identities as well, e.g. appear less mature than they really are by using babytalk strategically to evoke a heightened state of dependency. Children as young as four years old have been found to use a babytalk register to younger children and babies, showing that children are also learning to negotiate the caregiver identity as part of their acquisition of communicative competence (e.g. Sachs & Devin, 1975, Weeks, 1971). The issue of children's control over different codes points out the lack of attention to a developmental perspective in the codeswitching literature. Scholars of language acquisition and socialisation generally have ignored the work by codeswitching theorists and vice versa. More attention to research on how and when speech registers and codes are learned as part of the development of communicative competence will expand our understanding of the social dynamics underlying codeswitching. Finally, we should resist the temptation to conceptualise the meanings of code choices as a structural system apart from socially and historically constituted actors. We have seen in the example from Taiwan that a babyish communication style (sajiao) becomes imbued with gendered meanings, such that the metaphoric use of this style by adults has very different implications depending on whether the speaker is a man or a woman. The code choices and related RO sets might exist as mental scripts for social interaction, but meaning is ultimately negotiated and contested in the process.

< previous page

page_208

next page >

< previous page

page_209

next page > Page 209

Appendix I: Transcription Conventions Notation

Meaning

//

The point at which one speaker overlaps another

[]

Brackets around part of a speaker's utterance means the bracketed portion overlapped the previous speaker's utterance

If I// could [But] you can't ::

Indicates that the immediately prior syllable is prolonged

=

Indicates that no time has elapsed between the objects latched by the marks

()

Indicates that something was said but not caught by the transcriber

((lento))

double transcriptions enclose de criptions, not transcriptions

(Adapted from Zimmerman & West, 1975) Non-Segmental Phonological Features forte

loud

piano

soft

allegro

brisk in tempo, fast

lento

slow

accelerandogradually increasing in tempo rallentando gradually decreasing in tempo nasalisationresonance through the nasal cavity (Adapted from Crystal, 1971) Affective Sentential Particles1 and Interjections2 A1, A2

1: reduced forcefulness; 2: Ah!, Aha!

Le + A = La

high level tone = CRS (currently relevant state) + A1

La1**

falling tone = impatient, childish usage

< previous page

page_209

next page >

< previous page

page_210

next page > Page 210

Notation

Meaning

O1, O2

1: friendly warning; 2: Oh!

Ma1*

1: emphatic, softener

E1** (Ye), E2

1: pride, 2: vocative-Eh?!

Aiyou!2 (Aiya!)

2: Good Grief!, Drat!

Wa!2

2: Wow!

(Adapted from Li & Thompson, 1981) * see Chao, 1968 and Erbaugh, 1985 ** see Farris, 1988 Notes 1. The 1984 fieldwork was supported by grants from the Woodrow Wilson National Science Foundation Women Studies grant, a Northwest Center for Research on Women fellowship, and a Pacific Cultural Foundation grant, all of which I gratefully acknowledge. For both periods of fieldwork, I was a Visiting Scholar at Academia Sinica, Institute of Ethnology in Nankang, Taiwan. A version of this paper was presented at Columbia University department of anthropology research colloquium in February 1991. I benefited from participant comments, as well as feedback from Mary Erbaugh, Carol Eastman, and Yong-O Biq. I completed this paper while a post-doctoral Research Scholar at Columbia University, East Asian Institute, Taiwan Area Studies Program. I appreciate the kindness of the administrators, teachers, and students at the Sande and Xinyi preschools in Taibei. 2. When grammatical modifications in speech are directed at young children, presumably with the function of serving as language lessons, scholars usually refer to this speech as 'motherese'. The term explicitly shows the euro-centric bias in the research design; i.e. nuclear families with mothers as the main if not the sole socialisation agents, are taken as norm. 3. Most of the children who attended the smaller preschool were not from Catholic families. When I asked the priest how many were Catholic, he could only point out two who were. One of the teachers was a Catholic; the other was not. For Christian churches in Taiwan, operating preschools is an obvious method to reach out into a community that historically has not shown much interest in Western religions. 4. Transcription of the 1984 videotapes was carried out by Alice Liou Hsiaohsun, a PhD. candidate at the University of Washington. The 1988 data was transcribed by An Yijing, a teacher of Chinese, formerly with the Stanford Center. Both of my assistants are native Mandarin speakers from Taiwan. I am very grateful for their help. 5. Myron Cohen first suggested to me the importance of Taiwan's preschools in the development of a national culture. 6. Noun reduplication is not 'vivid', as Y. R. Chao (1968) noted, while adjective and adverb reduplication is (e.g. heiheide'really black'; manman-de'very slowly'). Rather, it has a diminutive effect, as in the dog example. This reduplication appears in many close kinterms, e.g. gege'older brother'; jiejie'older sister', and also in names of objects that a child might encounter: wawa'baby doll', wei doudou'student's bib'. Based on these models, adults apparently perceive a babytalk grammar to include a rule which generates noun reduplication as babytalk lexicon (Chao, 1968). 7. While I am not including a line of literal translation here, the glosses I do provide are not necessarily colloquial English. I have tried to convey something of Chinese discursive practices, particularly the topic/comment discourse structure and the lack of anaphora, both of which anchor speech more firmly in context than a grammar- or syntax-driven language such as English does. The expressive sentence final particles (a o ma e and la) are not translated either as their interpretation implicates the wider sociolinguistic context.

