Immigrant Languages in Europe

cover

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

next page >

Immigrant Languages in Europe Extra, Guus. Multilingual Matters 1853591793 9781853591792 9780585120850 English Immigrants--Europe--Language, Europe--Languages, Language acquisition, Sociolinguistics--Europe. 1993 P380.I45 1993eb 306.4/4/094 Immigrants--Europe--Language, Europe--Languages, Language acquisition, Sociolinguistics--Europe.

cover

next page >

< previous page

page_i

next page > Page i

Immigrant Languages in Europe

< previous page

page_i

next page >

< previous page

page_ii

next page > Page ii

Multilingual Matters Alsatian Acts of Identity LILIANE M. VASSBERG Attitudes and Language COLIN BAKER Breaking the Boundaries EUAN REID and HANS H. REICH (eds) Citizens of This Country: The Asian British MARY STOPES-ROE and RAYMOND COCHRANE Community Languages: A Handbook BARBARA M. HORVATH and PAUL VAUGHAN Continuing to Think: The British Asian Girl BARRIE WADE and PAMELA SOUTER Education of Chinese Children in Britain and the USA LORNITA YUEN-FAN WONG European Models of Bilingual Education HUGO BAETENS BEARDSMORE (ed.) Life in Language Immersion Classrooms ELIZABETH B. BERNHARDT (ed.) Linguistic and Communicative Competence CHRISTINA BRATT-PAULSTON Linguistic Minorities, Society and Territory COLIN H. WILLIAMS (ed.) Migration and Intercultural Education in Europe U. PORNBACHER (ed.) Minority Education: From Shame to Struggle T. SKUTNABB-KANGAS and J. CUMMINS (eds) Multilingualism and Nation Building GERDA MANSOUR One Europe - 100 Nations ROY N. PEDERSEN Opportunity and Constraints of Community Language Teaching SJAAK KROON Reversing Language Shift JOSHUA A. FISHMAN Sociolinguistic Perspectives on Bilingual Education CHRISTINA BRATT PAULSTON Please contact us for the latest book Information: Multilingual Matters Ltd, Frankfurt Lodge, Clevedon Hall, Victoria Road, Clevedon, Avon BS21 7SJ, England

< previous page

page_ii

next page >

< previous page

page_iii

next page > Page iii

Immigrant Languages in Europe Edited by Guus Extra and Ludo Verhoeven MULTILINGUAL MATTERS LTD Clevedon · Philadelphia · Adelaide

< previous page

page_iii

next page >

< previous page

page_iv

next page > Page iv

Library of Congress Cataloging In Publication Data Immigrant Languages in Europe/Edited by Guus Extra and Ludo Verhoeven p. cm. Includes bibliographical references and index. 1. Immigrants-Europe-Language. 2. Europe-Languages. 3. Language aquisition. 4. Sociolinguistics-Europe. I. Extra, Guus. II. Verhoeven, Ludo H. P380.I45 1993 306.4'4'094 dc20 British Library Cataloguing In Publication Data A CIP catalogue record for this book is available from the British Library. ISBN 1-85359-179-3 (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 © 1993 Guus Extra, Ludo Verhoeven and the authors of individual chapters. 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 WBC Print, Bridgend.

< previous page

page_iv

next page >

< previous page

next page >

page_v

Page v

CONTENTS Introduction Immigrant Groups and Immigrant Languages in Europe Guus Extra and Ludo Verhoeven

3

Part 1: Use of Immigrant Language Varieties in Europe 1. Sweden Finnish Jarmo Lainio

21

2. South Asian Languages in Britain Safder Alladina

55

3. A Bilingual Perspective on Turkish and Morrocan Children and Adults in the Netherlands Guus Extra and Ludo Verhoeven

67

4. Croatian or Serbian as a Diaspora Language in Western Europe Andrina Pavlinic

101

Part 2: First Language Acquistion in a Second Language Context 5. Turkish Language Development in Germany Carol Pfaff

119

6. Turkish Language Development in the Netherlands Anneli Schaufeli

147

7. Summative Assessment of Ethnic Group Language Proficiency Jeroen Aarssen, Jan Jaap De Ruiter and Ludo Verhoeven

159

8. Parental Attitudes towards Child Bilingualism in the Nordic Countries Sirkku Latomaa

181

< previous page

page_v

next page >

< previous page

next page >

page_vi

Page vi Part 3: Code-Switching 9. Code-copying in Immigrant Turkish Lars Johanson

197

10. Turkish-Dutch Code-switching and the Frame-Process Model Ad Backus

223

11. Code-switching and Borrowing in an Arabic-Dutch Context Jacomine Nortier

237

12. Finns and Americans in Sweden: Patterns of Linguistic Incorporation from Swedish Paula Andersson

249

Part 4: Language Maintenance and Language Loss 13. Immigrant Minority Languages and Education in Sweden Sally Boyd

273

14. Romani at the Crossroads Donald Kenrick

285

15. Methodological Issues in Language Shift Research Koen Jaspaert and Sjaak Kroon

297

16. Lexical Aspects of Language Attrition and Shift Kees De Bot and Bert Weltens

309

Authors and Affiliations

323

< previous page

page_vi

next page >

< previous page

page_vii

next page > Page vii

ACKNOWLEDGEMENTS This volume contains a collection of updated versions of papers presented at an international colloquium on immigrant language varieties in Europe. The colloquium was held in Gilze-Rijen, the Netherlands, in December 1990. We want to thank the Royal Dutch Academy of Sciences for the financial support of this colloquium and Concita Barbé, Rieky Jurriëns and Anneke Smits for their secretarial support in the process of editing this volume. We owe special gratitude to Jetske Folmer, who has been most helpful in both the organization of the colloquium and the compilation of the report. GUUS EXTRA AND LUDO VERHOEVEN

< previous page

page_vii

next page >

< previous page

page_1

next page > Page 1

INTRODUCTION

< previous page

page_1

next page >

< previous page

page_ii

next page >

< previous page

page_3

next page > Page 3

Immigrant groups and immigrant languages in Europe Guus Extra & Ludo Verhoeven The focus of this volume is on immigrant groups and immigrant languages with a recent or earlier background of migration to industrialized countries in Western and Northern Europe. After having presented some basic concepts in the area of language and immigration, we will focus on demographic statistics of immigrant groups in European Community countries and Scandinavia, and on research in the field of immigrant language varieties. Finally, we will outline the structure of the present volume. Basic concepts It is a common pattern across the world that immigrant groups are confronted with a mismatch between the language varieties of their source and target countries. During the twentieth century, the USA has been a target country for many European immigrants, whereas industrialized European countries have been the target area for many Third World immigrants during the past decades. Due to processes of immigration, there is a growing cross-atlantic resemblance between Europe and the USA with respect to psychological, sociological, linguistic and educational issues in the area of bilingualism and language contact. It is, however, important to note that American and European concepts of bilingualism are used within rather different contexts. Whereas English is unvariably the high prestige variety in the USA, it is not the only high prestige language in Europe. Due to the existence of different nation-states and different national languages in Europe, second language acquisition by immigrant adults or children does not always mean learning English, but a whole range of different languages. In addition, having to cope with more than one language is a common experience for immigrant groups in the USA, whereas it is a rare experience for AngloAmericans. Given the fact that monolingualism is the unmarked case of language use in the USA and bilingualism the

< previous page

page_3

next page >

< previous page

page_4

next page > Page 4

marked one, many American studies on bilingualism include chapters on effects of bilingualism and few studies pay attention to effects of monolingualism. In Europe, learning more than one language is a common and often highly valued experience of many citizens. To give an example: at the traditional and still highly prestigious gymnasium in the Netherlands, six languages are taught and learnt, i.e. Dutch, English, French, German, Latin and Greek; together, these languages make up more than half of the secondary school curriculum. As a consequence of socio-economically or politically determined processes of immigration, the traditional patterns of language variation across Europe have been considerably extended over the past decades. Many industrialized European countries show evidence of a growing number of immigrant populations which differ widely, both from a cultural and from a linguistic point of view, from the maistream indigenous population. It has been estimated that in the year 2000, one third of the population under the age of 35 in urban Europe will have an immigrant background. Within European Community countries, four major immigrant groups can be distinguished: people from Mediterranean EC countries, from Mediterranean non-EC countries, from former colonial countries, and political refugees. First, an economically or politically motivated process of migration took place. Especially in the case of Mediterranean groups, migration initially most often involved contract workers who expected - or were expected - to stay for a limited period of time. As the period of their stay gradually became longer, the pattern of economic migration was followed by a pattern of social migration of families. Subsequently, a second generation was born in the immigrant countries and grew up with parental uncertainty and ambivalence about whether to stay or return. These demographic shifts over time are accompanied by shifts of denotation for the groups under consideration (migrant workers, immigrant families, and ethnic minorities respectively), and by repeated demographic announcements of governmental authorities that their country ''should not be seen as an immigrant country". Shifts of denotation for the groups under consideration not only derive from chronological changes over time, but also from the perspective of self-definition versus other-definition. The concept of "ethnic minorities" commonly relates to groups whose members belong to different generations and who are considered by themselves or by others as different from a majority group in terms of number and spread, socio-economic status and power, and language and culture (often including religion). Differences in language, culture and religion are basic

< previous page

page_4

next page >

< previous page

page_5

next page > Page 5

to the concept of "ethnicity", although this concept does not necessarily include all three ingredients. Language use is a highly observable marker of group identity, and in many cases it is a core value of identity (Smolicz, 1980). It has, however, been acknowledged that language is not necessarily a prerequisite for the intergenerational maintenance of group identity. To give another example of Dutch experiences: Dutch immigrants in Australia commonly define themselves as Dutchmen; at the same time, they show a strong shift to mainstream English within one generation (Clyne, 1991). There are many instances of ethnic groups with distinct languages, but also many instances of distinct ethnic groups without distinct languages. Ethnic differences do not always find parallels in linguistic differences, and vice versa. With respect to these issues, two major opposite trends can be observed within European Community countries. Processes of unification and internationalization through mass media, trade, labor migration and tourism go together with a growing awareness of the significance of cultural and linguistic diversity. This concern, however, relates most commonly to nation-states and national languages, less frequently to regions and their indigenous language varieties (e.g., Welsh or Frisian), and least frequently to immigrant groups and immigrant language varieties. As a consequence, the concept of European Community languages has the unmarked connotation of referring to the official or national languages of EC member states (Coulmas, 1991) and only rarely refers to indigenous or non-indigenous minority language varieties. In contrast, the concept of community languages in Australia refers to both the indigenous language varieties of Maori groups and the non-indigenous language varieties of widely different immigrant groups (Clyne, 1991). European research cooperation on indigenous minority languages received an impetus in 1980, when the first international conference on this topic took place in Scotland (Glasgow). This conference turned out to be the beginning of a successful series of conferences, going from Finland (Abo/Turku 1983), via Ireland (Galway 1986) to Friesland (Leeuwarden 1989). Papers from the Leeuwarden conference were recently edited by Gorter et al. (1990). Additional background information on both indigenous and nonindigenous minority languages in Europe can be derived from Kroon et al. (1991) and Jaspaert & Kroon (1991). For a historical perspective we refer to Engman (1992). A basic policy question (especially in the area of education) is how to reconcile the opposite trends of unification and diversification of

< previous page

page_5

next page >

< previous page

page_6

next page > Page 6

different communities in Europe. This question also dominates the language debate, given the fact that languages are prominent features of both trends and also that, within educational contexts, languages are relatively easy to promote or neglect. In all European Community countries, a top-down insistence of indigenous majority groups on second language acquisition by nonindigenous minority groups can be observed in combination with a bottom-up insistence of minority groups on first language maintenance over time (Extra & Vallen, 1989). As yet, the European Council Directive (1977) on the education of immigrant children has been a rather weak attempt to cope with these opposite pressures (Reich, 1991). Demographic statistics Comparative quantitative information on population figures in European Community member states can be derived from the Statistical Office of the European Community in Luxembourg (EuroStat). In all 12 EC members states, an overall decrease of the indigenous population can be observed over the last decade. Apart from Ireland, birth rates show a tendency to remain below the level of population reproduction rates. Between 1964 and 1985, there was a decrease of 40-45% in the birth rate of Germany and Italy, and about 30% in France, Great Britain and the Netherlands (Wilpert, 1991, for further details). The widely supported prediction is that this trend is irreversible. During the same decades, an increase of immigration figures in EC member states can be seen. Although free movement of migrants between EC member states is legally permitted and promoted, most immigrants in EC countries originate from abroad. The largest immigration figures have been observed in France and Germany (8% and 6% respectively of the total population in 1989). For various reasons, reliable demographic information on immigrant groups in Western and Northern Europe is difficult to obtain. For some groups or countries, no updated information is available or no such data have ever been collected at all. Moreover, official statistics only concern immigrant groups with a legal residence status. Most importantly, however, the most widely used criteria for immigrant status, i.e., nationality and/or country of birth, show strong erosion over time because of an increasing trend toward naturalization and births within the country of residence. In addition, residents from former colonies often have the nationality of their country of immigration.

< previous page

page_6

next page >

< previous page

page_7

next page > Page 7

Table 1 Estimated numbers of 9 Mediterranean foreign resident groups in 10 EC countries (data derived from EuroStat 1991; --- = unknown, lower than 10.000) (data on Netherlands derived from Muus 1991) PO SP IT GR YU TU TN AL MO Belgium 13.500 52.500 241.000 20.600 5.400 79.500 6.200 10.600 135.500 300 800 1.900 500 9.100 26.100 300 300 2.500 Denmark France 764.900 321.400 333.700 7.900 64.400 123.500 189.400 795.900 431.100 Germany 74.900 127.000 519.500 293.600 610.500 1.612.600 24.300 5.900 61.800 Great Britain 18.000 28.000 78.000 14.000 --18.000 ------Greece 400 1.000 6.700 --- 1.900 3.200 300 200 300 Italy 2.200 7.500 --- 12.700 11.900 1.800 10.900 1.400 21.600 Netherlands 8.000 17.400 16.700 4.500 12.800 191.500 2.400 700 148.000 Portugal --- 7.300 1.100 50 60 30 20 30 50 Spain 32.900 --- 15.300 700 400 200 300 700 14.500 Total 915.100 562.900 1.213.900 354550 716.460 2.056.430 234.120 815.730 815.350

< previous page

page_7

next page >

< previous page

page_8

next page > Page 8

In contrast to countries with a longer immigration history, such as the USA, Canada, and Australia, there is no tradition in Europe with respect to collecting data on the basis of an alternative or at least complementary criterion: home language use as a major indicator of group-bound ethnicity (De Vries, 1989: 68). What is available are official statistics on "foreign residents" in the European Community, produced by the Statistical Office of the EC in Luxembourg (EuroStat, 1991). Table 1 gives the estimated numbers of nine major Mediterranean foreign resident groups in ten EC countries; four groups originate from EC countries (Portugal, Spain, Italy, Greece), five from non-EC countries (Yugoslavia, Turkey, Tunesia, Algeria, Morocco). The EuroStat data on the Netherlands have been adapted in upward direction on the basis of Muus (1991), and similar tendencies are expected for other EC countries. There are large differences among EC countries as regards the size and composition of immigrant groups. Owing to labour market mechanisms, immigrant groups concentrate in northern industrialized EC countries, whereas their presence in Mediterranean countries like Greece, Italy, Portugal, and Spain is rather limited. Immigrant groups in these four Mediterranean countries originate to a great extent from neighbouring countries, e.g., the Italian and Turkish communities in Greece, the Spanish community in Portugal, or the Portuguese community in Spain. Moreover, in most EC countries, there is a trend towards an increase of immigrants from non-EC countries, and a decrease of immigrants from EC countries. For the nine Mediterranean groups under consideration, France or Germany are the major countries of immigration. Portuguese, Spanish, and Magreb residents concentrate in France, whereas Italian, Greek, Yugoslavian, and Turkish residents concentrate in Germany. The largest immigrant group in EC countries is made up by the Turkish community. Not included in Table 1 are data on ex-colonial groups (apart from the people from Magreb countries in France) and on political refugees. Political refugees originate from a wide variety of countries, whereas ex-colonial groups concentrate only in specific EC countries (e.g., the South-Asian community in Great Britain or the Surinam community in the Netherlands). For Scandinavia, estimates on the previously mentioned groups can be derived from the Nordic Statistical Secretariat (1991). The relevant data are based on population figures according to citizenship and, apart from Denmark, presented in Table 2.

< previous page

page_8

next page >

< previous page

page_9

next page > Page 9

Table 2 Estimated numbers of previously mentioned groups in 3 Scandinavian countries (--- = unknown) Finland (1988) Norway (1990) Sweden (1989) Portugal 50 400 1.500 Spain 300 900 2.900 Italy 300 800 4.000 Greece 200 4.100 12.000 Yugoslavia 50 3.900 39.600 Turkey 100 5.300 24.200 Tunesia ------Algeria 100 300 500 Morocco 100 2.100 1.200 Total 1.200 17.800 61.700 Finland has the lowest figures and Sweden the highest ones. Yugoslavia, Turkey and Greece are the most important source countries of the groups under consideration in both Sweden and Norway. In all Scandinavian countries, however, figures for mutual population exchange are much higher. To give an example: the estimated number of Fins in Sweden was 124.000 in 1989, whereas the estimated number of Swedes in Finland was 5.400 in 1988. Research on immigrant language varieties Migration processes in recent history have led to a growing social and scientific interest in language contact phenomena at the levels of society and the individual. Traditionally, there has been a clear bias in the orientation of both policy and research towards second language acquisition (Extra, 1990). With respect to policy the basic philosophy of the melting pot called on immigrants to shed their own language and culture in favour of the majority language and culture. With respect to research this biased interest is reflected in the almost exclusive focus on the use and acquisition of dominant languages. In comparison with the large number of studies in the latter area, the number of studies on ethnic community languages has been extremely small. Research groups focusing on non-indigenous language varieties followed the philological tradition of historical analysis. There was only limited attention for research on these languages as they are spoken in the here-and-now and

< previous page

page_9

next page >

< previous page

page_10

next page > Page 10

most of the available data were generally ignored. This biased orientation in research on immigrant communities speaking a non-indigenous minority language could be seen in Northern America (Grosjean, 1986), Australia (Clyne, 1982), and Europe (Tosi, 1984). In the past decade there has been a shift in research from almost exclusive attention to the dominant language to a broader focus on sociolinguistic and psycholinguistic dimensions of multilingualism. Attention was given to the following rather new research paradigms: • the socio-political context of non-indigenous minority language use at the macro-level; • the attitudes towards non-indigenous minority languages and their speakers; • the use of non-indigenous minority languages in conversation, depending on the participants, the topic and the social setting; • the interactions between dominant and dominated languages in terms of language transfer and code-switching; • the process of acquisition of the mother tongue in a second language immersion or submersion environment; • the factors determining first language maintenance, shift and loss; • the emergence of new language varieties due to restricted language input and contact with dominant other languages. A landmark for this shift of orientation in Europe was the Linguistic Minorities Project (1985), followed by studies on nonindigenous language varieties in Sweden (Boyd, 1985), France (Vermes, 1988), Great Britain (Alladina & Edwards, 1991) and the Netherlands (De Ruiter, 1991; Extra & Verhoeven, 1992). Apart from these general overviews, a number of studies focussed on the status and use of specific immigrant languages, such as Italian (Fantini, 1985), Arabic (Nortier, 1989; De Ruiter, 1989), Finnish (Lainio, 1987), Serbo-Croatian (Pavlinic et al., 1988) and Turkish (Boeschoten, 1990; Pfaff, 1991; Schaufeli, 1991; Verhoeven, 1991a). Thanks to the research that has been done so far, several trends on the status and use of immigrant languages in Europe can be traced. With respect to the socio-political context there is a movement away from the acknowledgement of linguistic diversity towards more assimilatory policies. During the eighties, in many Northern and Western European countries the general policy allowed room for first language maintenance of immigrant groups. As to the mass media, immigrant languages were given the opportunity to serve particular public functions in the community. As to education, a number of initiatives were taken which

< previous page

page_10

next page >

< previous page

page_11

next page > Page 11

aimed at offering mother-tongue instruction (Reich, 1991). However, in recent years in several countries an assimilatory policy has been advocated, based on the argument that the adaptation of immigrant groups would be in their own interest. The socio-political context, among other factors, influences the attitudes towards immigrant language varieties. Studies by, for example, Boyd (1985), and Van de Wetering (1990) show that immigrant groups in European countries usually highly value their native language. Moreover, it turns out that the native language is used in many domains which are related to the family and the wider ethnic community. A number of studies have focussed on processes of borrowing and code-switching, resulting from the alternate use of dominant and dominated languages. The constraints of grammatical integration, as formulated by Poplack (1980) with reference to a more or less symmetrical use of two related languages (i.e. Spanish and English), were tested. The first constraint is the free morpheme constraint stating that a switch cannot occur between a lexical form and a bound morpheme unless the former has been phonologically adapted into the language of the latter. The second constraint is the equivalence constraint which states that the word order immediately before or after the switching point must exist in the two languages to make it possible for a switch to take place. It is an open question whether the constraints claimed so far can be generalized to speech communities with a quite different sociolinguistic background, such as that of Turks or Moroccans in Western Europe. Boeschoten and Verhoeven (1985; 1987), Nortier (1989) and Backus (1992) found that their code-switching data could not be adequately described in terms of the above-mentioned constraint rules. With respect to the acquisition of ethnic community languages it is clear that the submersion in a second language environment may result in stagnation (Verhoeven, 1991a; Verhoeven & Boeschoten, 1986; De Ruiter, 1991; Pfaff, 1991; Van de Wetering, 1990). Due to restricted first language input, ethnic minority children may have a lower level of mother-tongue competence. However, there is ample evidence that children in a first or second language environment rely on highly comparable intralingual strategies in developing language competence. There was only minor evidence of interlingual patterns of language use (Boeschoten, 1990; Schaufeli, 1991). From a sociolinguistic point of view, the preferred language network in ethnic minority children's homes and in the wider community can be seen as a strong determinant of the

< previous page

page_11

next page >

< previous page

page_12

next page > Page 12

degree to which the children will develop the mother tongue (Verhoeven, 1991b). A central topic in research concerns shift and loss of immigrant languages. Shift refers to a functional redistribution of L1 and L2 in various domains of language use. Loss refers to a decrease in L1 proficiency. In contrast with language acquisition studies, the focus of studies on language shift and loss is mainly on the rate of change over time rather than on structural aspects (Extra, 1992). A substantial amount of research on language shift derives from periodically collected national population census data which include answers to one or more questions on language use (De Vries, 1989). Intergenerational patterns of language shift amongst immigrant groups have been studied to a large extent in English-dominant countries, such as the USA (Fishman et al., 1966; Veltman, 1983), Canada (DeVries & Vallee, 1980) and Australia (Clyne, 1991; Pauwels, 1988). As a general pattern, language shift seems to take place over four generations. The first generation (G1) is bilingual with a strong dominance of the mother tongue, G2 is bilingual either with a dominance of one of the two languages or with an equal command of both, G3 is bilingual with a dominance of the second language, and G4 only masters the majority language of the country of immigration (Saltarelli & Gonzo, 1977). In Europe there is no longstanding tradition of periodical collection and analysis of census data on home language use. A notable exception is Finland. In most countries no language-related issues were included in census questionnaires, except for the use of specific indigenous minority languages in certain regions, such as in Britain, France and Italy. In some countries (for example Belgium) language-related questions were even abolished for political reasons. On the basis of her study on language contact of immigrant minority groups in Sweden, Boyd (1985) found six major groups of factors determining the rate of shift. First of all, the process of shift is influenced by historical factors, such as the sociolinguistic background of minority groups in the home country, the relationship between the home country and the host country, or the social and linguistic situation of other minority groups which arrived earlier. Second, demographic factors such as group size, birth rate, marriage patterns, and length of immigration period turn out to be relevant. Third, geographic factors, like the distance from the home country, the concentration of settlement and the self-sufficiency of the group are relevant. Fourth, there are social factors related to the socio-economic status of the group. Fifth, institutional

< previous page

page_12

next page >

< previous page

page_13

next page > Page 13

factors related to, for example, culture, religion or mass media, come into play. The final category concerns attitudinal factors. Given the fact that these factors overlap and that they are not mutually independent, their relative importance is hard to demonstrate. With respect to structural aspects of first language loss, Van Els (1986) presented a typology and an overview of the research that has been done in Europe. From the studies carried out so far, it is clear that the structure of L1 loss is determined by both L1-related interlingual and L2-related intralingual principles. However, real-time studies on language loss are faced with the problem of defining full proficiency as the point of reference (Jaspaert et al., 1986). Finally, research on the emergence of new language varieties due to restricted language input and contact with other languages is just in a nascent state. This topic is highly interesting in that processes of synchronic language change can be compared with the outcome of earlier contact situations of immigrant languages with Indo-European languages (Johanson, 1990). Moreover, the features of new language varieties can be compared with the range of expressions children use in language acquisition. The present volume In the present volume recent developments in the study of immigrant languages in Europe will be presented. The book is divided into four parts. Part 1 gives a general overview of immigrant language varieties in Europe. The focus of Part 2 is on processes of first language acquisition in a second language environment. Part 3 deals with the study of code-switching and Part 4 with the study of language maintenance and language loss. In order to allow for crosslinguistic comparisons, different immigrant language varieties in various European countries are taken into account in all four Parts. The first chapter is by Jarmo Lainio. He gives an overview of studies on the language situation of Finnish immigrants in Sweden. Special emphasis is put on the FIDUS project, in which first language patterns of Finnish immigrants have been studied. In chapter 2, Safder Alladina puts South Asian language varieties in Great Britain in a historical and contemporary perspective. With respect to the present distribution and use of these language varieties, also the language maintenance efforts of South Asian communities and the

< previous page

page_13

next page >

< previous page

page_14

next page > Page 14

implications of the Education Reform Act of 1988 on the use and maintenance of South Asian languages in Britain are discussed. In chapter 3, Guus Extra and Ludo Verhoeven present recent research evidence on ethnic minority groups and languages in the Netherlands. The central focus is on the Turkish and Moroccan communities, who belong to the largest Mediterranean immigrant groups. After a discussion of the language background of these communities, the outline is given of a current research programme on processes of language change over time among Turks and Moroccans in the Netherlands. In chapter 4, Andrina Pavlinic describes the development of Serbian and Croatian language varieties in Western Europe. She focuses on the socio-demographic background and the self-reported first and second language proficiency of Serbian and Croatian children and adolescents in European immigrant countries. Moreover, she discusses the actual language proficiency of her informants on a multiple-choice grammar task and a composition task. Part 2 of the volume opens in chapter 5 with a contribution by Carol Pfaff on the acquisition of Turkish in Berlin. She describes the outcomes of two major studies: the KITA project and the EKMAUS project. Her main focus is on the functional development of conversational competence in Turkish by children in a bilingual context. The second study on acquisition is by Anneli Schaufeli who focuses in chapter 6 on first language development of 11-year-old Turkish children in the Netherlands. She discusses aspects of language proficiency at the lexical, morpho-syntactic and discourse level. Her main concern is the degree to which these different linguistic levels are affected by language erosion. She concludes that this is the case at least for the lexical and morpho-syntactic levels of the children's first language proficiency. The next study on acquisition is by Jeroen Aarssen, Jan Jaap de Ruiter en Ludo Verhoeven. In chapter 7, they make an attempt to assess the ethnic group language proficiency of Turkish and Moroccan children at the end of primary school in a diagnostic way. Apart from oral measures for vocabulary and listening comprehension, written measures on the level of graphemes, lexicon, syntax and text are distinguished. In order to test the construct validity of the measures taken, the interrelationships between language skills and background variables are discussed. A question that is highly relevant to the acquisition of ethnic community languages concerns parental attitudes towards bilingualism.

< previous page

page_14

next page >

< previous page

page_15

next page > Page 15

This question is addressed in chapter 8 by Sirkku Latomaa. She gives an empirical account of such attitudes among immigrant communities in the Nordic countries. Her conclusion is that parents generally express a strong desire for public support of their native languages. Part 3 of the volume deals with code-switching. In chapter 9, Lars Johanson opens this part by discussing basic principles of language change among Turks in Northern and Western Europe. He criticizes the basic concepts of traditional contact linguistics, such as 'borrowing', 'transfer' and 'substitution', and shows how the concept of copying may provide better insight into the processes characterizing interactions between languages in contact. Various types of copying processes are outlined on the basis of data on Turkish. In chapter 10 Ad Backus focuses on Turkish-Dutch code-switching among Turkish adolescents in the Netherlands. He makes an attempt to explain his empirical data in terms of both the sociolinguistic markedness model and the 'frame-process model', and concludes that the principles of the latter model can be violated in order to use code-switching as a marked choice. The next contribution on code-switching is by Jacomine Nortier. In chapter 11 she discusses the status of Dutch single word switches in Moroccan-Arabic language use of Moroccan adolescents living in the Netherlands. One of her questions is whether it is appropriate to consider single word switches as a separate category rather than as a particular type of intrasentential switching. Moreover, she questions whether these switches can be considered as borrowings. In chapter 12, Paula Andersson focuses on code-switching devices among Finns and Americans in Sweden. In her comparative study, she finds that Finns tend to integrate Swedish items, while Americans tend to use the Swedish items in a non-integrated way. She explains such differences in code-switching patterns in terms of the structure of the languages involved, the social context of immigration and the degree of linguistic neutrality. Part 4 of the volume deals with various aspects of language maintenance and language loss. It opens in chapter 13 with a contribution by Sally Boyd on immigrant minority languages in Sweden. She shows that despite the official policy of freedom of choice and the provision of home language instruction, language shift in the second generation is widespread. Changes in this pattern can be expected, however, as immigrants from more distant countries have recently arrived in larger numbers.

< previous page

page_15

next page >

< previous page

page_16

next page > Page 16

The subsequent contribution is by Donald Kenrick. In chapter 14, he gives an empirical account of the use and acquisition of Romani at a crucial stage in its development. He points out that Romani is undergoing change under the differential influence of various majority languages. Consequently, the mutual understanding of groups speaking different Romani varieties becomes harder. Conscious attempts to reverse this growing divergence are discussed. The two final chapters deal with language attrition. In chapter 15, Koen Jaspaert and Sjaak Kroon discuss methodological issues in language shift research. At the same time, they present data on a sociolinguistic research project on the social conditions that influence the processes of language shift among Turkish and Italian immigrants in the Netherlands and Flanders. The final contribution is by Kees de Bot and Bert Weltens. In chapter 16, they go into lexical aspects of language attrition and shift. After a summary of previous research carried out in this domain they present an outline of two research projects in which the emphasis is on lexical aspects of language attrition. References ALLADINA, S. & V. Edwards (eds) (1991), Multilingualism in the British Isles. Vol. 1: The older mother tongues and Europe. Vol. 2: Africa, Asia, and the Middle East. London: Longman. BACKUS, A. (1992), Turkish and Dutch in contact. Wiesbaden: Harrasowitz (to be published.) BOESCHOTEN, H. (1990), Acquisition of Turkish by immigrant children. A multiple case study of Turkish children in the Netherlands aged 4 to 6. Wiesbaden: Harrasowitz. BOESCHOTEN, H. & L. Verhoeven (1985), Integration niederländischen lexikalischer Elemente ins Türkische: Sprachmischung bei Immigranten der ersten und zweiten Generation. Linguistische Berichte, 98, 347-364. BOESCHOTEN, H. & L. Verhoeven (1987), Language mixing in children's speech. Language Learning, 37, 191-215. BOYD, S. (1985), Language survival. A study of language contact, language shift and language choice in Sweden. University of Gothenburg (diss.). CLYNE, M. (1982). Multilingual Australia. Melbourne: River Seine Publ. CLYNE, M. (1991). Community languages. The Australian experience. Cambridge: University Press. COULMAS, F. (ed) (1991), A language policy for the European Community. Berlin: Mouton de Gruyter.

< previous page

page_16

next page >

< previous page

page_17

next page > Page 17

DE RUITER, J. (1989), Young Moroccans in the Netherlands. An integral approach to their language situation and acquisition of Dutch. University of Utrecht (diss.) DE RUITER, J. (1991), Talen in Nederland. Een beschrijving van de taalsituatie van negen etnische groepen. Groningen: Wolters-Noordhoff. DE VRIES, J. (1989), On coming to our census. A layman's guide to demolinguistics. In: D. Gorter et al (eds), Fourth International Conference on Minority Languages, Vol. 1, 57-76. Clevedon: Multilingual Matters. DE VRIES, J. & F. Vallee (1980), Language use in Canada. Ottawa: Canadian Statistics. ENGMAN, M. (ed) (1992), Ethnic identity in urban Europe. Comparative studies on governments and non-dominant ethnic groups in Europe, 1850-1940. European Science Foundation. Hants/England: Dartmouth Publishers. EUROSTAT (1991), Demographic statistics on the European Community. Luxemburg: Statistical Office of the EC. EXTRA, G. (1990), Ethnic minorities, language diversity and educational implications. A case study on the Netherlands. Balance and Perspective: 25 years of Dutch Applied Linguistics. Toegepaste Taalwetenschap in Artikelen 36 (1), 45-71. EXTRA, G. (1992), Language acquisition, shift, and loss of immigrant minority groups in Europe. In: B. Kettemann & W. Wieden (eds), Proceedings of the Inaugural Conference of the European Second Language Association. Tübingen: Narr (to be published). EXTRA, G. & T. Vallen (1989), Second language acquisition in elementary school. A crossnational perspective on the Netherlands, Flanders and Germany. In: L. Eldering & J. Kloprogge (eds), Different cultures, same school. Ethnic minority children in Europe. Amsterdam: Swets & Zeitlinger, 153-188. EXTRA, G. & L. Verhoeven (eds) (1992), Ethnic community languages in the Netherlands. Amsterdam: Swets & Zeitlinger (to be published). FANTINI, A. (1985), Language acquisition of a bilingual child. Clevedon: Multilingual Matters. FISHMAN, J. et al (1966), Language loyalty in the United States. The Hague: Mouton. GORTER, D. et al (eds) (1989), Fourth International Conference on Minority Languages, Vols. 1 and 2. Clevedon: Multilingual Matters. GROSJEAN, F. (1986), Life with two languages. Harvard: University Press. JASPAERT, K. & S. Kroon (eds) (1991), Ethnic minority languages and education. Amsterdam: Swets & Zeitlinger. JASPAERT, K., S. Kroon & R. van Hout (1986), Points of reference in first-language loss research. In: B. Weltens et al (eds), Language attrition in progress. Dordrecht: Foris, 37-49. JOHANSON, L. (1990), Studien zur turkeitürkischen Grammatik. In: G. Hazai (ed), Die türkische Sprachwissenschaft der Gegenwart. Budapest: Akademia Kiado. KROON, S., M. Kruger-Potratz, U. Neumann & T. Vallen (eds) (1991), Kultur- und Sprachenvielfalt in Europa. Münster/New York: Waxmann. LAINIO, J. (1987), Language use of Finns in Sweden. Implicational patterns in four domains. Scandinavian Working Papers on Bilingualism 7, 11-28. LINGUISTIC MINORITIES PROJECT (1985), The other languages of England. London: Routledge & Kegan Paul. MUUS, P. (1991), Migration, minorities and policy in the Netherlands. Recent trends and developments. Report for the OECD. University of Amsterdam: Dept. of Human Geography.