< previous page

page_210

next page >

< previous page

page_211

next page > Page 211

The reader is referred to the appendix for the approximate 'meaning' of each particle. See also Chao, 1968; Erbaugh, 1985; and Li & Thompson, 1981 for English language references on Chinese particles. For readers who wish to see the dialogues presented here in Chinese characters, please write to the author. 8. The reader should understand that some words are italicised because of prosodic marking, not because they are part of a special babytalk lexicon, as described in Table 1. For instance, in Example 1, line 3, wo:: ('I', 'me') is italicised because of exaggerated intonation (the vowel is prolonged). Whereas, in line 1, ai::ya! ('A-ya!' = 'yikes!') represents both exaggerated use of an interjection (thus part of a special lexical set) and has marked prosody. Except for noun reduplication, which is strictly a babytalk usage, other features of the Chinese babytalk register, such as sentential particles and tag questions, appear in ordinary adult speech. It is their exaggerated use in caregiver speech to young children which marks them as part of the babytalk register. 9. This ma is a question particle and not to be confused with the expressive particle ma which speakers add to a sentence to 'soften' its tone. 10. Weeks (1971: 1125) reports the use of a 'fuzzy' voice quality, often becoming nasal, used in private speech of the child, or, with a soft (piano) voice, to indicate futility, sadness, or, in lying. In the Samoan example, Ochs (1982) mentions whining as a pleading style. Surely the latter occurs in the speech of American children as well. 11. In fact, as Erbaugh (personal communication) points out, use of a person's name instead of a personal pronoun is part of polite adult speech in Chinese, in which personal pronouns are avoided as rude (second and third person) or too self-assertive (first person). The sociolinguistic rule which Chao cites is a patronising and hierarchical usage. In traditional China, a lexical set of euphemistic words (such as chen 'vassel' in speech to the emperor) was used in self-effacing first person reference. 12. During the school year preceding this one, when I did most of my participant observation, the school had about twice as many children (60) and was divided into three classes with one teacher each. When a preschool opened down the street, the new school year enrolment at the Catholic preschool dropped by half. The head teacher had previously taught the little class and her babytalk register was perhaps most closely keyed to this age group. 13. See also David Wu (1985: 127) on infant gratification and the expression of xiao in a Chinese immigrant community in Papua New Guinea. References Anderson, Elaine (1986) The acquisition of register variation by Anglo-American children. In Schieffelin and Ochs (eds) 1986 Language Socialization Across Cultures (pp. 153-64). Cambridge: C.U.P. Blom, J. P. and Gumperz, J. (1972) Social meaning in linguistic structures: Code switching in Norway. In J. Gumperz and D. Hymes (eds) Directions in Sociolinguistics: The Ethnography of Communication (pp. 407-34). New York: Holt, Rinehart and Winston. Blount, Ben (1981) The development of language in children. In Ruth Monroe, Robert Monroe and Beatrice Whiting (eds) Handbook of Cross-Cultural Human Development (pp. 379-403). New York: Garland STPM Press. (1990) Review of Elinor Ochs, 1988 Culture and Language Development: Language Acquisition and Language Socialization in a Samoan Village. Cambridge: University Press. In Language in Society 19 (4): 554-57. Chao, Y. R. (1956) Chinese Terms of Address. Language 22, 212-41. (1968) A Grammar of Spoken Chinese. Berkeley: University of California Press. Crystal, David (1971) Prosodic and paralinguistic correlates of social categories. In E. Ardener (ed.) Social Anthropology and Language (pp. 185-206). London: Tavistock. Demuth, Katherine (1986) Prompting routines in the language socialization of Basotho children. In Schieffelin and Ochs (eds) 1986. Language Socialization Across Cultures (pp. 51-79). Cambridge: C.U.P.

< previous page

page_211

next page >

< previous page

page_212

next page > Page 212

Erbaugh, Mary (1982) Coming to order: natural selection and the origin of syntax in the Mandarin speaking child. PhD thesis (Linguistics), University of California at Berkeley. (1985) Sentence final particles as an Asian areal feature. Eugene, OR, Proceedings of the First Pacific Linguistics Conference. Farris, Catherine (1988) Language and sex role acquisition in a Taiwanese kindergarten: A semeiotic analysis. PhD. thesis (anthropology). University of Washington. (1991) The gender of child discourse: Peer socialization through language use in a Taiwan preschool. Journal of Linguistic Anthropology, Dec. 1991. (pp. 198-224). (forthcoming) A semiotic analysis of sajiao, a gender linked communication style in Chinese. In Marshall Johnson and F. Y. L. Chiu (eds) Culture and Society in Taiwan. Center for Far Eastern Studies, University of Chicago Press. Ferguson, Charles (1977) Baby talk as a simplified register. In Snow and Ferguson (eds) Talking to Children: Language Input and Acquisition (pp. 209-35). Cambridge, MA: University Press. (1978) Talking to children: A search for universals. In Joseph Greenberg (ed.) Universals of Human Language Vol I (pp. 20324). Stanford: University Press. Gelman, Rochel and Shatz, Marilyn (1977) Appropriate speech adjustments: The operation of conversational constraints on talk to two-year-olds. In Michael Lewis and Leonard Rosenbaum (eds) Interaction, Conversation, and the Development of Language (pp. 227-45). New York: John Wiley & Sons. Gleason, Jean Berko (1973) Code switching in children's language. In T. E. Moore (ed.) Cognitive Development and the Acquisition of Language (pp. 159-67). New York: Academic Press. Gleitman, Lila, Newport, Elissa and Gleitman, Henry (1984) The current status of the motherese hypothesis. Journal of Child Language 11, 43-79. Harkness, Sara (1976) Cultural variation in mother's language. Word 27 (1,2,3), 495-98. Heller, Monica (1988) Introduction. In M. Heller (ed.) Codeswitching, Anthropological and Sociolinguistic Perspectives (pp. 123). Berlin: Mouton de Gruyter. Ho, David (1986) Chinese patterns of socialization: a critical review. In M. H. Bond (ed.) The Psychology of the Chinese People (pp. 1-37). Oxford: Oxford University Press. Ho, David and Kang, T. K. (1984) Intergenerational comparisons of child-rearing attitudes and practices in Hong Kong. Developmental Psychology 20 (6), 1004-1016. Hudson, R. A. (1980) Sociolinguistics. Cambridge and London: University Press. Hymes, Dell (1972) Models of the interaction of language and social life. In J. Gumperz and D. Hymes (eds) Directions in Sociolinguistics: The Ethnography of Communication (pp. 35-71). New York: Holt, Rinehart & Winston. Li, Charles and Thompson, Sandra (1981) Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Newport, E., Gleitman, H. and Gleitman, L. (1977) Mother, I'd rather do it myself: some effects and non-effects of maternal speech style. In Snow and Ferguson (eds) 1977. (pp. 109-49). Ochs, Elinor (1982) Talking to children in Western Samoa. Language in Society 11, 77-104. (1986) Introduction. In Schieffelin and Ochs (eds) 1986. (pp. 1-13). (1988) Culture and Language Development: Language Acquisition and Language Socialization in a Samoan Village. Cambridge, U.K.: University Press. Pye, Clifton (1985) Quiche Mayan speech to children. Journal of Child Language 13, 85-100. Sachs, Jacqueline and Devin, Judith (1975) Young children's use of age-appropriate speech styles in social interaction and roleplaying. Journal of Child Language 3, 81-98. Schieffelin, Bambi and Ochs, Elinor (eds) (1986.) Language Socialization Across Cultures. Cambridge: Cambridge University Press. Scotton, Carol (1983) The negotiation of identities in conversation: a theory of markedness and code choice. International Journal of the Sociology of Language 44, 115-36. (1988) Code switching as indexical of social negotiations. In M. Heller (ed.) 1988. (pp. 151-86).