< previous page

page_17

next page >

< previous page

page_18

next page > Page 18

NORDIC STATISTICAL SECRETARIAT (1991), Yearbook of Nordic Statistics, Vol. 29. NORTIER, J. (1989), Dutch and Moroccan Arabic in contact. Code switching among Moroccans in the Netherlands. Dordrecht: Foris. PAUWELS, A. (ed) (1988), The future of ethnic languages in Australia. International Journal of the Sociology of Language 72 (special issue). PAVLINIC, A., K. Brcic & N. Jeftic (1988), Supplementary mother-tongue education and the linguistic development of Yugoslav children in Denmark. Journal of Multilingual and Multicultural Development 9 (1-2), 151-167. PFAFF, C. (1991), Turkish in contact with German. Language maintenance and loss among immigrant children in West Berlin. International Journal of the Sociology of Language (to appear). 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. REICH, H. (1991), Developments in ethnic minority language teaching within the European Community. In: K. Jaspaert & S. Kroon (eds.), Ethnic minority languages and education. Amsterdam: Swets & Zeitlinger, 161-174. SALTARELLI, M. & S. Gonzo (1977), Migrant languages: linguistic change in progress. In: H. Giles & B. Saint-Jacques (eds), Language and ethnic relations. Oxford: Pergamon Press. SCHAUFELI, A. (1991), Turkish in an immigrant setting. A comparative study of the first language of monolingual and bilingual Turkish children. University of Amsterdam (diss.) SMOLICZ, J. (1980), Language as a core value of culture. Journal of Applied Linguistics 11 (1), 1-13. TOSI, A. (1984), Immigration and bilingual education. Oxford: Pergamon Press. VAN DE WETERING, S. (1990), Onderwijs in eigen taal en cultuur aan Marokkaanse kinderen in Nederland. Utrecht University (doct. diss.). VAN ELS, T. (1986), An overview of European research on language attrition. In: B. Weltens et al. (eds), Language attrition in progress. Dordrecht: Foris, 3-18. VELTMAN, C. (1983), Language shift in the United States. Berlin: Mouton. VERHOEVEN, L. (1991a), Acquisition of Turkish in a monolingual and bilingual setting. In: H. Boeschoten & L. Verhoeven (eds), Turkish linguistics today. Leiden: Brill, 113-149. VERHOEVEN, L. (1991b), Predicting minority children's bilingual proficiency: child, family and institutional factors. Language Learning 41, 2, 205-233. VERHOEVEN, L. & H. Boeschoten (1986), First language acquisition in a second language environment. Journal of Applied Psycholinguistics, 7, 241-256. VERMES, G. (ed) (1988), Vingt-cinq communautés linguistiques de la France: les langues immigrées. Paris: Editions l'Harmattan. WILPERT, C. (1991), Demographische Entwicklung und Migration zwischen den Mitgliedstaaten der europaischen Gemeinschaft und in sie hinein. In: S. Kroon, M. Kruger-Potratz, U. Neumann & T. Vallen (eds), Kultur- und Sprachenvielfalt in Europa. Münster/New York: Waxmann, 20-37.

< previous page

page_18

next page >

< previous page

page_19

next page > Page 19

PART 1 USE OF IMMIGRANT LANGUAGE VARIETIES IN EUROPE

< previous page

page_19

next page >

< previous page

page_ii

next page >

< previous page

page_21

next page > Page 21

1 Sweden Finnish Jarmo Lainio It has become acknowledged that the use of two or more languages by the same individual is a normal and globally unmarked situation. Recent immigration has increased the number of bilingual speakers in various parts of Europe. This has also been the case in Sweden, where Finns constitute by far the largest immigrant group. Finnish as a minority language in Sweden was in some respects one of the original case studies of linguistic assimilation among minorities in Europe, following the sociopolitical changes that took place during the 1960's and 1970's. The liberal views expressed by the Swedish governments, summarized in the idealistic goals of 'equality, freedom of choice and partnership', supported the study of immigrants and their linguistic and cultural integration, and made way for an early (1976) version of a state level policy regarding the basis of 'active bilingualism'. Practice has, nevertheless, largely been one of implicit assimilation pressure for the second and later generations. In Sweden, the concept of 'semilingualism' was adopted early in descriptions of the Finnish-speaking second generation's difficulties and efforts to deal with the bilingual minority situation. It was also believed by many to be the ultimate outcome of sending a child to a school which allowed teaching in two languages. Now, while the concept has been abandoned in Sweden, it seems to have spread to other language contact contexts (Baetens Beardsmore, 1986; Romaine, 1989). A balance between the first and second language was difficult to attain both for Sweden Finnish children and adults in every-day life, and as it seems, also within the field of research. In addition to this imbalance at language level, the use of non-standard varieties has had an impact on the views of language users and researchers. It has even been stated by adult almost monolingual speakers of Finnish in Sweden that they considered themselves 'semilingual', since they did not use "proper language", the standard. Comparatively little attention has been paid to the first language (L1) and its varieties, for several theoretical and practical

< previous page

page_21

next page >

< previous page

page_22

next page > Page 22

reasons. With the growth of sociolinguistic studies, their importance has been increasingly acknowledged (cf. Preston, 1989). In the present study the first language varieties of Finnish in Sweden will be dealt with. First of all, the sociolinguistic background of Finnish-speaking populations in Sweden will be discussed. Some results of studies conducted so far on Finnish varieties used in Sweden will be discussed, especially those of the completed FIDUS-project. Finnish-speaking populations in Sweden The predecessors of post-war Sweden Finns Finnish-speaking people have resided within the borders of Sweden for about 700 years. The Finnish mainland constituted one of the easternmost parts of Sweden for 600 years, since Swedes had received papal blessings to finally impose a 'proper Christianity' on the Finns in the 12th century, and to compete with the orthodox, eastern influence. Other more profane factors have at various times exerted temptations on Finns to move to the Swedish mainland. In 1809, the Swedish kingdom had to cede the Finnish part to the Russian empire. One result of this forced peace treaty was that the border in the north divided a culturally and linguistically fairly homogeneous area (with Finnish and Sami spoken, both Finno-ugric languages). Finnish-speaking inhabitants were thus to be found on both sides of the SwedishFinnish/Russian border. The population remaining within the borders of Sweden, frequently referred to as the Tornedalen Finns, has made up a shrinking linguistic minority since that time. A promotion of Swedish among the Tornedalen Finnish (=ToFi) population was anticipated as necessary by the state and was officially sanctioned from the 1880's and onward. The number of speakers with an at least receptive command of ToFi today is estimated at 20,000 (Jaakkola, 1973; Wande, 1982, 1984). Measures have been taken to postpone a total language shift, but the number of speakers is decreasing in most former ToFi communities. On the one hand, postwar immigration from Finland to these north Swedish communities has meant a support for Finnish language use, but on the other hand, the variety of Finnish that might survive is not necessarily the old ToFi, but rather the Finnish spoken by the immigrants, with its closer connection to stan-

< previous page

page_22

next page >

< previous page

page_23

next page > Page 23

dard Finnish. Attempts have been made lately to stabilize the ToFi dialects in writing. Since the 13th century there has been a small but steady migration between the central and eastern coastal regions of Sweden and the western parts of Finland. The Baltic sea has connected the areas west and east of the sea to the capital area. For example, the Finnish parish in Stockholm celebrated its 500th anniversary during the 1980's. In the continuous flow of Finns to Sweden, the Swedish-speaking people regularly have made up a relatively large proportion. The best-known larger settlements, however, were the Finnish-speaking ones in the western and central forest-regions. Initially they consisted of burnbeating Savonians, who were asked to cultivate Swedish forest land. A large proportion of these Savonians ended up in the province of Värmland, which is why these so-called 'forest Finns' have been equalled to Värmland Finns. In the 1960's, after more than 400 years, single descendants of this linguistic minority could still be found, who knew and had spoken Finnish in their homes as children. The last speaker of Värmland Finnish died in the 1970's. Both the Tornedalen Finns and the Värmland Finns have been the objects of historical, ethnographic and dialectological studies for more than one hundred years. Data have been collected on the Värmland Finns since the early 1800's and data on the Tornedalen Finns from at least the 1880's (Wande, 1982; Virtaranta, 1986; Winsa, 1991; cf. the review in Lainio, 1989: 30-32). Socio-demographic characteristics of Sweden Finns Since the Second World War, about 450-500,000 Finns have migrated to Sweden, about one third of which have returned to Finland. Today, second and third generation Sweden Finns, i.e., of Finnish ethnic origin, are growing up in Sweden. Allardt & Starck (1981) estimate that there are about 300,000 Sweden Finns. An estimated 50-60,000 of these are Swedish-speaking Finns, i.e., former minority speakers of Swedish in Finland. The number of Sweden Finns is believed to be decreasing continuously (Reinans, 1988). A main reason is the migration pattern between Sweden and Finland during the last decade, which has been one of net re-migration to Finland, largely due to years of economic prosperity in Finland, versus economic recession in Sweden, especially in 1980-83.

< previous page

page_23

next page >

< previous page

page_24

next page > Page 24

Already in the 1970's the proportion of Sweden Finns living in urban municipalities was above 90 % (Majava, 1981). There is no reason to believe that this proportion has decreased. One third of all Sweden Finns are estimated to live in the capital Stockholm region. The socioeconomic (=SEC) characteristics of Sweden Finns have generally been those of a working-class, both from statistical points of view and as a stereotype picture of Finns. These characteristics have changed during the last decade, especially regarding the second generation. A shift from production to service occupations has also taken place among first generation immigrants, especially among women (SCB, 1984). The SEC traits of the Sweden Finns are no longer as uniformly 'working class' as during the first decades of immigration. The ethnic revival ideologies that swept through other European countries also concerned Sweden (cf. Allardt & Starck, 1981). The school and language debate has had a key position among Sweden Finns from the 1970's and onward. From 1976, when the home-language legislation was put through, there has also been a constant need for teachers and other educational staff from Finland, until the Swedish educational system had a chance to catch up with the situation and 'native' Sweden Finnish teachers had been trained. Even if the legislation of 1983 (SOU, 1983), weakened the position of minority languages in the school system, the formal training in Finnish for pupils had already had a chance to have some sociolinguistic impactit had increased the knowledge of the standard variety among second-generation speakers. The social participation of teachers has further contributed to a more heterogeneous sociolinguistic map of Sweden Finnsmany of them have been involved in Sweden Finnish organizations and in language policy matters. This social differentiation has included other occupational groups as well. During the late 1980's, multinational Finnish companies established themselves on the Swedish market, by buying Swedish companies. A new category of Finns in Sweden emerged, that of the upper-class managerial staff. This has revised the picture of Finns in general among Swedes and may have had some positive effects for the self-image of Sweden Finns and their attitudes towards Finnish. Concerning the language use patterns among Finnish-speaking people in Sweden, this has probably only had minor effects thus far. If the concept 'Sweden Finn' is used to cover not only immigrant labourers from Finland and their descendants, the SEC characteristics of Sweden Finns are quite different from the stereotypical picture, just as is the language use. This is complicated by ethnic identification,

< previous page

page_24

next page >

< previous page

page_25

next page > Page 25

however, since e.g. some Finnish 'middle' and 'upper-class' immigrants who have come to Sweden, not as labourers, but in order to acquire useful experience abroad, do not identify with Sweden Finns, presumably because of the negative connotations the concept has had. However, this reason for emigrating is frequently given by so-called working-class Sweden Finns as well. The non-identification of minority speakers with the minority has for example been taken as a criterion for their exclusion in some other language contact studies (e.g., Poplack, 1989). Typically, Sweden Finns have earlier had an all-Finnish ethnic background, with both parents originating from Finland. Sweden Finns today make up a continuum, with highly different social and ethnic characteristics. If we find at one end, those who identify as Finland Finns, we have at the other end those who have assimilated in one way or other with the Swedish society, by identification, marriage and mixed family background. This concerns to a larger extent the second and later generations. For studies of the first generation the degree of identification with one group or the other has been observed to overlap with ability to speak Swedish (e.g., Jaakkola, 1983). The question arises whether the Finnish of those who have almost fulfilled a languageshift should be studied on a par with those who have not. This is a methodological difficulty in language contact studies. These questions have not been discussed at any length for Sweden Finnish. Another problem, little discussed so far, is the fact that many second-generation immigrants of (partial) Finnish ethnic origin, even among those who participate in Finnish home language instruction, are trilingual (cf. Natchev & Sirén, 1988). It seems to be the case that Finnish often becomes the stronger language for children in these multilingual families. This is usual in families where the mother is Finnish and the father of ethnic background other than Swedish or Finnish, a frequent pattern during the 1980's in, e.g., the capital area (Natchev & Sirén, 1989). Other minority languages may not have the same possibilities as Finnish regarding instruction in school, due to the smaller number of speakers and availability of teachers etc. Furthermore, one cannot, according to the Swedish school legislation (regarding instruction), have two mother tongues other than Swedish used in the home. One of the minority languages thus tends to become a merely spoken language, not formally taught in school. In the capital Stockholm region and in the two other larger urban areas of Sweden, Gothenburg and Malmö, interethnic relationships are

< previous page

page_25

next page >

< previous page

page_26

next page > Page 26

more common than in smaller municipalities. Finnish-ethnic women seem to marry interethnically to a larger extent than Finnish men (cf. Natchev & Sirén, 1989). In the mentioned regions the proportion of single mothers has been noted to grow significantly, which however also was a characteristic among Swedish women. These matters bear more on the question of 'who is a Sweden Finn' now than they did only ten years ago. Still, the majority of Finnish-origin children either belong to all-Finnish families or to Finnish and Swedish families. In studies concerning the situation 10-15 years ago, Finns were typically an ethnically (though not linguistically) homogeneous population (e.g., Lainio, 1989, cf. below). Studies covering the situation today have a more complex reality to account for (Huss, 1991; Sirén, 1991). The second generation's future perspectives in earlier comparisons were generally (when, e.g., proportions of pupils continuing to secondary education was compared for Finnish background children to those of majority background), that the Finnish children did not indicate high social mobility in relation to their parents' generation. The picture of future secondary education of pupils of Finnish background has also begun to change (e.g., Olkiewicz, 1990). In several aspects, then, Sweden Finns are becoming more like the majority population. Patterns of language use The use of Finnish among the adult first generation speakers has had a strong position. It has been the exclusive language used in the homes and in other private domains, according to self-reports (cf. Jaakkola, 1983; Boyd, 1985; Lainio, 1984, 1987). The second generation that grew up during the 1960's, before the largest waves of immigration had arrived, did not have favourable opportunities to receive Finnish instruction. They may be expected to have shifted to Swedish to a larger extent than may be assumed to be the case among later born second-generation Finns (Lainio, 1987). Later, when there was a choice between Finnish and Swedish, a strong polarization took place among the adult generation regarding the choice of linguistic and cultural identification. This probably reflects the simultaneous polemic situation in earlier researcheither you remain Finnish or you don't (cf. Skutnabb-Kangas, 1981, 1987; Jaakkola, 1983; Savolainen, 1987). Predictions of the Sweden Finnish speech community have indicated a language shift in later generations (Boyd, 1985), thus following the fate of for example the Finnish emigrants and their descendants in

< previous page

page_26

next page >

< previous page

page_27

next page > Page 27

the US and in Canada (Virtaranta, 1988; cf. Edwards, 1985). Other similarly hypothetical developments, though not in full contradiction with these views, are that the number of speakers using Finnish in Sweden may well be decreasing, but similarities with the Finnish spoken in Finland may remain fairly constant or even increase (Wande, 1989). Roughly one third of the Finnish-origin pupils in the Swedish school system do not participate in any tuition of Finnish, which may be assumed to make them, or at the latest their children's generation, participate in a shift to Swedish. The other two-thirds constitute two main categories of pupils, who have had Finnish in their curriculum (Hammar, 1985). One category has followed hour-based tuition in Finnish (home-language instruction), the other has had most instruction in Finnish classes (all-Finnish classes initially, later Swedish share is increased). Students in the latter group would thus largely retain some of the impact of Finland Finnish varieties, e.g., standard Finnish, while at the same time becoming functionally bilingual. The outcomes of the Swedish homelanguage legislation is only by now becoming apparent for research: the first pupils who have participated in the various homelanguage programs completed their secondary education during the late 1980's. Another future scenario is that Finnish immigrants who are concerned with their children having a chance to learn Finnish in the Swedish school, will tend to move to municipalities, where there is an existing large population of Sweden Finns, and where the Finnish language has a more favourable situation in school. Some municipalities may therefore face a faster language shift than others, as a result of local differences. It is not known how widespread this kind of migration is, but many families tend to postpone their final decision on whether they will stay or not in Sweden, to the point when the children are to start their first grade in school. Internal migration in Sweden has become an option just as re-migration to Finland. Studies on immigrant Finnish in Sweden If language in a broad sense is considered here, the earliest studies on Sweden Finns can be said to have been done in the early 1970's, mainly by social scientists. Some linguistic observations were published as well (e.g., Hormia, 1976). These were preceded by Hansegård's (1968) observations of Tornedalen Finns, that they were "semilingual" in

< previous page

page_27

next page >

< previous page

page_28

next page > Page 28

several (psycholinguistic) respects, which was followed up by psychological, educational and social-psychological test studies and surveys (Toukomaa & Skutnabb-Kangas, 1977; Ekstrand, 1978; Rönmark & Wikström, 1980; Ekstrand et al., 1981; Toukomaa, 1983; cf. Hyltenstam & Arnberg, 1988), as well as some linguistic studies (Loman, 1974). Many of the studies concerned the linguistic competence of Sweden Finnish school-children (not the Tornedalen Finns). Naturalistic language data, outside test situations, were seldom considered in these studies (however, Loman, 1974 for Tornedalen). Adult Sweden Finnish speakers were studied later, mostly in language survey studies based on sociological methodology. The question of identity and language recurred in studies on first generation adults (e.g., Jaakkola, 1983; Hujanen, 1986). This direction of research was continued for second-generation adolescents as well (Ouvinen-Birgerstam, 1987; Liebkind, 1989; Olkiewicz, 1990). Other aspects, like the study of discourse strategies, have been included later (e.g., Skutnabb-Kangas, 1987; cf. Strömqvist, 1989). Jaakkola (1983) used a social network method to study language use among adults. One large-scale survey (Boyd, 1985) on adolescents' preference and some small-scale case studies of language use (e.g., Knookala, 1982; Lainio, 1984, 1987) were conducted. Adults' languages were only sporadically included in linguistic studies, and almost exclusively in L2 acquisition studies (Hyltenstam, 1978; Hyltenstam & Magnusson, 1982; cf. Hyltenstam, 1989). Most linguistic studies on children and adolescents were carried out within this framework as well. A polarization took place from the mid-1970's and onward between researchers who in addition to making research on minority matters, also took an open stand for or against the assimilation of the minorities. When the government decided to evaluate the mother tongue issue in an effort to dictate a future general policy, it was agreed that a neutral observer should be used, but since most debates in Sweden were tied up by earlier polemics, it was decided that a scholar with cultural and linguistic knowledge of the situation in Sweden, but from a third country (i.e., neither Sweden nor Finland), be consulted. This evaluation (Bratt Paulston, 1982) was, not unexpectedly, criticized from both sides as being shallow and simplified. The semilingualism debate, which turned out to be an unsolvable polemic discussion implicitly dealing with the question of whether immigrants and minorities should or should not become assimilated Swedes, and on whose terms this should take place, was understood by

< previous page

page_28

next page >

< previous page

page_29

next page > Page 29

most researchers to best be abandoned for more positive exchanges of ideas. The concept has also been seen as a pretheoretical one for communicative competence (e.g., Martin-Jones & Romaine, 1987). However, Hansegård, who initiated the debate with his book (1968), felt, presumably rightly, that his idea was misinterpreted and misunderstood by most combatants. He therefore returned to the topic and tried to clarify his standpoint (Hansegård, 1990). The factual confusion and the individuals' problems resulting from (socially) specific kinds of language contacts, have been discussed elsewhere in other recent overviews (e.g., Baetens Beardsmore, 1986; Appel & Muysken, 1987: 107-8; Romaine, 1989: 232-44). Many of them rely on or react against Skutnabb-Kangas (1984). In addition to the literature referred to above (cf. also Wande, forthc., for a review), some later works can be mentioned. Nesser (1990) has described, on the basis of the written composition material by school-children (cf. her thesis, Nesser, 1986), their use of the possessive suffix. The same linguistic feature has been studied by Kalin (in Boyd & Andersson, 1991). Lainio has followed up his thesis (1989) with two studies on the spoken language material of the FIDUS project, consisting of recordings of 54 first-generation immigrants from four Finnish dialect regions. One paper (Lainio, 1990) is an extended study of five female immigrants in one of the dialect groups studied earlier, who showed clearly diverging patterns regarding their retention of dialect features. Another is an extended study of a syntactic feature, the use of verb forms and accusative objects of the first person plural (Lainio, 1991a). Lainio & Wande (1991) have quantitatively studied the use of the 1st person singular pronoun in the same (FIDUS) corpus, incorporating prosodical and pragmatic factors into the analysis. In the project ''Variation and change in linguistic contact" (Boyd & Andersson, 1991), the types of code-switching that occur among Sweden Finns (and American-English immigrants in Sweden) have been studied, as well as the integration of Swedish lexical items into Finnish. The project also addresses the question of language death, or language attrition, as a characteristic of language shift. Andersson (forthc.) has studied the lexical changes and integrations more in detail, based on the same material. Boyd (forthc.) has been concerned with code-switching typologies developed during the last decades, and their application to their Finnish and American-English material (cf. Boyd, Andersson & Thornell, 1991).

< previous page

page_29

next page >

< previous page

page_30

next page > Page 30

The FIDUS project The study conducted by Lainio was one of about ten different projects included in a Nordic joint research programme, Urbanisation and language change in the Nordic countries, lead by Bengt Nordberg from the sociolinguistic research unit of the department of Nordic languages in Uppsala. The different, fairly independent projects dealt with migrants', immigrants' and stable populations' apparent-time linguistic adaptations to more urbanized living conditions (see Lainio, 1989: 395-98, for references on published works). Lainio's project, FIDUS (short for Swedish 'Finnish dialectal development in a Swedish industrial town'), was aimed at establishing a base description of spoken Finnish of adult first-generation immigrants. It had as its aim to try to test some hypotheses on what varieties of Finnish would influence other Finnish varieties. One hypothesis concerned whether or not the sheer numeric majority of speakers originating from the rural, eastern and northern parts of Finland, with their specific dialectal backgrounds, would direct the whole speech community away from the development of spoken Finnish in Finland. Since any geographical dialect and most sociolects represented in Finland may be represented within one municipality in Sweden, the outcome of these multifarious language contacts is not predictable and its characteristics are little known. It was further believed to be necessary to establish a systematic and somewhat representative linguistic description based on naturalistic speech data, either to confirm the views expressed by laymen and some scholars of language that 'Sweden Finnish', as it has become known, was a chaotic, deteriorated, mixed and/or vulgar mode of language, or conversely, that this was not the case. In retrospect, on the basis of the results of sociolinguistic research of the 1970's and 1980's, it is easy to tell that the first view was not adequate from a linguistic point of view. That both laymen and linguists could hold such views, can partly be attributed to the fact that linguists as well as laymen are victims of surrounding mainstream attitudes and categorical perception, and partly to the fact that linguistics has suffered from the long-term impact of the bias of written language imposed on research methodologies and paradigms (Linell, 1982; Rommetveit, 1983). The results from the project were also hoped to offer basic L1 descriptions for later studies dealing with Finnish in Sweden.

< previous page

page_30

next page >

< previous page

page_31

next page > Page 31

Data collection The 54 speakers studied could be counted as Sweden Finns by time of residence (5-35 years in Sweden), and as first-generation immigrants by age of immigration (at least 10 years old at the time of immigration). They mostly held typical working class occupations and/or could, by level of education, be counted as belonging to this social stratum. They were endogamic, if married at all. The age groups, which both included only first-generation immigrants, were 20-30 and 40-60 years at the time of the study. Four main dialect regions, represented by major dialects, were chosen as the areas of origin for the informants (see Fig. 1), namely the southwest Finnish, the central Häme, the north Finnish (Kemi-Rovaniemi areas), and the North Karelian dialects. The informants were supposed to have been born within one of the four dialect regions and not have moved out of this region before the age of 10.

Figure 1 The four dialect areas and the Swedish municipality studied in the FIDUS study.

< previous page

page_31

next page >

< previous page

page_32

next page > Page 32

These regions represent both demographically, regarding the origin of Finnish immigrants, and linguistically, regarding the division into main dialect areas, optimal samples for a quantitative study of Sweden Finns. A majority of Finnish immigrants have originated from the northern and eastern rural parts of Finland, whereas the highest prestige has generally been attributed to varieties of the southern, central and southwestern parts, especially the urban centre of the capital Helsinki. From these areas it was also possible to find some recent quantitative studies indicating what the language as spoken today is like, in comparison to older dialect descriptions. The study was conducted in the city of Eskilstuna, Sweden, which has a high concentration of Finnish immigrants. About ten per cent of the 90,000 inhabitants in the municipality are estimated to have Finnish ethnic background. Eskilstuna is situated about 110 km from Stockholm and 220 km from two other main ferry harbours, with frequent connections to Finland. With a long tradition of metal industry production, Eskilstuna was one of the typical targets for Finnish immigrants, especially those of rural background, during the 1950's to the 1970's. There is also a fairly large regional hospital, which during the later decades has offered work for women. Eskilstuna was one of the first municipalities which independently experimented in order to solve language problems among Finnish pupils in school, by offering them a kind of compensatory teaching in Finnish. Though this partly emanated from the problems teachers faced, Eskilstuna has been one of the cities which have tried to follow government recommendations regarding home language instruction. This has, as in most Swedish municipalities, largely been the result of the parents' struggle to guarantee some formal teaching in Finnish for their children. All-Finnish classes as well as secondary level Finnish classes and home language instruction have been offered Finnish-speaking pupils from the early 1980's. A Finnish infrastructure has developed in Eskilstuna, with shops, minor business, organizations and leisure time activities. The two local newspapers regularly have had Finnish pages in their Swedish editions. The informants of the FIDUS study were recorded in their homes in informal, self-chosen group discussions among friends (30120 minutes, average 60 minutes). During the recordings, the participants started off discussing pre-chosen topics, which were typically abandoned and turned into free conversations. They had before the meeting answered a questionnaire, for the purpose of a basic description of the language use habits in the speech community (Lainio, 1984). The

< previous page

page_32

next page >

< previous page

page_33

next page > Page 33

speech data can be seen to reflect the vernacular of the informants, although some stylistic variation does occur (for more details of the study, see Lainio, 1989, 1991b). The creation of Standard Finnish For any understanding of what is going on in the Sweden Finnish variety contact situation, it is necessary to recall how Finnish in Finland has been developed and evaluated from about the 1820's. The historical background of the linguistic variables studied and the present-day distribution of their variants are, e.g., for (d), quite complex (for closer descriptions, see Lainio, 1989, 1991b.) One reason is that the west Finnish and East Finnish dialects are not on an identical 'distance' from the Germanic languages from a typological point of view, in addition to other dialectal variation. Furthermore, instead of being created by one variety, which would have been raised to the privileged position of a standard variety, StFi was formed by conscious compromises. Finnish writing was, from its beginning in the 16th century, based on west Fi, mainly southwest Fi dialects, which were closer to Swedish on all linguistic levels than were for example the east Fi Savo dialects. A continuous adoption of east Fi features took place during the following three centuries, and during the romantic and nationalistic movements of the 19th century these discussions developed into a systematic re-codification of the written language. During this so-called 'battle of dialects' the principles of future 'purification' of Finnish were set. Many strictly west Fi and southwest Fi features were replaced by forms that were considered 'purer', more Finnish, i.e., east Fi. At all linguistic levels, from phonology, over morphology and syntax, to lexicon, the language was restructured. Non-existent but structurally seen plausible innovations were also accepted. One result of this language planning is that speakers of all dialects have features in common with StFi. Which these features are depends on which dialect region the speaker is from. A firm prescriptive attitude, which connected language with political ideologies in a wider sense, implemented the linguistic innovations made by linguists, writers and laymen. This caused an extensive change of the linguistic usage among Finns. Since Finns to a very high extent have become literate already from the end of the last century, the impact has been even greater. These issues were connected in complex ways to the question of linguistic equality between Swedish, thitherto the dominant language, and Finnish, the language of the majority although dominated by

< previous page

page_33

next page >

< previous page

page_34

next page > Page 34

Swedish (and partly Russian during the 19th century). The first monolingual Finnish-speaking intellectual generationsintellectual in the sense that they represented an educated middle-class and not a rural population or an uneducated working classwhich had the ability and need to promote the linguistic ideals and socioeconomic characteristics of a middleclass, evolved during the first decades of the 20th century. Sociolinguistic variability for Finnish had before this period mainly been connected to areal differences, mainly between rural dialects. The knowledge of what the speech of Finns living in towns was and is like, is limited. The social variation as expressed by linguistic variants correlating with urban socioeconomic background has thus been vague (Mäkelä, 1985). It has not been associated with social class directly, but rather with the factor of education, which is more accessible in Finland than in many other countries. The possibility to exert power by the use of (standard) Finnish as an instrument for this, then, has been possible for less than 80-90 years. In present-day Finland there is reason to expect that the sociolinguistic stratification is progressing (cf. Horvath, 1985: 4-5, for a similar development). The characteristics of this development still have a bearing on the generations which grew up until the 1950's. Most emigrants to the US and Canada, and most first-generation immigrants to Sweden have been heavily influenced by these ideas. This is also one explanation as to why traditional dialectology has played an important role in the study of spoken emigrant Finnish. Though this by necessity has meant that some aspects of language have been less studied than others, a rich material of spoken language has been collected. This did not concern Sweden Finnish however, since the speakers did not fit the informant criteria of this paradigm. What is somewhat ironic, is that this knowledge based on features in Finnish dialects could, to some extent, have been adapted to clarify some misconceptions in the evaluations of linguistic competence testing among school-children during the 1970's and 1980's. What was considered aberrant from the standard was not necessarily the result of 'semilingualism' or interference, it was at least partly plain, spoken language. Data analysis The seven variables studied in Lainio (1989) included phonological, morphological and morphosyntactic features of Finnish, selected for their ability to make comparisons possible between Sweden Finnish and spoken Finland Finnish. The variables were chosen in ways as to ex-

< previous page

page_34

next page >

< previous page

page_35

next page > Page 35

clude obvious impact of Swedish. The variables have shown sociolinguistic variation in Finland and constitute features that have different variants in the west Fi and east Fi dialects. The variables studied were the following (details can be found in Lainio, 1989, 1991b): 1) (d)-variable: In writing occurs; pronounced as [d] in StFi, a voiced alveo-dental stop, similar to the English pronunciation of /d/. The StFi /d/ is a result of gradation, a main linguistic feature of Finnish, where /d/ is the 'weak grade' variant of 'strong grade' /t/ (cf. Karlsson, 1983). Present-day loan words are slowly, if at all, integrated into the gradational system. Three linguistic contexts are considered: VdV, VVdV and /hd/. The Finnish /d/ is a shibboleth phenomenon: its use is largely socially, situationally and geographically determined. It is the most frequently studied isoglossic feature of Finnish dialectology. The variant d has traditionally occurred in formal standard language and in some dialects in contact with Swedish (in Table 1 the basic variants for each dialect are given; all variants do not occur in all linguistic contexts mentioned). The north Fi dialects studied have a mixed historical background, which has made both west Fi variants for some variables, and east Fi variants for other variables, like (d), become the native dialect variants. Example words, for VdV-: StFi: katu 'street' : kadulla 'on the street', west Fi karulla etc., east Fi kavulla; for /hd/: StFi lehti 'newspaper' : lehden 'the newspaper's', west Fi lehren, east Fi lehØen (Ø = zero). 2) (ts)-variable: occurs in writing and StFi. This variant is only considered to be native in the southeasternmost Finnish dialects (not included in the study). Historically, the phonological positions of /ts/ have followed the gradational principle. In the dialects this is still reflected in the shift between a strong grade and a weak grade variant, according to their respective phonological position. In StFi thus, e.g., metsä : metsässä 'forest, in the forest', which in the dialects may have retained a shift between grade variants: mettä : metässä, like in some west Fi dialects (gradation /tt/ : /t/), or, mehtä : metässä, like in east Fi dialects (cf. Table 1). The (d)- and (ts)-variables have been introduced into formal spoken Finnish, mostly due to a strong orthoepic pressure from the 19th century and onward. Part of their origin is due to the influence of Swedish writing. 3) (schwa)-variable: The insertion/epenthesis of vowels is a widespread phenomenon in the east Fi and north Fi dialects (in Fennistics this is called a schwa-vowel). The epenthesis has developed in heterorganic C clusters like /1/+/k/, /l/+/m/, /l/+/p/; /r/+/j/; /n/+/h; /h/+/m/, and in some