< previous page

page_212

next page >

< previous page

page_213

next page > Page 213

Snow, Catherine and Ferguson, Charles (eds) (1977) Talking to Children: Language Input and Acquisition. Cambridge, MA: Cambridge University Press. Tse, John Kwock-ping (1982) Language policy in the Republic of China. In R. Kaplan (ed.) Annual Review of Applied Linguistics, 1981 (pp. 33-47). Rowley, MA: Newbury House. Watson-Gegeo, Karen and Gegeo, David (1986) Calling-out and repeating in Kwara'ae children's language socialization. In Schieffelin and Ochs (eds) 1986. (pp. 17-50). Weeks, Thelma (1971) Speech registers in young children. Child Development 42, 1119-31. Wu, David (1985) Child training in Chinese culture. In Wen-Shing Tseng and David Wu (eds) Chinese Culture and Mental Health (pp. 113-34). New York: Academic Press. Zimmerman, Don and West, Candice (1975) Sex roles, interruptions and silences in conversation. In B. Thorne and N. Henley (eds) Language and Sex: Difference and Dominance (pp. 105-29). Rowley, MA: Newbury House Publishers.

< previous page

page_213

next page >

< previous page

page_214

next page > Page 214

< previous page

page_214

next page >

< previous page

page_215

next page > Page 215

Codeswitching for Humour and Ethnic Identity: Written Danish-American Occasional Songs Marianne Stolen Danmarks Lærerhøjskole, Emdrupvej 101, Copenhagen NV 2400, Denmark Abstract This study examines individual bilingual language performance in occasional songs, focusing on the use of Danish and English, by a female member of a Danish-American organisation, Harmonien, based in Seattle, Washington. The subject occupied the dual role of Harmonien's song writer producing occasional songs for its celebrations, or for events in the lives of members, and its secretary who took minutes of its regular meetings. A prior study of her written performance as secretary, conditioned by the constraints of social role and format of the minutes, forms the backdrop to this study. An analysis of the two types of data is performed comparing the scope for creative employment of bilingual resources in the two genres. The findings confirm the hypothesis of a conditioning effect of features of genre and social role on the exploitation of two codes in writing. Occasional songs, as an expression of Harmonien group belonging and ethnic distinctiveness, contained not only types of transfer characteristic of the core-vocabulary of members and of the semi-formal minutes, e.g. integrated loans and loan translations, but also specific markers of the oral mode of interaction, not present in the minutes, e.g. nonce loans, tagswitches and codeswitching used as contextualisation cues. Introduction Theoretical Framework Codeswitching in language contact situations is a well-known strategy used by interactants to convey, metaphorically, an attitude or point of view of a socio-cultural, political or psychological nature (e.g. Blom & Gumperz, 1972; Gumperz & HernandezChavez, 1978; Valdez-Fallis, 1976). The language alternation that marks typical bilingual discourse can be compared to the phenomenon of style variation in monolingual discourse:

< previous page

page_215

next page >

< previous page

page_216

next page > Page 216

. . . The same social pressures which would lead a monolingual to change from colloquial to formal or technical styles may lead a bilingual to shift from one language to another (Gumperz, 1967:48) Where a monolingual speaker by replacing a formal address pronoun with its informal equivalent conveys an attitude of informality towards the interlocutor, a bilingual speaker may encode similar types of social meaning, such as attitude towards the dominant culture, by shifting from target language to native tongue. Such bilingual patterns of variation may characterise interaction in a whole speech community (Gumperz, 1971; Haugen, 1950; Poplack, 1979). They constitute part of a community's mode of speaking (Hasselmo, 1970). Codeswitching can be prompted, consciously or unconsciously, by socio-linguistic features of the situation, e.g. the locale, the interlocutor, or the role relationship of interactants. In addition, the process may be triggered by certain words or phrases in the discourse. This phenomenon of triggering seems to take place below speakers' conscious awareness (Clyne, 1980). The bulk of sociolinguistic studies of bilingual discourse has been on the spoken channel of communication, i.e. on interactants' oral performance in a variety of situations. The present study of a bilingual song writer's creative use of strategies such as codeswitching departs from most research on language contact. Its focus is on the genre of occasional songs as a unique written channel of communication through which a bilingual song writer is able to express social and ethno-cultural meaning. Furthermore, it incorporates the notion of social role as an important function of bilingual written expression. Methodology Data and Method of Analysis For my understanding of my subject K's production of occasional songs I examined the records of Harmonien, a DanishAmerican organisation which was established in Seattle in 1911. K functioned as Harmonien's secretary during two terms, ten years apart (from 1945-47 and again from 1953-73) and as the organisation's acting secretary at different periods of its early life. The Danish-language minutes of Harmonien meetings (covering the years 1911 till the middle of the 1970s), theatre programmes and scripts from members' performances of Danish plays, primarily from the 1920s, 1930s and 1940s, are all in the Archives and Manuscript Department of the Suzzallo Library at the University of Washington (the Elfrida Pedersen Collection). In the collection are also approximately 125 Danish occasional songs written for special events in Harmonien's life.