< previous page

page_35

next page >

< previous page

page_36

next page > Page 36

additional contexts. Recently it has diffused as a mainly male dialect feature in Finland. Examples: StFi jalka 'foot', jalaka, StFi kylmä 'cold', kylymä. 4) (diphth)-variable: The lowering of diphthongs has been going on for several centuries in west Fi dialects. Three of the Finnish diphthongs have participated in the process: /uo/ >> /ua/, /ie/ >> /ia/, or /iä/ (depending on vowel harmony), and /yö/ >> /yä/. Examples: StFi juo 'drink', jua, StFi vieras 'guest', viaras, StFi siellä 'over there', siällä, and StFi syö 'eat', syä. 5) (apoc)-variables: Apocope of the final vowel of some case endings is widespread in the dialects, especially in west Fi ones, like for Adessive {-llA} (A = vowel harmony variants), which expresses, e.g., locality ('at, on, by'), and which further is used in the habere-construction. The apocope is frequent in some sandhi contexts, like the habere construction and in adverbs like siellä StFi 'over there' : siellØ on 'there is/are'. The extent to which apocope is used in the dialects studied differs considerably. Another case ending, the directionality case Allative {-lle}, differs even more: in the southwest Fi dialects Adess and All have merged, but in the other dialects the apocope of All is clearly more restricted, or is not considered to occur at all. Hypothetically, merging of the cases, especially for southwest Fi speakers, may give rise to confusion when speakers have acquired the Swedish prepositional system regarding locality vs. directionality, which differs from the Finnish one. 6) (III INF)-variable: The variation is mainly morphologically determined, with various phonological variants interfering in the variation. For the 3rd Infinitive Illative, the morpheme of the 3rd Inf does not occur in some dialects. The omission of the morpheme has been observed as a superregional high-status feature for some decades in Finland. The StFi variant contains the 3rd Inf morpheme {-mA}, e.g., menit teke+mä+än 'you went to do/make' (2nd p sg), southwest Fi menit tekemähä (phonol. variation), Häme Fi menit tekeØen ~ tekemään (morphol. and identical (StFi) variant), north Fi menit tekeØen ~ tekheØen ~ tekemä(än) ~ tekemhään (phonol. variants of different morphol. variants), east Fi menit tekemään ~ tekemmää(n) (phonol. and identical (StFi) variants). Especially for second generation Finnish, this variety level variation may cause confusion for the 1st Inf and 3rd Inf Ill of some frequent verb types. 7) (1st P P1)-variable: Lack of subject - verb concord of 1st p pl has become a general, informal variant in many Finnish varieties and espe-

< previous page

page_36

next page >

< previous page

page_37

next page > Page 37

cially east Fi dialects. The Passive (=Pass) voice (impersonal) variant of the verb is used instead of the Active (=Act), inflected forms. Several morphosyntactic structural conflicts arise from this (cf. Lainio, 1991b). StFi has concord variants, e.g., (me) mene+mme 'we go' (both the optional pronoun and the suffix express '1 p pl'), the Pass voice, non-concord variant me mennään 'we go'. Both the concord and Pass variants occur, with phonological variants of the verb, in the dialects involved. Table 1 The linguistic variants of seven variables studied by Lainio (1989) and the pronoun variable studied by Lainio & Wande (1991). Variable StFi west Fi east Fi 1) (d) d r, ¶, I Ø,j, v, h, t (incl. north Fi) 2) (ts) ts:ts tt:t, tt:tt ht:t, ht:h (incl. north Fi) 3) (schwa) no schwa V no schwa V +schwa V (incl. north Fi) 4) (diphth) no diphth. +diphth. no diphth. (incl. north Fi) 5 (apoc) no apocope +apocope no apocope (incl. north Fi) 6) (III Inf) +{-mA} +/-{-mA) +{-mA) no phonol. +phonol. +phonol. variation variation variation (incl. north Fi) 7) (1st P Pl) active voice, active voice active voice, subj.-verb subj.-verb subj.-verb agreement agreement agreement/ no phonol. +/-phonol. +/- phonol. variation variation variation, (incl. north Fi) passive voice, no subj-verb agreement 8) (minä) minä mää, mä mie, (minä) (minä) (incl. north Fi) 8) (minä)-variable: StFi has the pronoun minä, which should be used when emphasized Since the verb carries the 1st p sg suffix as well, it is, according to this prescriptive view, not used otherwise. A main variant is the superregionally spread monosyllabic {mA}: mä 'I', which

< previous page

page_37

next page >

< previous page

page_38

next page > Page 38

further is considered the main Häme variant. The StFi variant also occurs in Häme and some east Fi dialects. Other main present-day variants for the dialects in question are given in Table 1. In addition to the seven first variables, Lainio & Wande (1991) studied the eighth variable, the use of the pronoun for 1st p sg. The informants were identical to the ones in the FIDUS project. In addition to the mere occurrence several other factors were taken into account, among others prosodic and contextual features (pro-drop, topicalization and verb form). One aim of this study was to investigate typological differences between the dialects. Results of the study When considering the results, it is worth recalling the hypotheses. It was believed that west Fi dialects, with a traditionally higher prestige might influence speakers of east Fi dialects, and vice versa, that east Fi dialects, represented by a higher number of speakers, might influence speakers of west Fi dialects. Several aspects have to be considered when evaluations of the results are made. First, the control data of Finland Finnish use is only to some extent represented by corresponding data, i.e., present-day, quantitative data from speakers of the same age groups. This is one reason why the use of StFi is seemingly clearly increasing among the Sweden Finns. The Finland Finns from the identical areas have probably adopted StFi to an even higher degree, but data confirming this are lacking. Second, data above are averages, which means that some speakers do not show these tendencies. There are actually four or five different main types of adopters among the informants: it is rather a question of continuous adaptations than either-or-choices. Third, there are differences between the sexes as well, but the reasons for these differences have not been considered depending on sex per se, but on circumstances in the bibliographical backgrounds of the speakers. Some of the data suggest that similar changes, typical for women and men respectively in Finland, are taking place among these groups of speakers. Fourth, most variables contain several competing variants (e.g., the (d)-variable includes nine variants). This means that the tendencies described in these tables (2 and 3) only show what variants have been added to the ones considered inherent in the respective (traditional) dialect. The results here do not show to what extent the older variants have become obsolete (for a discussion, cf. Lainio, 1989, 1990, 1991b; Lainio & Wande, 1991). In

< previous page

page_38

next page >

< previous page

page_39

next page > Page 39

spite of these and other, minor reservations (see also Lainio, 1991b), the following conclusions about the west Fi speakers are feasible. Table 2 Changes indicated for speakers of West Finnish origin in Sweden Concluding main tendencies for the FIDUS project. West Finnish Variable Southwest Fi Häme Fi +StFi, +east Fi 1) (d) +StFi, +east Fi +StFi +StFi 2) (ts) +east Fi +east Fi 3) (schwa) +StFi=east Fi +StFi=east Fi 4) (diphth) 5) (apoc)* 6) (III Inf)**

+StFi +StFi, +east Fi

+superreg. +StFi

7) (l P Pl) 8) (minä)***

+superreg. +StFi=east Fi, +superreg. N=10

+superreg. +StFi=east Fi N=7

*= concerns Adessive; **= concerns Illative; *** variable studied by Lainio & Wande. Italics = the east Fi variant identical with the superregionally spread variant in Finland. 'StFi = east Fi' = the variants are identical. Underlining = change indicated has not been stated for corresponding dialects in Finland. Variables 14 = phonological, 5-6 = morphophonological, 7-8 = morphosyntactic variables. Some east Fi influence can be observed (Table 2). The possible east Fi influence is of three kinds, however, and it follows a generalizable pattern. East Fi variants are preferred and probably supported in the linguistic milieu in Sweden, not only due to use among speakers of east Fi dialects, but also since they occur or are in fact identical with the StFi variant (e.g., variables 4 (diphth) and 8 (minä)), or, the superregional development of spoken FiFi (variable 1). The general pattern is, that east Fi features which have support from a high prestige variety in Finland, and consequently, Sweden, are spread. One variable does,

< previous page

page_39

next page >

< previous page

page_40

next page > Page 40

however, show a kind of independent Sweden Finnish development, variable 3 (schwa). The corresponding use of schwa-vowels among speakers in Finland from these dialect groups has not been stated. Furthermore, this is a mainly male feature, both for speakers of the dialects in Finland, and for the majority of speakers of west Fi origin that start using it in the Sweden Finnish milieu. Another generalization traceable in the results, is that the use of highly dialectal features is diminishing on a general basis, even if they are retained by some speakers, who fairly consequently retain other dialectal features as well. Both women and men can be found among these speakers. If one considers the changes indicated, some variables show new east Fi influence, (five instances together for both the southwest Fi and the Häme speakers; cf. underlined indications, Table 2). The difference compared to Finland is only to a limited extent a sharp, qualitative one. Some of the new east Fi influence indicates an extension with regard to the grammatical structure where the new variant is used, compared to the use observed in Finland. Some of the changes imply a closer connection to a standardized, or maybe rather a less dialectal, and therefore more superregional Finnish. The reservations mentioned for the earlier dialect groups are to be considered here as well (Table 3). For the north Fi and east Fi speakers several tendencies similar to those of the west Fi dialects are found. The StFi influence is increasing. But also an extended use of west Fi variants is obvious. For two variables the west Fi variant is identical with the StFi one (variables 3 (schwa) and 7 (1 P Pl), the latter for east Fi only). For variable 6 (III Inf), the west Fi variable is identical with a superregional variant. The only variable showing an exclusive west Fi influence is 4 (diphth). Several of the variants that do not represent prestigious varieties in present-day Finland, did so at the time of the emigration of a majority of the Sweden Finns (e.g., west Fi (d) and (ts) variants). This is corroborated by the fact that the mixed north Fi dialect group only has adopted one new east Fi variant (variable 6 (III Inf)), compared to five west Fi.

< previous page

page_40

next page >

< previous page

page_41

next page > Page 41

Table 3 Changes indicated for speakers of north and east Finnish origin in Sweden. Concluding main tendencies for the FIDUS project. Variable West/East Finnish East Finnish north Fi east Fi (Kemi dialects) (North Karelian) +StFi, +west Fi 1) (d) +StFi, +west Fi +StFi +StFi, + west Fi 2) (ts) +StFi=west Fi +StFi=west Fi 3) (schwa) +west Fi +west Fi 4) (diphth) 5) (apoc)* 6) (III Inf)**

+StFi, +superreg. +superreg. +StFi, +east Fi, +StFi, +west Fi +west Fi 7) (1 P Pl) +StFi=west Fi +superreg. +west Fi, +superreg. 8) (minä)*** +StFi, +west Fi, +superreg. N = 16 N = 21 *= concerns Adessive; **= concerns Illative; *** variable studied by Lainio & Wande. Italics = the west Fi variant identical with the superregionally spread variant in Finland. 'StFi = west Fi' = the variants are identical. Underlining = change indicated has not been stated for corresponding dialects in Finland Variables 1-4 = phonological, 5-6 = morphophonological, 7-8 = morphosyntactic variables. The north Fi and east Fi speakers constitute a majority of speakers, and they show a willingness to adopt variants from the west Fi varieties, but the representatives of the prestigious west Fi dialects show less inclination to adopt variants from the north and east Fi majority varieties. A rough and very general conclusion would be that prestige as a factor of importance for linguistic change is more important than number of speakers in this minority context, at least for these first generation speakers. In principle, however, the ways by which the speakers adopt new variants are the same: general features, with some prestige support are spread, but not features typical of a local dialect. Archaic features generally seem to be abandoned.

< previous page

page_41

next page >

< previous page

page_42

next page > Page 42

The respective variables which indicate a different direction, i.e., west Fi speakers adopt east Fi variants and vice versa, without the support of a prestigious variety, are mainly continuous phonological variables. One can probably conclude that these changes are unconscious initially. On the other hand, e.g., the use of the 1 p sg pronoun also shows that changes that probably to some degree are conscious, have taken place. For example, the north Fi speakers, to whom the indigenous variant mie has been considered an identity marker of north Fi speakers in Finland have generally stopped using it. The use of StFi variants is not very widespread, despite the fact that the dialects have become more standardized. This may partly be explained by the recording situation. The speakers' performance was only analysed in one situation: an informal gettogether with friends or relatives. The socio-cultural orientation and identification of these speakers is still fairly much towards the Finland Finnish development, of which they maintain knowledge, via visits to and from Finland, through close media contacts, and until recently, fresh immigration. If one should talk of typically Sweden Finnish features, these would rather consist of new combinations of old features, than of totally new features. No doubt, however, can some changes be observed in the speech of the informants, that are not probable for corresponding speaker groups in Finland, not even among speakers who have migrated within Finland. Most changes are nevertheless related to similar changes already started in the Finland Finnish milieu. The variants that seem to diffuse follow a pattern noted in other language contact milieux: features shared by speakers survive more easily (e.g., Le Page & Tabouret-Keller, 1985, on 'focusing' in Creole language contacts). In their view, this is connected to identification between speaker and listener and her/his group, and to other psychosocial factors. Similar phenomena have been observed among the Norwegian settlers of Spitsbergen (Maehlum, forthc.). Poplack (1990) has seen this as one type of linguistic convergence in studies performed in different language contact situations in North America and elsewhere. Appel & Muysken (1987, and the literature referred to there) discuss similar phenomena in terms of 'strategies of neutrality', which include both ingroup and inter-group purposes. It may further be seen as a basic function and may be tied in with the social network theory (Milroy, 1980: 114-15; Milroy & Milroy, 1985). However, the mere sharing of a linguistic feature by speakers of different varieties did not seem to promote this development among the Sweden Finns.

< previous page

page_42

next page >

< previous page

page_43

next page > Page 43

Reality is more complex than these generalizations would have us believe, however. Four or possibly five 'types of variety shifters' seemed to emerge, similar to the results in other urbanisation studies (cf. Lainio, 1989, Appendix). One type was the dialect preserver, who to a large extent used dialect stereotypes and dialect markers, but who, nevertheless, had adopted some variants that were not inherent in her/his own dialect. A second speaker type, rare in studies in Finland, is one who goes from using one non-standard variety to start using another in a new surrounding, a local variety adopter. This type was rare also in the FIDUS study, since it would imply a person who fairly indiscriminately chooses variants from the options of a total language and variety contact milieu. This kind of wild mixer does not occur. Some speakers have adopted features that may not occur in Finland Finns' language use, which could be signs of the development of a Sweden Finnish variety. Its ultimate representative would be a speaker of 'Sweden Finnish'. A third type was the kind of speaker who distances her-/himself from the features typical of local dialects, a dialect avoider (by choosing this term, I have stressed the avoidance of dialect features, not the adoption of the superregional variants). This type seemed to head towards a non-identifiable spoken language profile, using features that are common and known from the urban superregional variety of Finland, some standard language features, and features that may be common in the Sweden Finnish context. A fourth type observed in studies of people migrating within a language community, the standardized speaker, did not seem to be a model adopted frequently among the Sweden Finns (partly as a consequence of method, as was noted above). A fifth category, rather than type, in the FIDUS study, were speakers who have cautiously retained easily identifiable features of their original dialect, but who have otherwise adopted features from other regional dialects, from superregional speech varieties, and from StFi. The dialect features they use seem to be matters of conscious choice. These features, often morphological, appear to be known by the speech community at large to be dialect markers. The whole Sweden Finnish speech community does not seem to follow the route taken by, for example, Finnish emigrants to North America in earlier decades, who, according to most sources, are characterized by an abundant use of archaisms, and in some recent studies (Jönsson-Korhola, 1989; Martin, 1989; cf. Wheeler, 1987) by internal changes quite different from the Finland Finnish language development Though not studied in depth yet, it is possible that variety contact

< previous page

page_43

next page >

< previous page

page_44

next page > Page 44

among American Finns has lead to diffusion of features in similar ways as among Sweden Finns. A methodological difficulty remains: in neither case is it possible to say what has actually been acquired by the emigrant when still living in Finland. The local speech communities in Finland have constantly changed, presumably variably, but mainly in more standardized and urbanised directions. The language use of this first generation of immigrants in the FIDUS project showed that internal influences between varieties were largely parallel to those of speakers in Finland. One reason for this, in addition to those mentioned, may be that changes initiated already in Finland are not easily reversed by the first generation in the new surrounding. A reason for variety change that has not yet been discussed here is the linguistic/communicative distance between the dialects. Even though two of the dialects involved in the FIDUS project could be said to represent a maximal dialectal distance, the southwest Fi (=west Fi) and the North Karelian (=east Fi), this is generally not a problem for first-generation speakers: the mutual understanding between speakers of different dialects is relatively good. Only single dialect maintainers of these two dialects expressed concern about the problem of understanding, this being a reason why they could not use 'maximal dialect' when speaking to other Sweden Finns. Nevertheless, speakers of all four dialect groups seem to be involved in similar processes of change. Some of the 'classical characteristics' in sociolinguistic studies could be observed in the FIDUS study. Women generally used, especially for some of the phonological variables, more StFi variants. Younger women also used some other innovations to a higher extent than any other subgroup of speakers. The same variants have been seen as typically introduced to new speech communities in Finland by younger women. Even if this pattern is classical, it is still not well understood beyond the surface correlations (Milroy & Milroy, 1985). A major Finnish social setting in Sweden Finnish daily life, at least for men, has been the factory where they work. High concentration of Finns in the same plant and even during the same shift has been another characteristic. The use of schwa/epenthetic vowels and other phonological variants suggest that there is a male variety developing. These are parallels to developments in Finland, both regarding linguistic variants and social subgroups. The use and distribution of variants in the original dialects in Finland, and how they overlap with either another dialect, superregional

< previous page

page_44

next page >

< previous page

page_45

next page > Page 45

speech patterns/varieties, or the standard, are crucial factors for the outcome. Another obvious feature was that in-between variants occurred, that have not been stated in the source dialects, or as allophonic (or corresponding) variants. Often it was felt that these fit into patterns as if they were floating, 'on their way', from one main variant to another. This was typically so on the phonological level, but similar variants could be observed on the other levels as well. Presumably, this kind of structural 'aberration' is of greater importance in extensive language contact situations, where the impact of a prescriptive, stabilizing variety may be reduced. On the individual level, evidence for the reasons as to why a speaker chose to use particular variants could not be found. This is due first of all to the heterogeneous migration patterns, which have included several stops on the way to Eskilstuna. Only about one fourth of all potential informants of the FIDUS study, i.e., the respondents of the initial survey (Lainio, 1984), had moved directly from their birth place to Eskilstuna. Secondly, it is impossible to state from the present results how deeply integrated the forms and variants of the linguistic variables studied are into the vernacular of these speakers, or how frequent they are outside the mostly informal settings created in their homes during these recordings. What further seems obvious from studies like this on Sweden Finns' L1, is that the mother tongue variety the speakers have been socialized into, needs to be taken into account when dealing with a description of the spoken L1 development in a new, immigrant milieu, and probably also when dealing with L2 use and acquisition. Reversely, L2 contact has a variable impact on speakers from different L1 background. Lainio & Wande (1991) found, for example, L1 variation that indicated varying typological distance to Swedish between the dialects studied (cf. Andersson, forthc.). Discussion As indicated above, Sweden Finnish seems to be developing its own combinations of linguistic variation, maybe even an internal contact variety. In the first generation the contacts with Finland Finnish are largely retained. The basis for the second generation consists of the original varieties of their parents, the developing common variety, and most likely, an increasing impact of Swedish. This could be a starting

< previous page

page_45

next page >

< previous page

page_46

next page > Page 46

point for further research on Sweden Finns' L1, as well as L2. One way to carry out this research is to narrow the gap between L1 studies and L2 learning studies. This raises methodological difficulties for the study of both their Finnish and Swedish. Other problems in this respect are, e.g., that Finnish has been in constant contact with Swedish for seven or eight centuries, and before that, with other Indo-European (Germanic, Baltic, Slavonic) languages. There may be reason to posit a 'typological drift', which works on different grammatical levels, in addition to the obvious, mostly lexical, borrowings. Even though the individual speaker decides what innovations will be adopted, there are many instances of this process on language level that have been out of reach for linguistic descriptions (cf. Milroy & Milroy, 1985; cf. Leino, 1982, for Finnish). Especially for syntactic changes these processes may be difficult to verify, also with quantitative methods, due to the relative infrequency and the referential side of the features (cf. Sankoff, 1988). Further, when should such influence be counted as a long-term cultural and linguistic impact on the whole mother tongue language community and its varieties in their native surrounding, and when should it be seen as the impact of, in this case, Swedish in Sweden? On the one hand, Finnish vocabulary has received hundreds of innovations directly or indirectly via Swedish, which by monolingual majority speakers may be considered 'pure' Finnish (e.g., Pulkkinen, 1984; cf. also, Poplack, Sankoff & Miller, 1988). On the other hand, reborrowed items are sometimes believed to be fresh innovations (Grönholm, 1989; Andersson, forthc.). Many contact phenomena remain diffuse and can perhaps only be clarified by attempts to quantify the use of the features in both the old and the new, minority contact surrounding of L1. Such features may, however, be unequivocally described in grammars (cf. Poplack, forthc.). When the minority L1 is passed on to second-generation speakers in a present-day emigrant situation, the speakers' core contact options move in the direction of the surrounding majority language, L2. Among the L1 varieties, which the speakers first lose contact with, is the standard, while standard L2 is introduced late, at school age (but it is implicit in the spoken language to some degree). Similarly, the less well-defined superregional, prestigious varieties of L1 in the old country have their impact slackened in the new milieu. There is then more space for internal development and for the second language to have an impact on L1 in the minority milieu. The extent to which contact with the standard is upheld becomes crucial. In the early, formative years

< previous page

page_46

next page >

< previous page

page_47

next page > Page 47

there is probably more space for internal development in both L1 and L2 than later. In addition, regular extensions of internal linguistic rules may cause deviations from monolingual native speakers' use, which may be based on socially determined exceptions. Similar developments have even influenced majority language use (e.g., Horvath, 1985). In the case of Sweden Finns, projects dealing with both L1 and L2 are at the present integrating and studying the bilingual aspect to a higher extent (e.g., Viberg, 1988; Juvonen, Lindberg & Viberg, 1989; Boyd & Andersson, 1991; Boyd, Andersson & Thornell, 1991; Wande, 1991). Earlier methods to study L1 and L2, and theoretical concepts used apparently need some reevaluation. For the study of L2 use, e.g., 'fossilization', and, for L1 use, its recently developed counterpart 'attrition' may need to be discussed. Most adult learners do not attempt to achieve a full competence in L2 in the long run. 'Full competence' is a rough idealization based on an ideal standard language speaker image and the structure of the standard, even for a native speaker. Second-generation immigrants usually acquire native skills in L2, and may never have developed fully monolingual-like L1 skills. Furthermore, if L2 learners, according to Selinker (1972, cf. Ellis, 1985), fossilize in 95 % of cases, i.e., only about 5% reach the monolingual level as learners, what is the value of the concept, and what implications does it have for connotations about the speakers? The 'interlanguage' concept does not seem to consider the problem that a speaker at some point in time is no longer mainly a language learner but a language user. Practical problems in defining the phases presumably contribute to this: where does interlanguage/learning stop and become variable L1-like use, revealing e.g., ethnic markers (cf. Tarone, 1988: 13338). There are few variational studies of long-term immigrants' use of L2, as well as adult native L1 speakers' adoption of new linguistic features and de-learning of old ones, which could be related to concepts like 'fossilization' and 'attrition'. There is a gap between studies of initial L2 learning/acquisition and the essentially L1-like variability of a L2 speaker. There is also a gap between the few minority L1 studies so far and large-scale studies of the integration of L2 features into L1. The role of L1 and its internal variation at these different stages needs further studies. A possibility to integrate some of these factors can probably be found in various attempts to study individuals' language use on the continuum from code-switches to fully integrated features. There is also reason to study more in depth the reasons for speakers of second and later generations of immigrants not to retain L1 sty-

< previous page

page_47

next page >

< previous page

page_48

next page > Page 48

listic distinctions in a minority milieu (c.f. Gal, 1979, 1984; Lavandera, 1984). The identical situations may not arise in the new surrounding, and when they do, it may be more appropriate to use varieties of L2. Do, for example, new non-standard varieties and new linguistic variants in the minority milieu replace the former features connected to different degrees of formality in L1 discourse in a majority situation? There is no need to assume that the perception of and need for linguistic expression of informality - formality (or corresponding counterparts) as such has disappeared among immigrants. Even if the relation to the home country's prescriptive ways of dealing with style of L1 remains crucial in the emigrant community, it is presumably so mainly when the minority speaker communicates with someone from the former L1 speech community. In this respect, bilingual immigrant spectra of style may just be other ways of expressing stylistic differences than the prescriptive monolingual ones. The FIDUS project and the bilingual studies referred to above constitute the very first attempts to link these different aspects together in the study of Sweden Finnish. Acknowledgements The present chapter was written during my stay as a visiting researcher at the University of Ottawa, Canada. The stay was made possible thanks to funding from the Sweden Tercentiary Foundation, scholarships from the Sweden-America Foundation and the Canadian Government. I also would like to express my gratitude to Keltie Purcell, Ottawa, and Erling Wande, Stockholm, for comments on this chapter. References ALLARDT, E. & C. Starck 1981), Spräkgränser och samhällsstruktur. Finlandssvenskarna i ett jämförande perspektiv. Stockholm: AWE/Gebers. ANDERSSON, P. (forthc.), Blandat och klart: svenska inkorporeringar i minoritetsfinska och -engelska. To appear in: Wande, E. et al. (eds.), Finska varieteter utanför Finland. (Stockholm Studies in Finnish Language and Literature 8). Stockholm University. APPEL, R. & P. Muysken (1987), Language contact and bilingualism. London: Edward Arnold. BAETENS BEARDSMORE, H. (1986), Bilingualism: Basic Principles. 2nd edition. Clevedon, Avon: Multilingual Matters.

< previous page

page_48

next page >

< previous page

page_49

next page > Page 49

BOYD, S. (1985), Language Survival. A Study of Language Contact, Language Shift and Language Choice in Sweden. (Gothenburg Monographs in Linguistics 6.) University of Göteborg. BOYD, S. (forthc.), Paper at the conference Language progression and regression, Fittja Gård, Botkyrka, August 1990. Department of Linguistics, University of Gothenburg. BOYD, S. & P. Andersson (1991), Linguistic change among bilingual speakers of Finnish and American English in SwedenBackground and some tentative findings. International Journal of the Sociology of Language 90, 13-35. BOYD, S., P. Andersson & C. Thornell (1991), Patterns of incorporation of lexemes in language contact: language typology or sociolinguistics? In: Papers for the Symposium on Code-Switching in Bilingual Studies: Theory, Significance and Perspectives. Vol. II. Barcelona, 21-23 March 1991. Strasbourg: ESF, 463-488. BRATT PAULSTON, C. (1982), Swedish research and debate about bilingualism. A critical review of the Swedish research and debate about bilingualism and bilingual education in Sweden from an international perspective. A Report to the Swedish National Board of Education. Stockholm: National Board of Education. EDWARDS, J. (1985), Language, Society and Identity. Oxford: Basil Blackwell. EKSTRAND, L-H. (1978), Bilingual and bicultural adaptation. Studies in assessment of second language learning and of factors related to bicultural adjustment, with special reference to immigrant children. Institute of International Education. University of Stockholm. EKSTRAND, L-H., S. Foster, E. Olkiewicz & M. Stankovski (1981), Interculture: Some Concepts for Describing the Situation of Immigrants. In: Journal of Multilingual and Multicultural Development, 2: 4, 269-95. ELLIS, R. (1985), Understanding Second Language Acquisition. Oxford: Oxford University Press. GAL, S. (1979), Language Shift. Social Determinants of Linguistic Change in Bilingual Austria. London and New York: Academic Press. GAL, S. (1984), Phonological Style in Bilingualism: The Interaction of Structure and Use. In: D. Schiffrin (ed.), Meaning, Form and Use in Context: Linguistic Applications. Washington D.C.: Georgetown University Press, 290-301. GRÖNHOLM, M. (1989), Ruotsalaiset lainasanat Turun murteessa. Åbo: Åbo Academy Press. HAMMAR, T. (1985), Immigration to Sweden in 1985 and 1986. In: SOPEMI report: Immigration to Sweden in 1985 and 1986. (Center for Research in International Migration and Ethnic Relations, Report 4.) Stockholm University, 1-37. HANSEGÅRD, N-E. (1968), Tvåspråkighet eller halvspråkighet? Stockholm: Aldus/Bonniers. HANSEGÅRD, N-E. (1990), Den norrbottensfinska språkfrågan. En återblick på halvspråkighetsdebatten. (Uppsala Multiethnic Papers 19). Centre for Multiethnic Research. Uppsala University. HORMIA, O. (1976), Finska invandrares språksituation. (Unpublished ms.) Finska institutionen, Stockholms universitet. HORVATH, B. M. (1985), Variation in Australian English. The sociolects of Sydney. Cambridge: Cambridge University Press. HUJANEN, T. (1986), Kultamaa ja kotimaa. Tutkimus Ruotsin ensimmäisen polven suomalaissiirtolaisten SUOMI JA RUOTSIkuvasta. (Acta Universitatis Tamperensis. Ser. A, Vol 205.) Tampereen yliopisto.

< previous page

page_49

next page >

< previous page

page_50

next page > Page 50

HUSS, Leena (1991), Simultan tvåspråkighet i svens-finsk kontext. (Acta Universitatis Upsaliensis. Studia Uralica Upsaliensia 21.) Uppsala. HYLTENSTAM, K. (1978), Progress in Immigrant Swedish Syntax. A Variability Analysis. University of Lund. HYLTENSTAM, K. (1989), Svenska som andraspråk. Forskning och undervisning i Sverige. In: Nuopponen, A. & R. Palmberg (eds.), Special Languages and Second Languages: Methodology and Research. AFinLA Yearbook 1989. Vaasa, 27-49. HYLTENSTAM, K. & L. Arnberg (1988), Bilingualism and education of immigrant children and adults in Sweden. In: C. Bratt Paulston (ed.), International Handbook of Bilingualism and Bilingual Education. New York: Greenwood Press. HYLTENSTAM, K. & E. Magnusson (1982), Typological markedness, contextual variation, and the acquisition of voice contrast in stops by first and second language learners of Swedish. Paper presented at the 6th International Congress of Applied Linguistcs, Lund, August 9-14, 1981. In: Indian Journal of Applied Linguistics. JAAKKOLA, M. (1973), Språgränsen. En studie i tvåspråkighetens sociologi. Stockholm: Aldus/Bonniers. JAAKKOLA, M. (1983), Finnish Immigrants in Sweden: Social Networks and Life Styles. (Research Group for Comparative Sociology. Research Report 30/1983.) University of Helsinki. JUVONEN, P., I. Lindberg & Å. Viberg (1989), Narrative skills in two languages. In: Scandinavian Working Papers on Bilingualism, 8, 49-69. JÖNSSON-KORHOLA, H. (1989), Amerikansuomen kaasussyntaksin erityispiirteitä. (Unpublished lic. thesis). Suomen kielen laitos, Helsingin yliopisto. KARLSSON, F. (1983), Finnish Grammar. Porvoo, Helsinki and Juva: WSOY. KNOOKALA, T. (1982), En beskrivande studie över tvåspråkighet bland finska invandrare. Kodväxling och interferens. (FUMS rapport 105). Uppsala. LAINIO, J. (1984), Finsk dialektutveckling i en svensk industristad (FIDUS). Slutrapport 1. Språksociologisk del. (FUSKIS/FIDUS 8). Uppsala. LAINIO, J. (1987), Language Use of Finns in Sweden. Implicational Patterns in four Domains. In: Scandinavian Working Papers on Bilingualism, 7, 1-28. LAINIO, J. (1989), Spoken Finnish in Urban Sweden. (Uppsala Multiethnic Papers 15). Centre for Multiethnic Research, Uppsala University. LAINIO, J. (1990), Medelålders nordkalottsemigranterhur profilerar de sin finska? In: I. Almqvist, P-E. Cederholm & J. Lainio (eds.), Från Pohjolas pörten till kognitiv kontakt. (Stockholm Studies in Finnish Language and Literature 6.) Stockholm University, 85-104. LAINIO, J. (1991a), Syntactic Consequences of Varietal Contact. In: H. Sigurdsson et al. (eds.), Papers from the 12th Scandinavian Conference of Linguistics. Reykjavçk: Linguistic Institute, University of Iceland, 205-18. LAINIO, J. (1991b), Från finsk landsbygd till mellansvensk stadsbygd. In: K.L. Berge & U-B. Kotsinas (eds.), Storstadsspråk och storstadskultur i Norden. (MINS-rapport 34). Department of Scandinavian Languages, Stockholm University, 108-44. LAINIO, J. & E. Wande (1991), Pronominet minä 'jag' i urban sverigefinska. In: Fenno-Ugrica Suecana, 10, 17-60. LAVANDERA, B.R. (1984), The Variable Component in Bilingual Performance. In: D. Schiffrin (ed.) (1984), Meaning, Form and Use in Context: Linguistic Applications. Washington D.C.: Georgetown University Press, 391-409. LEINO, P. (1982), Suomen kielen lohkolause. (Suomi 124: 2.) Helsinki: SKS.

< previous page

page_50

next page >

< previous page

page_51

next page > Page 51

LE PAGE, R. & A. Tabouret-Keller (1985), Acts of Identity: Creole-based Approaches to Language and Ethnicity. Cambridge: Cambridge University Press. LIEBKIND, K. (1989), Patterns of ethnic identification amongst Finns in Sweden. In K. Liebkind (ed.), New Identities in Europe. Aldershot: Gower Publishing Co Ltd/European Science Foundation, 116-39. LINELL, P. (1982), The Written Language Bias in Linguistics. (Studies in Communication 2.) University of Linköping. LOMAN, B. (ed.) (1974), Språk och samhälle 2. Språket i Tornedalen. Lund: CWK Gleerup. MAJAVA, A. (1981), Finns in Sweden: Characteristics and Living Conditions. In: M.G. Karni (ed.), Finnish Diaspora I: Canada, South America, Africa, Australia and Sweden. Toronto: The Multicultural History Society of Ontario, 231-53. MARTIN, M. (1989), Amerikansuomen morfologiaa ja fonologiaa. (Unpublished lic. thesis) Department of Finnish, University of Jyväskylä. MARTIN-JONES, M. & S. Romaine (1987), Semilingualism: A Half-Baked Theory of Communicative Competence? In: E. Wande et al. (eds.), Aspects of Bilingualism. (Acta Universitatis Upsaliensis. Studia Multiethnica Upsaliensia 2). Uppsala, 87104. MILROY, J. & L. Milroy (1985), Linguistic change, social network and speaker innovation. In: Journal of Linguistics, 339-84. MILROY, L. (1980), Language and social networks. Oxford: Basil Blackwell. MILROY, L. (1987), Observing and Analysing Natural Language. Oxford: Basil Blackwell. MÆHLUM, B. (forthc.), Dialect Socialization in Longyearbyen, Svalbard (Spitsbergen): A Fruitful Chaos. (Manuscript). Tromsø University. MÄKELÄ, K. (1985), Kulttuurisen muuntelun yhteisöllinen rakenne Suomessa. In: Sosiologia, 4, 247-60. NATCHEV, E. & U. Sirén (1988), Goddag, Hassan Heinonen. Förskolebarn i Stockholm med dubbelt minoritetsursprung. Stockholms Socialförvaltning, Hägersten. NATCHEV, E. & U. Sirén (1989), Språkanvändningfrån avsikter till förverkligande. Stockholms Socialförvaltning, Stockholm. NESSER, A. (1986), ''Se oli kurja käya koulusa". Bruket av inre och yttre lokalkasus i sverigefinska skolbarns språk. (Acta Universitatis Upsaliensis. Studia Uralica et Altaica Upsaliensia 16). Uppsala. NESSER, A. (1990), "Mie meinasin istoa minun enon hevosen päälle". Om possessivkonstruktionen i sverigefinska skolbarns språk. In: I. Almqvist, P-E. Cederholm & J. Lainio (eds.), Från Pohjolas pörten till kognitiv kontakt. (Stockholm Studies in Finnish Language and Literature 6.) Stockholm University, 160-74. OLKIEWICZ, E. (1990), Invandrarfamiljer i förändring. En studie över finska och jugoslaviska invandrarungdomar och deras föräldrar. (Studies of Comparative and International Education 17). Institute of International Education, Stockholm University. OUVINEN-BIRGERSTAM, P. (1987), Identitetsutveckling hos barn. En jämförelse mellan finska, jugoslaviska och svenska barn. In: E. Wande, et al. (eds.), Aspects of Bilingualism. (Acta Universitatis Upsaliensis. Studia Multiethnica Upsaliensia 2.) Uppsala, 265-81.