< previous page

page_216

next page >

< previous page

page_217

next page > Page 217

Most songs stem from the organisation's very active years mentioned above. Of the occasional songs in the collection fifty, at least, can be attributed to K (recognisable by either a signature, her special pen name, or other evidence such as idiosyncracies of spelling). Whereas all the Harmonien minutes are written in longhand, the occasional songs are typed, some of these printed in booklets and some with illustrations. According to a present long-term member of the organisation who knew K well, K would write a song '. . .every time there was a festive occasion, a Harmonien party, or a silver wedding anniversary. She produced hundreds of them'. Thus, there may have been even more songs than those preserved in the Archives (lost, or in the possession of Danish-Americans in the Community). For my linguistic analysis of K's songs I selected those which give clear evidence of K's creative use of the two codes, English and Danish. Once my scope was narrowed to twenty songs, I focused on types of linguistic expression that seemed to reflect K's use of her bilingual code in the minutes. During the analysis phase I kept in mind both the norm of the minutes and also the norm of speaking which I have observed among present-day Harmonien members and Seattle Danish-Americans in general. When examining my data for conformity or deviation in regard to these two norms, I applied Fossestöl's (1978) principle of stylistic analysis. Decoding of a text, he says, is: . . .dependent both on how a text is registered as to type, and how it differs from other texts of the same type. Thus, there is a dynamic interaction between these two forces, conformity and deviation. 1 The Subject K was born in Denmark in 1886. She emigrated with her parents to Seattle via Montana in 1904 (at the age of eighteen) and spent the rest of her life, 69 years, in Seattle, continuously involved in different aspects of Harmonien's life. She, if anyone, was familiar with the Harmonien local colour and mode of speaking. She also knew a wealth of anecdotes about members and amusing events in Harmonien's history which, as its secretary, she not only drew on in her song writing, but also entered as part of the official Harmonien records in the books of minutes. Her education in the basic skills of writing and spelling took place in Denmark, an education which bore the mark of turn-ofthe-century Danish teachers' somewhat puristic attitude to the use of foreign loans in the native language (Albeck, 1953:64). Some of K's metalinguistic comments in the Harmonien minutes bear witness to her strong desire to speak and write 'pure Danish'. As active member of Harmonien, K held two roles, both of which were constrained by established conventions: those belonging to secretarial

< previous page

page_217

next page >

< previous page

page_218

next page > Page 218

functions and those associated with being a song writer for the organisation. Furthermore, the genres themselvesminutes and occasional songscontained features conditioning her mode of expression in the bilingual and bicultural setting. The Genre of Occasional Songs The occasional song is in Danish called lejlighedssang (plural: lejlighedssange) literally translated 'event song' or 'occasion song'. The translation captures the essential characteristics of this genre: it is a song celebrating a special occasion. Thematically the Danish lejlighedssang shares features with the late nineteenth century Danish street ballad. Both can be seen to function as bearers of news, commentaries on current events, and as entertainment. For the Danish emigrants to America, a special version of the street ballad existed in the form of the emigrant ballad. It described events such as the farewell, the uprooting, and the journey, and illustrated sentiments like homesickness, nostalgia, humour, and satire (Wright & Wright, 1983). The Seattle-based Danish-Americans who emigrated around the turn of the century were probably familiar with the emigrant ballad and most certainly with the genre of the lejlighedssang. To this day, in both Denmark and Danish-American settings where heritage plays an important role, the lejlighedssang forms a set item at family celebrations and occasions such as confirmations, weddings, and anniversaries. In terms of language and style the lejlighedssang in many ways captures what Eastman (1985:3) characterises as group talkthe informal talk of a close-knit group reflecting the close relationship of its memberswhat in Danish is known as the group's fællesskab (closest translation: togetherness). Instances of use of the colloquial idiom of informal talk are frequently found in the lejlighedssang, within the constraints imposed by the verse form. Transplanted to the setting of a theatrical and social organisation such as Harmonien (in its peak years with a membership of several hundred young and recent immigrants), the lejlighedssang, in the hands of a song writer like K, became a cross between the emigrant ballad and the family lejlighedssang. It focussed on the traditional important events in the lives of the 'family' of Harmonien members, but also included the trials and tribulations associated with becoming a hyphenated American. In addition, many lejlighedssange were produced by K and other Harmonien members for the organisation's performances of light comedies, or for annual events such as its New Year Revue. Both in the country of origin and in America the lejlighedssang is normally modelled on a popular melody. This was also a feature of K's songs. She often used songs from the well-known repertoire of heritage songs brought to America by Harmonien members, their Danish cultural luggage as it