< previous page

page_51

next page >

< previous page

page_52

next page > Page 52

POPLACK, S. (1989), The care and handling of a mega-corpus: the Ottawa-Hull French project In: R. Fasold & D. Schiffrin (eds.), Language Change and Variation. Amsterdam: Benjamins, 411-51. POPLACK, S. (1990), Variation theory and language contact: concepts, methods and data. In: D. Preston, (ed.), American dialect research: An anthology celebrating the 100th anniversary of the American Dialect Society. Philadelphia: University of Pennsylvania Press. POPLACK, S. (forthc.), The inherent variability of the French subjunctive. In: T. Morgan & D. Lauefer (eds.), Proceedings of LSRL XIX. Amsterdam: John Benjamins. POPLACK, S., D. Sankoff & C. Miller (1988), The social correlates and linguistic processes of lexical borrowing and assimilation. In: Linguistics 26, 47-104. PRESTON, D. (1989), Sociolinguistics and Second Language Acquisition. (Language in Society 14.) Oxford: Basil Blackwell. PULKKINEN, P. (1984), Lokarista sponsoriin. Englantilaisia lainoja suomen kielessä. Helsingissä: Otava. REINANS, S.A. (1988), Finländare i Sverige 1985 - 2015. In Sverigefinländarna år 2000. (DEIFO Rapport 10.) Stockholm: DEIFO, 28-74. ROMAINE, S. (1989), Bilingualism. Oxford: Basil Blackwell. ROMMETVEIT, R. (1983), Prospective Social Psychological Contributions to a Truly Interdisciplinary Understanding of Ordinary Language. In: Journal of Language and Social Psychology, 2: 2-4, 89-104. RÖNMARK, W. & J. Wikström (1980), Tvåspråkighet i Tornedalen. Sammanfattning och diskussion. Pedagogiska institutionen, Umeå universitet. SANKOFF, D (1988), Sociolinguistics and syntactic variation. In: F.J. Newmeyer (ed.), Linguistics, the Cambridge survey. Vol. 4: Language; the socio-cultural context. Cambridge: Cambridge University Press, 140-61. SAVOLAINEN, M. (1987), "Jag skäms inte för att jag är finne". In: Y. Lithman, (ed.) Nybyggarna i Sverige. Stockholm: Carlssons Bokförlag, 80-121. SCB (1984), Tema Invandrare/Theme Immigrants. Living conditions, report 38. Official Statistics of Sweden. Stockholm/ Örebro: Statistics Sweden. SELINKER, L. (1972), Interlanguage. In: IRAL, 10, 209-31. SIRÉN, U. (1991), Minority language transmission in early childhood. (Studies in Comparative and International Education 21). Institute of International Education. Stockholm University. SKUTNABB-KANGAS, T. (1984), Bilingualism or Not: The Education of Minorities. Clevedon, Avon: Multilingual Matters. SKUTNABB-KANGAS, T. (1987), Are the Finns in Sweden an Ethnic MinorityFinnish Parents Talk about Finland and Sweden. (Research Project The Education of the Finnish Minority in Sweden, Working Paper 1.) Institute VI, Roskilde University. SOU 1983: 57, Olika ursprung - Gemenskap i Sverige. Utbildning för språklig och kulturell mångfald. Huvudbetänkande av invandrarutredningen. Ministry of Education. Stockholm. STROMQVIST, S. (1989), Perspectives on second-language acquisition in Scandinaviawith special reference to Sweden. (Gothenburg Papers in Theoretical Linguistics 55.) University of Gothenburg. TARONE, E. (1988), Variation in Interlanguage. London: Edward Arnold.

< previous page

page_52

next page >

< previous page

page_53

next page > Page 53

TOUKOMAA, P. (1983), Education through the medium of the mother tongue of Finnish immigrant children in Sweden. In: L. Dabène, M. Flasaquier & J. Lyons (eds.), IV. Status of Migrants' Mother Tongues/Le Statut des Langues d'Origine des Migrants. Strasbourg: European Science Foundation, 181-88. TOUKOMAA, P. & T. Skutnabb-Kangas (1977), The intensive teaching of the mother tongue to migrant children at pre-school age. Department of Sociology and Social Psychology, University of Tampere. VIBERG, Å. (1988), Tvåspråkighet i skolanen projektpresentation. In: B-L. Gunnarson, C. Liberg & A. Nesser (eds.), Barns tvåspråkighet. Association Suedoise de Linguistique Appliquée (ASLA). Uppsala, 155-97. VIRTARANTA, P. (1986), Utforskningen av språk och traditioner i mellersta Skandinaviens finnskogar. In: S. Huovinen (ed.), Värmlandsfinnar. Om finnskogens historia och kultur. Stockholm: Kulturfonden för Sverige och Finland, 86-113. VIRTARANTA, P. (1988), Considerations on American Finnish, Particularly its Vocabulary. In: M.G. Karni, O. Koivukangas & E. Laine, (eds.), Finns in North America. (Migration Studies C9). Institute of Migration, Turku, 320-34. WANDE, E. (1982), Tornedalsfinskan och dess särdrag. In: Finska språket i Tornedalen. Stockholm: Föreningen Norden/Kulturfonden för Sverige och Finland/Finn-Kirja, 40-73. WANDE, E. (1984), Two Finnish Minorities in Sweden. In: B. Molde & D. Sharp (eds.), Second International Conference on Minority Languages, June 1983, Turku/Åbo, Finland. (Journal of Multilingual and Multicultural Development, 5: 3-4, 225-41). WANDE, E. (1989), Socio-educational and linguistic aspects of Finnish in the Swedish school. In: Scandinavian Working Papers on Bilingualism, 8, 122-32. WANDE, E. (1991), Sverigefinnars två språk: språkbruk och attityder hos två generationer. In: K. Herberts & C. Laurén (eds.), Flerspråkighet i och utanför Norden. Papers from the Sixth Nordic Conference on Bilingualism, 4-6.6. 1990 Vasa. Åbo, 184203. WANDE, E. (forthc.), Fennistisk forskning utanför Finland. To appear in: M. Norman & E. Söderholm (eds.), Papers from the First Norwegian Symposium on Finnish Linguistics and Literature, April 1991,. Tromsø (Nordlyd. Tromsø University Papers on Language and Linguistics). Tromsø. WHEELER, S. (1987), Code-switching and Borrowing in a Finnish-English Bilingual Situation. (Unpublished M A thesis). Department of Linguistics, University of Ottawa WINSA, B. (1991), Östligt eller västligt? Det äldsta ordförrådet i gällivarefinskan och tornedalsfinskan. (Acta Universitatis Stockholmiensis. Studia Fennica Stockholmiensia 2). Stockholm.

< previous page

page_53

next page >

< previous page

page_ii

next page >

< previous page

page_55

next page > Page 55

2 South Asian languages in Britain Safder Alladina The presence of South Asian languages in Britain is not a phenomenon of this century. The presence of people of South Asian origin, and therefore their languages in Britain, dates back to centuries (see Alladina, 1986; Fryer, 1984 and Visram, 1986). The presence of South Asian languages in academic institutions occupied another realm. In 1897, the Indian Institute was set up at Oxford University and in 1889, the Oriental School was established at the Imperial Institute in London. Both these institutions offered South Asian languages. University College in London had appointed a professor of Gujerati in 1855. The Honour School of Oriental Studies at the University of Oxford, in 1866, offered Sanskrit, Hindustani, Hindi, Marathi, Bengali, Tamil and Telegu (OUP, 1887). In 1907, the Raey Committee stressed the urgent need for suitable teaching in London for those about to take administrative and commercial posts in the Indian Empire (SOAS, 1984). In 1916, the site occupied by the London Institute for the Advancement of Literature and Diffusion of Useful Knowledge was acquired for the School of Oriental Studies as a school of the University of London. Nine students were admitted in February 1917 and by July of that year the number had increased to 125. Within the first ten years over 3,000 students had been enrolled (SOAS, 1984). In 1938, the Royal Charter was amended to change the title of the school to 'School of Oriental and African Studies'. Language was an important component in the examinations for the recruitment of personnel for the Empire, and the London School of Oriental and African Studies serviced that need. There had been courses at Oxford, Cambridge and London to prepare candidates for the rigours of the Indian Civil Service Examinations. Urdu was a compulsory language for those wishing to join the Indian Civil Service. A 'language allowance' was awarded to those who had acquired the rudiments of the language of the subject nations they were going to administer. Urdu was a major language in this. Hailbury College in Hertfordshire, which specialised in preparing youngsters for the Civil Service examination, gave 'rudimentary instruction in Oriental languages' (Moorehouse,

< previous page

page_55

next page >

< previous page

page_56

next page > Page 56

1983). The Civil Service Commission Examination of Officers in the Army in Modern Foreign Languages gave tests in Urdu. Thus South Asian languages were well established as modern, foreign but relevant languages in academic institutions in Britain. The debates about the suitability and acceptability of these languages in educational institutions in Britain were only initiated in the last three decades. History It was after the Second World War that larger number of people of South Asian origin came to the metropolitan centres of Britain to meet the demand for labour in industries. The languages that were used daily in the homes, families and social networks were totally disregarded by the British education system. Some distinct phases of responses from the British education system can be identified since the middle of the century. No response or 'sink or swim' There had always been groups of people arriving into the British education system who brought with them their own cultures and languages. Historically, they had somehow managed to become part of the British school life. Children of South Asian origin were expected to fit into this pattern. Although individual teachers may have tried to respond to the needs of these children, generally, it was a situation of 'sink or swim'. There were many who swam and survived to tell their experience. However, there are no records of those who sank. Assimilation In the decade between the 50's and the 60's, the presence of a large number of children of South Asian origin in British schools began to be seen as a problem. These children were not becoming a part of the British school system like the earlier groups of other language speakers had. They were a disruption to the normal run of the classroom. It was claimed that the children were bringing about a decline in the standard of education. The solution was seen to be to take these children into Reception Centres and teach them English. The method of teaching English was a mixture of the colonial and missionary traditions. Parents

< previous page

page_56

next page >

< previous page

page_57

next page > Page 57

were advised by teachers to speak to their children in English at home. It was claimed that more than a certain percentage of children from other language backgrounds was detrimental to their own process of assimilation and the general standard of education in the school. Certain education authorities began to disperse these children among schools within their area (cf. Killian, 1979; Troyna & Williams, 1986). In an education system that claimed to be child-centred, the children would have to change while the school would remain unchanged and Anglo-centric. Integration By the early 70's, it began to be recognised that assimilation was an unrealistic goal. Integration, where elements of language and culture of 'alien' groups would be tolerated within the British society, was advocated. In terms of language education, children were taken into Language Centres for a part of the school day. English as a Second Language (ESL) became a wellestablished sector of the British education system. The ESL industry was still Anglo-centric in approach, content and personnel. The children continued to be segregated from the life and substance of state schools. The British school system continued to under-achieve in its relation with children of South Asian origin (see, for example, the Rampton Committee Report, 1981). Multicultural education It was in the mid-70's that the notion of multicultural Britain, and its corollary, multicultural education, began to be articulated. The cultural and linguistic diversity of Britain were recognised, and they were to be celebrated in schools. In the area of language education, the home languages of children were to be valued and made use of in the classroom. The ESL industry was losing its credibility and segregation of children was seen to be insupportable. The Bullock Committee Report (Bullock, 1975) acknowledged the importance of home language in a child's educational development. For the first time, an important government document recommended that: In a linguistically conscious nation in the modern world we should see bilingualism as an asset, as something to be nurtured, and one of the agencies which should nurture it is the school. Certainly the school should adopt a positive attitude to its pupils' bilingualism and wherever possible should help maintain and

< previous page

page_57

next page >

< previous page

page_58

next page > Page 58

deepen their knowledge of their mother-tongues. (p.294). And: No child should be expected to cast off the language and culture of the home as he crosses the school threshold, and the curriculum should reflect those aspects of his life. (p.543). The views of the Bullock Report reflected the views expressed in a large number of researches on bilingual education around the world. This was also a period, on the one hand, of a growth of a large number of research projects on the speech communities of Britain, and on the other a flowering of language awareness on the part of schools and teachers. Although this was a period when a large proportion of educational research money was spent on studying children who spoke a language or languages other than or in addition to English (Alladina, 1986: 353-4), very few of the benefits reached the children, the schools and the teachers working there. With hindsight, it would seem that the research projects helped to provide information not to promote the maintenance of languages in Britain but to bring about their demise. Even then the information was incomplete and sketchy. For example, after a decade of data collection on language use in Britain, as late as in 1984, the Department of Education and Science could only give an estimate that the number of pupils aged between 5 and 16 years whose first language was not English was 'somewhere between 375,000 and 500,000' (Commission of the European Communities, 1984). To this day, there has not been a comprehensive survey of languages and language use in Britain. In this, the biennial language censuses carried out by the Inner London Education Authority (ILEA) since 1979 are a valuable source of information. As a part of the education reforms in Britain, however, the ILEA, the largest education authority in the world, was dissolved in 1989 and data collection of this kind has therefore ceased. The ILEA language censuses give an indication of the variety of languages spoken in the inner city schools. In the last of the language censuses carried out by the ILEA in 1988, a total of 184 different languages were identified among ILEA pupils. Sixteen of the languages had more than a thousand speakers each: Bengali (20113), Turkish (4625), Chinese (4242), Gujerati (3904), Urdu (3821), Spanish (3095), Panjabi (3062), Arabic (3427), Greek (2241), French (2342), Yoruba (2941), Portuguese (1939), Italian (1619), Vietnamese (1573), Twi (1111) and Tagalog (1102), (ILEA, 1989). A decade after the Bullock Committee Report, The Swann Committee Report (1985) on the education of ethnic minority children was published. Although it recognised the existence of linguistic diversity (and racism) in British schools and the usefulness of home language in the acquisition of Eng-

< previous page

page_58

next page >

< previous page

page_59

next page > Page 59

lish in the early years, it sounded the death knell for mother-tongue maintenance by recommending that: Mainstream schools should not seek to assume the role of community providers for maintaining ethnic minority community languages (p. 427). Anti-racist education By this time there was a strong feeling among teachers and parents of South Asian origin that racism was not enabling their children to succeed in the British education system. A sizeable number of teachers and educationists were of the opinion that to bring about equal opportunities in education, not only multicultural education but anti-racist strategies were necessary. Many education authorities supported this view and formally instituted anti-racist educational policies (see, e.g., ILEA, 1983, Berkshire Education Authority, 1984). In the area of language education, bilingual teaching strategies and multilingual approaches to education began to be developed by many schools. This also necessitated recruitment and training of teachers who shared the language background of the children in schools. It should be noted, however, that the five phases of responses did not appear evenly in different parts of Britain. For example, when language centres and ESL approaches had been clearly abandoned by many education authorities and they had issued anti-racist educational policies, in 1986, a local education authority in the north of England was accused by the Commission for Racial Equality of discriminating against children of Asian origin by having a policy of screening these children on entry to school and identifying a large number of these children as needing separate English as a Second Language (ESL) instruction. The problem was compounded by the fact that ESL units were set up which kept these children away from mainstream classrooms (CRE, 1986). Equal opportunities in education may have been accepted by many teachers and educationists. However, the majority of the British population was either indifferent, hostile or ignorant of the idea. This situation has made it easy for the government to bring about major changes in education generally and specifically in the areas of racial and linguistic equality. An education system has a crucial role to play in the life or death of a language. It is interesting to note also that with universal compulsory education, the state has greater control over language reproduction. The implications of this are not confined to young children and their education because language reproduction, maintenance and recognition have significant repercussions on

< previous page

page_59

next page >

< previous page

page_60

next page > Page 60

participation of all in the political processes in a society. Although there is a history of educational policies and recommendations to eradicate the regional languages of Britain (see Edwards, 1991; Hancock, 1991; Hickey, 1991; MacKinnon, 1991). There were no clearly stated aims and objectives in response to languages that were establishing themselves in Britain. The Education Reform Act (ERA) of 1988 is a new departure in that it puts a legal obligation on schools and local education authorities to provide a particular kind of education which will have certain repercussions on the life and vitality of the various languages of Britain. To discuss the fate and future of these languages, information on their use and distribution is essential. Data on South Asian languages in Britain The decennial censuses of Britain do not give any details of racial, ethnic or linguistic backgrounds. The 1981 Census of Britain included a question on the country of birth of a respondent but this information throws little light on language use in the home and social networks. Head counting of people from various ethnic groups in Britain, particularly of people from the "New Commonwealth" is a taboo area in Britain because politicians from all shades of politics have played number games on immigrant population to serve their political ends (Sivanandan, 1982). The debate on ethnic background question is chronicled by Lomas (1980). The 1991 Population Census of Great Britain will elicit information on the ethnic origin of respondents, but still the question of language use will remain unaddressed. In absence of a comprehensive language use survey, education policy makers have to rely on linguistic sketches that have been drawn by researchers in Britain. In this, the Linguistic Minorities Project (LMP) based at the University of London Institute of Education, between 1979 and 1983, is an important source. The LMP carried out in-depth language use surveys among samples of adults in five urban areas. The ILEA Language Censuses mentioned above, are another important source. Several other educational authorities in Britain have carried out language censuses among their school populations. However, they still remain sketches with varying degrees of reliability. There are five major factors that introduce a high degree of unreliability in these figures.