< previous page

page_218

next page >

< previous page

page_219

next page > Page 219

were. The pattern of rhymes in the stanzas of the lejlighedssange was never complicated, in keeping with the music and diction of the genre. The Previous Harmonien Study and Hypotheses for the Present Study A prior study (Stolen, 1990:107-45) of secretary K's minutes of Harmonien meetings covering two terms (from 1945 to 1947, and again from 1953 to 1973) revealed that K had internalised the Harmonien norm of writing minutes in a number of ways. She thus seemed very conscious of giving a coherent, well organised account of meetings and observed the unwritten rules for length of a report and scope for possible digressions. Her elementary school education in Denmark which, as mentioned above emphasised correct use of the native idiom was evident in her, at times, over-zealous rewriting of those items in the text which showed intrusion from the English language. This trend was particularly noticeable during the years when Harmonien struggled for survival as an ethnic organisation (during the latter part of the 1950s and early 1960s) and linguistic performance at the microlevel could therefore be related to the dimension of macro-social phenomena characteristic of the Danish-American setting. A similar macro-micro situation was observed towards the end of K's term as secretary which coincided with the onset of the ethnic revival years (the beginning of the 1970s). During this period K allowed herself more license within the accepted scope of the formal writing. Her deliberate manipulation of the two codes with humorous intent, or as a strategy for reinforcing ethnic bonds, took the form of creative loan-blends and language play in the form of transliteration (bilingual punning). Very few instances of codeswitching were found in K's minutes. I concluded that lack of evidence for this particular form of bilingual expression could be attributed to the specific type of activity in which K was engaged, i.e. reporting in concise, formal and edited language. Furthermore, the constraints of the genre did not allow for typical features of bilingual spoken discourse, e.g. codeswitching caused by time constraints, prompted by interactional discourse markers, or by trigger words or phrases. For the present study of K's use of linguistic resources in her production of occasional songs for the celebration of Harmonien events, I hypothesised that both the lejlighedssang genre and K's social role as composer of songs for ethno-cultural events would create conditions conducive to a variety of linguistic strategies within the constraints and framework of the verse form. I thus predicted that semantic transfer in the form of integrated loans, a marked feature of Harmonien's discourse, would be as evident in the stanzas of the songs as in K's minutes of Harmonien meetings. In addition I predicted that nonce loans would also form part of K's creative use of material from the dominant code, and that finally the strategy of codeswitching would be used frequently in her songs. This type of language alternation

< previous page

page_219

next page >

< previous page

page_220

next page > Page 220

would have constituted a marked feature of the informal interaction with which K was familiar from her extensive socialising with members of the ethnic organisation. Whereas in the minutes K's preferred strategy for expressing ethnic meaning took the form of loan translations (e.g. transliteration with humorous intent), a wider range of bilingual strategies for encoding social and ethnic meaning would be available in the occasional songs. The Present Study: Analysis of the Songs Using the model of the emigrant ballad, I have arranged my extracts from various songs according to typical themes, in our case depicting the Seattle-based lives of the Danish-Americans. The first song about Seattle, in which K describes the immigrants' adopted city, falls in the category of Songs Describing the Virtues of the New Country. It stems from the Harmonien New Year's revue of 1924. Another category I call Songs Containing Harmonien Local Colour, i.e. songs composed for Harmonien events with anecdotal and/or biographical information, nicknames, etc., related to members. The third category is labelled Anniversary Songs. Here we find songs celebrating special events in members' lives (silver wedding anniversaries, for instance). All of the songs analysed in my study stem from the exceptionally active years in Harmonien's life, the 1920s, 1930s and 1940s. My English versions of the extracts below are close, not poetic, translations of the texts. Songs Describing the Virtues of the New Country (1.) Seattle er en By af den rigtig gode Slags, Tacoma, Everett, Enumclaw er ej saa snavs. Og allevegne hvor du gaar-ja! you bet! Pr. Bil dertil du naar, oh! saa let. Og vi har Sø og Skov og Bjerg og Dale kan du tro. (Seattle is a city of the really nice kind, Tacoma, Everett, Enumclaw are not bad at all And everywhere you goyeah! you bet! By car you get there, Oh so easily. And we have lakes and forests and mountains And valleys I assure you.) The switching in line 3 ('you bet') is a tag-switch typical of spoken interaction among Danish-Americans (both Haugen (1950) and Hasselmo (1961) found numerous examples of these tags in their data). 'You bet' is a movable constituent which could have been inserted two other places in the quoted passage without violating Poplack's (1985:6) equivalence constraint 'which requires that the surface word order of the two languages be homologous in the vicinity of the switch point'. The example is an indication of the

< previous page

page_220

next page >

< previous page

page_221

next page > Page 221

bilingual writer's ease in manipulating her two codes in a rule-governed manner. In terms of content it seems obvious that the item conveys social rather than factual information. To the immigrant the tag would be highly symbolic of conversations in the second culture. It signals, metaphorically, an attitude (of humour) in regard to the dominant culture and can therefore be classifed as a typical example of metaphorical switching (Gumperz, 1982: 60-61). Songs Containing Harmonien Local Colour (2.) San kom vi da tilbage fra 'Camp and shore and (So we came back then fair', from 'Camp and shore and fair', med hver sin Ægtemage each with his spouse, 'say Charlie was you there', 'Say Charlie was you there', og nu med nye Kræfter And now, with renewed strength vi alle starter a´ we all start again Ja ja ja ja ja ja ja Yes, yes, yes, yes, yes, yes, yes, de samme nu og da the same back then and now) (broadcast this) (broadcast this) Glad ser vi vore Venner fra baade fjern og nær ja ræk hinanden Hænder til trofast Haandtryk hver bort med 'old man depression' ham er vi trætte a´ ja ja ja ja ja ja ja ja Klø det skal han ha' (we believe in the new deal)

(Happy we are to see our friends from near and far yes, let's join hands and sincerely shake each others' hands away with 'old man depression' we are tired of him yes, yes, yes, yes, yes, yes, yes, let's give him a beating (we believe in the new deal)

En Formand ny ved Roret Queen Anne Chris kaldes han slaar Buler nu i Bordet og vaagner alle Mand men Sekretær, Kasserer vi valgte andre ej nej nej nej nej nej nej nej forandret er de ej (none better)

(We have a new president at the helm Queen Anne Chris he is called makes dents in the table [with his gavel] and everyone wakes up, but our secretary and book keeper we've elected no one else no no no no no no no they have not been replaced (none better)

K wrote this song for Harmonien's birthday on September 16, 1933. After a long summer vacation (and a break in Harmonien meetings) the meeting at this time of year functioned as a reunion, as evidenced by the content of the song. Instead of staying in the native code, K codeswitched

< previous page

page_221

next page >

< previous page

page_222

next page > Page 222

to English a number of times. It is interesting to observe that the political issue of the time, the Depression of the 1930s, is put into focus not only by K's description in stanza two, but by her marked use of language, that is, her codeswitching, part of which appears in the chorus of the stanza. It seems as if the political climate discussed was given a certain ironic distance by being rendered in the dominant language. The exchanges of everyday conversation ('say Charlie was you there') with the typical conversational opener 'Say', followed by what seems to be a deliberate incorrect use of the verb form 'was' (instead of 'were') may be K's humorous way of describing her fellow Danes' slips of grammar in their second language. In the third stanza we see an example of inside information about Harmonien's life. The nickname, 'Queen Anne Chris', by which the Harmonien president of the time was known to members, is incorporated into the second stanza, without missing a verbal beat. As we saw in example 1, K, when switching to the other code, observes the equivalence constraint. The codeswitched item in quotation marks (stanza 1, line 4) expresses a social ritual familiar to bilingual members of Harmonien. Again we have an example of the in-group mode of communication. (3.) Naar Walter fortæller en Joke eller to sidder Anna og skænder, 'Du maa ikke No!' (When Walter tells a joke or two Anna sits and scolds, 'You mustn't, No!') Chris og Maria paa Queen Anna Hill Goodness de giver mig altid en Thrill (Chris and Maria on Queen Anne Hill goodness they always give me a thrill.) The two couplets in the example above belong to an eighteen-stanza song which K composed for a 1926 birthday celebration in Harmonien. Each single stanza presents anecdotal/biographical information about one or two of the members of the organisation. In this example I selected the two stanzas in which K skilfully used the two codes for special effect. In stanza one, an integrated loan from the Harmonien lexicon, 'Joke', fits the syllabic pattern of the first line (note that the initial letter is written with a capital 'J', a sign of the word's adoption in thee Harmonien lexicon of Danish nouns which, in keeping with Danish spelling conventions at the time, were capitalised). In line two, 'No' only rhymes with the Danish numeral to (two) if the latter is given dialectal (Jutland) pronunciation (dipthongisation of the vowel). Although there are almost no dialectal traces in K's minutes, she may have used distinctive features of her native dialect in the lejlighedssange for purposes of rhyme and to appeal to the shared geographical background of her audience, a great many of whom had a Jutland background. (In the

< previous page

page_222

next page >

< previous page

page_223

next page > Page 223

song about Seattle above, K used other dialectal features of pronunciation to create the rhyme Slags and snavs.) Stanza two contains the word, 'Goodness', a very frequent exclamation in the spoken language of Danish-Americans. What was stated about 'No' in relation to the equivalent Danish word also holds for 'Goodness'. If K had used the Danish equivalent, Du store, for instance, the line would have lost in evocative meaning. The substitution test reveals the connotative meaning of the words K chose for her songs. K codeswitches at the end of stanza two, presumably for purposes of rhyme. 'Thrill' both rhymes with 'Hill' and fits the syllable pattern of the line. (4A.)

Saa i Julivarmen kvek! kvek! Komedie vi spilled'aa! gaa væk! til den aarlige Sommerfest 'Slægtningerne' vi gay som best, Blomsterpigernes søde Blikke, modstaa dem, det ku' man skam ikke, Pic-nik fik vi slet ingen a´ alle de 'camped' hver evige Da´.

(Then in the heat of July, kvek! Kvek! we played a comedy You'd better believe me! for the annual summer party The Relations we played as best we could the flower girls' coy glances you couldn't resist, resist them, no, you really couldn't. A picnic we never had everyone was camping everyday.)

(4B.) Om Som'ren vi fredelig hviler os, (In the summer we take it easy, paa vore Laurbær 'Pionæren's Ros. resting on our laurels, credit to Men nu skal I høre hvordan det the Pioneer gik, But now listen to what happened omtrent hver Søndag vi Pic-nic fik. almost every day we had a picnic Tra-la-la-la-la osv. Tra-la-la-la-la etc.) (4C.) Hvor herligt deter og hvor mange Picnic vi har Til Renton Juction den store picnic sidst jo var.

(How wonderful it is and how many picnics we have. At Renton Junction we had our last big picnic, as you know.)

(4D.) Hos Møller Lehman malede jeg baade Mel og Gryn. Hos Pluimberen jeg kørte rundt med W.C. sikket syn Jeg lirkede og limed paa Adolf Hansens Shop. Hos Mikkelsen som Mælkekusk, det var min sidste Job.

< previous page

(At miller Lehman's I ground both flour and grain. At the plumber's I drove around with toilets, what a sight. I fitted and glued things in Adolf Hansen's shop, at Mikkelsen's as milkman I had my last job.)

page_223

next page >

< previous page

page_224

next page > Page 224

(4E.) Som Chief for vores Politi Staar T.I. som den første Og Bageren som dry squad gaar Mens Folk maa gaa og tørste. Og lille Peet som Traffic-Cop Og Chris som Garbage Formand. Ja Danskeren, han gør det godt Her i hans nye Hjemland.

(As chief for our police T.I. is the first and the baker works as the dry squad, while people go around feeling thirsty and little Peet is traffic-cop and Chris is garbage-president, Yes, the Dane is dong fine here in his new country)

I chose all of the examples above to illustrate K's use of L2 loans in her songs. In the analysis of data from K's minutes was discussed K's variant spelling of the word picnic before it became a fully integrated loan, capitalised and without quotation marks. In the example above, dating from the latter part of the 1920s, variations in her spelling of this word are also apparent, even within the same stanza (example 4C). In excerpts 4A and 4B, we observe a variant not found in the minutes, a separation of the two-syllable word with a hyphen. This may be a poetic spelling. In 4D and 4E there are six examples of K's use of loans, most of which seem to be used for the description of typical occupations held by Harmonien members. The word Garbage-formand exemplifies K's use of a loan-blend. In excerpt 4E, K seems to have made particularly good use of her bilingual resources for creating effective meter and rhyme. Thus, 'cop' makes a nice rhyme for the Danish word godt. The loan 'dry squad', presumably describing a well-known 1930s occupation, is interestingly enough inserted in the line in the English form without capitalisation. One might guess that the item was more of a nonce word than the nouns describing more traditional occupational roles. Anniversary Songs (5A.) Stork, Stork, vor gamle Ven Vi drikke til Din Ære Thi uden dig og dine lange Ben Her kunde intet Sølvbryllup være

(Stork, stork, our old friend let's drink to your honour since without you and your long legs here would be no silver wedding anniversary).

Stork, Stork, (What again?) Ja, jeg fløj tilbage Gay Doris og Mand en lille Øjesten Som gjorde Bestefaer af Tage

(Stork, stork, (What again?) yes, I flew back gave Doris and her husband a little sweetie who made a granddad of Tage)

< previous page

page_224

next page >

< previous page

page_225

next page > Page 225

(5B.) Der stod den lille Laura, hvor hun rødmed, hun var sød, hans Hjerte næsten smelted' ja-det blev ganske blød, hvordan det vid're gik ja-den Historie er lang, jeg troer at først de taltes ved i Portens Trappegang, snart de vandrede Arm i Arm til Lunden ud. efter han fik sagt, vil Du vær' min little Brud, nu han fløjted'sang, oh hvor Høvl og Saven gik, medens Laura nynned' glad i sin Bageri Butik. (There stood little Laura, how she blushed, she was cute, his heart almost melted, yes it became quite soft, what happened after that, well, that's a long story, I believe they first met in the stairwell, soon they were walking arm-in-arm out to the grove after he had asked, would you be my bride, now he whistled, sang, Oh how he worked his plane and his saw, while Laura was happily humming in her bakery store.) (5C.) I Aftenskole Emmy i Vinter gik vi dette si'er og lærte engelsk tem'li godt, hun en expert bli'er 'she speaks a perfect lingo, perfect lingo with lots of pep and bingo yes! English spoken here'. (To evening school went Emmy this winter, we want to mention this, and learned English rather well, she becomes an expert, 'she speaks a perfect lingo perfect lingo with lots of pep and bingo yes! English spoken here'.) Above are three examples of wedding anniversary songs from the 1920s which illustrate K's competence in this subcategory of the lejlighedssang. In text 5A, in six out of the seven stanzas in the songs, K has Stork as the first word in the first line, sometimes repeated for effect. Using the codeswitched item in parenthesis ('What again?') rather than the Danish equivalent term Nu igen?, gives stylistic variety to the statement. By using words from the English code, K also manages to observe the end rhyme pattern: a,b,a,b. In text B, K uses the English word 'little' in combination with the near-

< previous page

page_225

next page >

< previous page

page_226

next page > Page 226

identical Danish adjective lille, apparently for purposes of stylistic variation. In the stanza following the one quoted here, K combines the English adjective with the proper noun, Laura ('little Laura'), again, it seems, for stylistic purposes. The song is almost generic in type, relating the successful immigration story of Adolf and his lille Laura. The attributive adjective 'little' is seen the first time in the stanza describing how Laura becomes Adolf's bride ('lille Brud'). The couple soon after settles in Copenhagen, catches America fever, and emigrates. After approximately five years in Seattle, during which Laura 'spared hver en cent' (saved every cent), homesickness drives the couple back to 'Københavnerlivet' (Copenhagen life). However, after a year in Copenhagen they have had enough of this life and return to 'denne store Flok' (literally, this big flock, that is, the Seattle Danes). In this example K successfully manages to fuse two categories of stories into her lejlighedssange, the prototypical Danish-American Seattle emigration story and the silver wedding anniversary story. K's second use of an English item in the text, the word 'cent', occurs at a time in the narrative when Laura is going through acculturation in her adopted country. The word could therefore be said to function as an effective ethnic marker in the text. Text C, a celebration of a couple's one-year anniversary, similarly contains elements of local colour relating to Danish-American life in Seattle. The extract is taken from the section of the narrative in which K describes Emmy's (the wife's) struggle with the English language. In Seattle, new immigrants desiring citizenship would normally study the English language at night school in order to pass the citizenship exam. The quoted stanza refers to this stage of Emmy's life. She attends Aftenskole (night school) and becomes an 'expert' in the second language ('she speaks a perfect lingo'). This example clearly shows K's deliberate use of codeswitching to communicate subtleties of meaning. K's matephorical switching may have created in her audience a sense of ethnic fællesskab as members identified with K's adjustment problems as newcomer. As a summary of our analysis of K's skills as bilingual composer of lejlighedssange, we can state that her creative exploitation of her language resources was very much in evidence. In verse form she seems to have captured some essential features of Danish-American life and mode of interaction at the time. Haugen (1984) in his introduction to his selection of Peter J. Rosendahl's comic strips (based on life in a Norwegian-American community in Minnesota in the 1920s), maintains that Rosendahl's recreation of local speech patterns reflected 'Spring Grove Norwegian' (the bilingual mode of speaking in Spring Grove, Minnesota). Thus, Rosendahl's comic strips would strike chords of recognition in his audience of readers (the strips appeared in the Norwegian language newspaper Decorah Posten from 1918 to 1935). 'This is our kind of talk', they would say, 'just as funny as we ourselves' (p. 28). Possibly the Harmonien members felt the same about the songs authored by K for their many gatherings. The songs were

< previous page

page_226

next page >

< previous page

page_227

next page > Page 227

certainly much in demand, one good sign that K had tuned into an interaction style that Harmonien members recognised as their own. A present-day member, Harmonien's oldest member, commented that the songs had the effect of bringing her back in time. In K's themes and diction she seemed to recognise the ways of the 'Store Flok', to use K's own term for Seattle's tightly-knit Danish community of the 1920s, 1930s and 1940s. Some of the categories we discussed above, interjections and tags for instance, are specimens of a large category of discourse markers with high emotional content. They are habitually used in informal interactions of Danish-Americans (for example, 'you bet' and 'goodness'). Observing genre-imposed constraints and also textual constraints such as the Poplack equivalent principle, K used her two languages (via codeswitching) to convey humorous aspects of Harmonien's life and to reflect the group's linguistic and ethnic identity. If we compare K's songwriting with the data of her minutes, the prior study referred to above, we can see that our hypothesis has been confirmed. The language of the songs contains markers of bilingual conversation not present in the minutes. In the latter type of data we found virtually no examples of codeswitching (in a total of 223 pages only two clear examples of this linguistic phenomenon were found both of which seemed to be instances of unconscious mixing) whereas this strategy was frequently used in the songs with deliberate communicative intent. For instance, K switches back and forth between 'little' and 'lille', presumably with the intent of evoking special sentiments in her audience, or switches to English for a longer passage of speech (example 5B) to make an effective statement about a well-known aspect of the newcomer's life. We also saw how switching to the other code (for example, 'thrill' and 'No', in example 3) served both a poetic (rhyme/meter) and a humorous function, unique to this type of data. K's creative use of loans gives similar evidence of her ability to mirror the insertion of borrowed material which characterises casual bilingual conversation (for example, 'Joke' and 'Garbage Formand'). Conclusion In conclusion, the songwriting mode and K's role as song composer for Harmonien both seem to have prompted her imaginative use of vocabulary from the combined pool of her two languages for stylistic variation and to create an atmosphere of humour and shared ethnic experience (fællesskab). We see here a parallel to instances of her creative and intentional juxtaposition of the two codes in the minutes. In the latter data the role of secretary and the genre limited the scope of her strategies for language play and bilingual humour to mainly loan translations. In the lejlighedssange, on the other hand, the conventions of this specific genre of song writing, K's social role of song writer, and the thematic possibilities offered by the immigrant

< previous page

page_227

next page >

< previous page

page_228 Page 228

lejlighedssang made for a wider range of effective bilingual strategiescentral among them codeswitchingfor both language play, play with meter and rhyme, and the signalling of common heritage and ethnic distinctiveness. Notes 1. S. Zempel's translation. In Zempel, S. (1980) Language Use in the Novels of Johannes B. Wist: A Study of Bilingualism in Literature. PhD thesis, Univ. of Minnesota, p. 18. References Albeck, U. (1953) Dansk Stilistik. Copenhagen: Gyldendal. Blom, J. and Gumperz, J. J. (1972) Social meaning in linguistic structures: code switching in Norway. In J. Gumperz and D. Hymes (eds) Directions in Sociolinguistics: The Ethnography of Communication. New York: Holt, Rinehart and Winston. Clyne, M. (1980) Triggering and language processing. Canadian Journal of Psychology/(Rev. Canad. Psychol.) 36, 400-406. Eastman, C. (1985) Establishing social identity through language use. Journal of Language and Social Psychology 4, 1-20. Fossestol, B. (1978) Tekstlingvistiske stilmarkorer. Norskrift: Arbeidsskrift for nordisk språk og litteratur 20, 1-14. Gumperz, J. (1967) On the linguistic markers of bilingual communication. Journal of Social Issues 22, 48-57. (1971) Language in Social Groups: Essays by John J. Gumperz. Stanford: Stanford University Press. (ed.) (1982) Discourse Strategies. Cambridge: Cambridge University Press. Gumperz, J. J. and Hernandez-Chavez, E. (1978) Bilingualism, bidialectalism and classroom interaction. In M. Lourie and N. Conklin (eds) A Pluralistic Nation. Rowley, Mass.: Newbury House. Gumperz, J. J. and Hymes, D. (eds) (1972) Directions in Sociolinguistics: The Ethnography of Communication. New York: Holt, Rinehart and Winston. Hasselmo, N. (1961) American Swedish. PhD thesis, Harvard University. (1970) Code-switching and modes of speaking. In G. Gilbert (ed.) Texas Studies in Bilingualism. Berlin: de Gruyter. Haugen, E. (1950) The analysis of linguistic borrowing. Language 26, 210-31. (1984) The Language of 'Han Ola og han Per'. In J. N. Buckley and E. Haugen (eds) J. Rosendahl. 'Han Ola og han Per'. Oslo: Universitetsforlaget. Poplack, S. (1979) Sometimes I'll start a sentence in English y termino en espanol: toward a typology of code-switching. Language Policy Task Force. New York. (1985) Contrasting patterns of code-switching in two communities. Ms, 1-39. Stolen, M. (1990) Harmonien: An ethnohistorical sociolinguistic analysis of a Danish-American organization. PhD thesis, University of Washington. Valdez-Fallis, G. (1976) Social interaction and code-switching patterns: A case study of Spanish/ English alternation. In G. D. Keller et al. (eds) Bilingualism in the Bicentennial and Beyond (pp. 53-85). New York: Bilingual Press. Wright, R. and R. L. Wright (eds) (1983) Danish Emigrant Ballads and Songs. Carbondale: Southern Illinois University Press. Zempel, S. P. (1980) Language use in the novels of Johannes B. Wist: A study of bilingualism in Literature. PhD thesis, Univ. of Minnesota.

< previous page

page_228