< previous page

page_60

next page >

< previous page

page_61

next page > Page 61

1. All the language surveys rely on self-reports. In addition to the inherent unreliability of self-reports, some specific factors obtaining in Britain add a greater degree of unreliability to this data. The socio-psychological climate in Britain leads school children to abandon, conceal and/or disown their linguistic and cultural backgrounds (Alladina, 1985). In this light, the suspiciously low figures of 32 speakers of Konkani and 15 of Marathi, as examples, out of a population of 70221 in the ILEA 1988 Language Survey need to be statistically verified. 2. The elicitors themselves add to the degree of unreliability. As Labov (1978) observed in his study of Afro-American respondents, "since the adult is negro man raised in Harlem, who knows this particular neighbourhood and these boys very well. He is a skilled interviewer who has obtained a very high level of verbal response with techniques developed for different age levels, and he has an extraordinary advantage over most teachers and experimenters in these respects. But even his skills and personality are ineffective in breaking down the social constraints that prevail here." Compared to this, the degree of unreliability introduced by classroom teachers operating under many disadvantages can be imagined. 3. In surveys of the type conducted by LMP, the sample strategy itself introduces a degree of unreliability. The built-in unreliability of samples extracted from telephone directories and electoral rolls is well-documented. The question of identifying names with the language of the home also has its own problems. Attempts have been made in Britain to use computers programmed to extract names from electoral rolls and identify them with possible home language (LMP, 1982). The margin of error in this exercise can be up to 15%, not in the identification of name with home language but in the first step of identifying name with an ethnic group. As an illustration, "Shah" is quite a common Gujerati name. But it can also be a family name of a Panjabi, Urdu or Farsi speaker. If the computer is programmed to extract names using the first four letters, as was done by the LMP, then the "wrong language response" in that sample introduces an immense margin of error. For example, using a name like Shah again, the London telephone directory lists thirty different family names beginning with S-H-A-H. The possible languages that could be related to these

< previous page

page_61

next page >

< previous page

page_62

next page > Page 62

names are: Gujerati, Urdu, Panjabi, Farsi, Armenian or Turkish. Furthermore, an example is given in Alladina (1985) of how three different workers extracted 200, 495 and 714 Gujerati names working from the same electoral roll. The first compiler had not taken into account any Muslim or Parsee Gujerati names, the second list had included some Muslim names whereas the third list had taken into account all the Hindu, Parsee and five major groups of Muslim Gujerati residing in the area. Each group has its own patterns of language use, maintenance and loyalties. For a more reliable linguistic picture, each of these groups has to be represented in a sample of Gujerati speakers. 4. The way languages are classified and defined also leads to unreliability of data. To focus on South Asian languages, the 1988 ILEA Language Survey, for example, gave the following information: Bengali (20113), Gujerati (3904), Urdu (3821), Panjabi (3062), Hindi (593), Tamil (398) and Singhalese (167). There were 274 speakers of other languages of India, Pakistan, Bangladesh and Sri Lanka. In this, for example, figures for Gujerati include speakers of Kachchi. No distinction is made between speakers of Panjabi who are of Hindu, Muslim or Sikh origin. It will be appreciated that each of these groups has its own dynamics of use, maintenance, loyalties and writing traditions. Similarly, the number of Urdu speakers does not clarify the situation of the Urdu/Panjabi bilinguals. The fact that the ILEA Language Surveys give no information on levels of literacy is a shortcoming that is additional to the discussion on unreliability. Information is elicited by teachers. Although detailed instructions are given to teachers regarding language classification and the sensitivity required when eliciting personal information, in the ILEA Survey of 1979, 1129 students were returned as speaking ''Indian" or "Pakistani". It can only be speculation to discuss if the speakers described their languages as such or the teachers, due to their ignorance of language names and language groups, used those terms. The fact that the ILEA 1981 Language Survey returned only 43 students as speaking "Indian" or "Pakistani" indicates that there may have been an increased awareness among teachers who elicited and recorded information. Thus, discussion of South Asian languages in Britain takes place on the basis of sketchy information with a high degree of unreliability. It is therefore not surprising that the recent education

< previous page

page_62

next page >

< previous page

page_63

next page > Page 63

reforms in Britain have put the clock back in response to the languages education in the country. The Education Reform Act of 1988 The Education Reform Act of 1988 provided for the establishment of a national curriculum comprising of core subjects (English, mathematics and science) and other foundation subjects to be taught to all pupils of compulsory school age in the state-maintained schools. In the primary education sector, no allowance is made for the teaching of languages other than English. An exception is made for Welsh in Wales where, in the Welsh-speaking areas, Welsh is the fourth core subject or one of the foundation subjects in the English-speaking areas of Wales. The Statutory Order (HMSO, March 1990) recognises that the development of English and Welsh should be seen as mutually supportive; transfer of skills, knowledge and understanding can take place from one language to another; and the similarities and differences between the two languages can be studied by children. However, no such recognition is given to the other languages of Britain. Although recognition of Welsh seems to be tied to the notion of "land and language" and territorial rights, a point that is elaborated by Bourne & Camaron (1988), even Welsh is not given the status of a school subject outside Wales. In the national curriculum, there is no room for languages other than English in primary education. Although the School Examinations and Assessment Council (SEAC), one of the government agencies set up under the ERA 1988, advises teachers to 'introduce home-tongue materials and invite home-tongue adult support into the classroom so that language development continues smoothly' (SEAC, 1990), 'language development' is understood to be English language development. The role of home language is thus seen to be to integrate children into the English-speaking classroom. In secondary schools, Bengali, Gujerati, Hindi, Panjabi and Urdu will be available to students after the school has also provided one of the 'working languages of the European Community'. Ultimately, it is a matter of student or parental choice. Moreover, if ERA has ensured that in the primary schools, home languages will only be used in the early years to facilitate the assimilation of children into the English-speaking classroom, it is quite likely that the number of pupils willing and able to take up the offer of the study of home languages at secondary level is going to decrease yearly. It would seem

< previous page

page_63

next page >

< previous page

page_64

next page > Page 64

that the sentiments expressed by the Bullock Committee Report in 1975 of a 'linguistically conscious nation in the modern world' are valid no longer. ERA 1988 has initiated another phase in the response to the multilingual reality of Britain. In an age when huge leaps have been made in the areas of information technology, simultaneous translations, speech synthesizers and in the understanding of thought and language, it is sad to see certain societies regressing to notions of 'one nation one language'. It has been claimed that equality is possible when there is only one language and that, in modern society, two languages are a nuisance, three languages are uneconomic and many languages are absurd. However, Pattanayak (1987) argues that "in the multilingual reality, many languages are a fact of existence, any restriction in the choice of language is a nuisance, and one language is not only uneconomic but absurd." References ALLADINA, S. (1985), Research methodology for language use surveys in Britain - A critical review. In: P. Nelde (ed.), Methods in Contact Linguistic Research, Bonn: Dumler. ALLADINA, S. (1986), Black people's languages in Britain - A historical and contemporary perspective. In: Journal of Multilingual and Multicultural Development 7/5, UK. BERKSHIRE EDUCATION AUTHORITY (1984), Education for Racial Equality. Royal County of Berkshire. BOURNE, J. & D. Cameron (1988), No common ground: Kingman, grammar and the nation. In: Language and Education 2(3), UK. BULLOCK, Sir A. (1975), A language for life. London: HMSO. COMMISSION FOR EUROPEAN COMMUNITIES (1984), A Report from the Commission to the Council on the Implementation of Directive 77/486/EEC. Brussels: COM(84) 54. COMMISSION FOR RACIAL EQUALITY (CRE) (1986), Teaching English as a second language, Report for a Formal Investigation in Calderdale Local Education Authority. London. EDWARDS, V. (1991), The Welsh speech community. In: S. Alladina & V. Edwards, Multilingualism in the British Isles, Volumes 1 and 2, London: Longman Linguistic Library. FRYER, P. (1984), Staying Power, the History of Black People in Britain. London: Pluto Press. HANCOCK, I. (1991), The Romani speech community. In: S. Alladina & V. Edwards, Multilingualism in the British Isles, Volumes I and 2, London: Longman Linguistic Library.

< previous page

page_64

next page >

< previous page

page_65

next page > Page 65

HICKEY, T. (1991), The Irish speech community. In: S. Alladina & V. Edwards, Multilingualism in the British Isles, Volumes 1 and 2, London: Longman Linguistic Library. HMSO (1990), Teaching of Welsh in Wales, Statutory Order, March. ILEA (1983), Anti-Racist Policy Statement. Inner London Education Authority. ILEA (1989), 1988 Language Survey, Inner London Education Authority, Rs 1261/89. London. KILLIAN, L. (1979), School Bussing in Britain: Policies and Perceptions. In: Harvard Education Review 49(2). LABOV, W. (1978), The design of a sociolinguistic research project. In: D. Pattanayak (ed.), Papers in Indian Sociolinguistics, India Central Institute of Indian Languages. LMP (1982), The geography and demography of South Asian languages in England: Some methodological problems. In: Linguistic Minorities Project Working Paper 2, London: University of London, Institute of Education. LOMAS, G. (1980), Race, Colour, Ethnicity: The collection of data. In: New Community VIII/1-2, London: Commission for Racial Equality. MACKINNON, K. (1991), The Gaelic speech community. In: S. Alladina & V. Edwards, Multilingualism in the British Isles, Volumes I and 2, London: Longman Linguistic Library. MOOREHOUSE, G. (1983), India Britannica. London: Paladin Books. OUP (1887), Record of the Establishment of the Indian Institute in the University of Oxford. Horace Hart, Printer to the University. PATTANAYAK, D. (1987), Multilingualism and Multiculturalism: Britain and India, Occasional Paper No. 6. London: Centre for Multi-cultural Education, University of London, Institute of Education. RAMPTON COMMITTEE REPORT (1981), The West Indian Children in our Schools. London: HMSO. SEAC (1990), A guide to teacher assessment, Pack C, School Examination and Assessment Council. Department of Education and Science, UK. SIVANANDAN, A. (1982), A different hunger, writings on black resistance. London: Pluto Press. SWANN, Lord (1985), Education for all. London: HMSO. TROYNA, B. & J. Williams (1986), Racism, education and the state. London: Croom Helm. VISRAM, R. (1986), Ayahs, princes and lascarsc. London: Pluto Press.

< previous page

page_65

next page >

< previous page

page_ii

next page >

< previous page

page_67

next page > Page 67

3 A bilingual perspective on Turkish and Moroccan children and adults in the Netherlands Guus Extra & Ludo Verhoeven It is a common misconception that multi-ethnicity and multilingualism as a result of immigration are recent phenomena in Dutch society. As in other industrial European countries, the patterns of migration to the Netherlands at any time seem to correspond with the country's relative economic prosperity. Lucassen & Penninx (1985) studied these patterns from a historical perspective. In the seventeenth or so-called "Golden" century, about 10% of the Dutch population originated from abroad. At that time, the Netherlands became a refuge for socio-economic elites (e.g., Jewish-Portuguese merchants), who mainly immigrated for religious or political reasons. Their expertise and capital were received in the Netherlands with great hospitality and tolerance. From 1600 to 1800 the immigrant population fluctuated between 6% and 10%. Less people immigrated at times when the economic climate in the Netherlands was less prosperous, e.g., at the turn of the twentieth century or shortly after World War II. Most people immigrated because of economic depression in their homeland and most of them had a relatively low socioeconomic status. A well-known group were the families of Chinese harbor workers and seamen in the Rotterdam area, who constituted the largest Chinese community on the European continent between 1920 and 1940. With the economic revival of the sixties and seventies, Dutch companies started to recruit cheap labour from less developed Mediterranean countries. As in most other Western and Northern European countries, this recruitment was done with the political and financial support of the national government. The early pattern of economic migration was followed by a pattern of social migration. Many Mediterranean families joined their husbands and fathers who lived in exile abroad. Due to economic prosperity in the Netherlands and changing political conditions abroad, a similar pattern of social migration held for families

< previous page

page_67

next page >

< previous page

page_68

next page > Page 68

originating from (former) Dutch colonies, like the Dutch Indies, Surinam, and the Dutch Antilles. In the present chapter processes of language change among ethnic minority groups in the Netherlands will be dealt with. First of all, we present basic demographic data. Derived from these data we will focus on the Turkish and Moroccan community who belong to the largest ethnic minorities. In addition, we will take into account the overall characteristics of ethnic minority research in the Netherlands. Special attention will be given to a research program focusing on processes of language change over time among Turks and Moroccans living in the Netherlands. Table 1 Main non-indigenous groups in the Netherlands, January 1990 1. Ex-colonial territories 214.000 Surinam 70.300 Dutch Antilles 40.000 Moluccas 2. Mediterranean countries 185.500 Turkey 143.700 Morocco 17.000 Spain 16.100 Italy 12.400 Yugoslavia 8.100 Portugal 4.600 Greece 2.500 Tunesia 2.100 Cape Verde 60.000 3. Chinese

< previous page

4. Refugees Vietnam Pakistan Ghana Hungary Turkish Christians Poland Ethiopia Sri Lanka Czechoslovakia Chile 5. Countries with similar SES Germany Great-Britain Belgium USA France Japan Ireland Austria Canada Australia

5.100 4.200 4.600 4.100 3.000 3.500 3.100 2.400 2.000 2.000 41.000 36.400 23.200 11.400 8.400 3.900 3.400 2.900 2.700 2.200

page_68

next page >

< previous page

page_69

next page > Page 69

General demographic statistics The demographic statistics presented in this section derive from recent longitudinal trend studies of Ankersmit et al. (1989, 1990) and Muus (1989, 1990). The former studies are based on longitudinal cooperation with the Dutch Central Bureau of Statistics, whereas the latter are part of a continuous reporting system on ethnic minority groups in the OECD member states. Table 1 gives a survey of actual or estimated figures (n>1500) for the main non-indigenous groups in January 1990; at that time, the indigenous Dutch population amounted to nearly 15 million inhabitants. Within the total non-indigenous population of the Netherlands, the following main groups can be distinguished: 1. Immigrants from former Dutch colonies. Surinamese may be speakers of Sranan-Tongo, Sarnami, Hindustani, Javanese, Hakka, Indian languages or Surinamese Dutch. Antillians may be speakers of Papiamentu, Dutch or English. Both in Surinam and on the Antillian Islands, Dutch has so far preserved the status of official language. Within the group that hails from the former Dutch East Indies (the present Republic of Indonesia), the Moluccans take a special position from a cultural, linguistic and religious perspective, and because many of them "repatriated" involuntarily to the Netherlands in 1951. Moluccans may speak Moluccan-Malay, High-Malay, Melaju-Sini or Dutch. 2. Labour migrants and their families from Mediterranean countries. Most of them originate from Turkey and Morocco. Turks may speak Turkish, Kurdish, Arabic or Azeri, whereas Moroccans may be speakers of Moroccan-Arabic or mutually different Berber varieties like Tarifit, Tamazigt or Taselhit. 3. Chinese immigrants from countries like China, Taiwan, Vietnam, Malaysia, Singapore or Hongkong. Chinese may speak Mandarin-Chinese, Cantonese, Whenzhou or Shanghai dialect. 4. Political refugees from Eastern-Europe, Asia, Africa and South-America. 5. Immigrants from countries with a socio-economic status comparable to the Netherlands. Immigration is often an initial step in the emergence of minority groups. Most groups of immigrants and their descendants in the Nether-

< previous page

page_69

next page >

< previous page

page_70

next page > Page 70

lands are nowadays referred to as 'ethnic minorities', although ethnicity does not necessarily imply a minority positionnor does the reverse apply. Until now, the Dutch government has refrained from officially defining the concept of 'ethnic minority'. In fact, the governmental concept only refers to an exhaustive list of ethnic groups which have a disadvantaged socio-economic status and which are eligible for financial support to reduce these disadvantages. Due to this deficit perspective, the Chinese are excluded from access to special educational facilities (e.g., instruction in Chinese). Apart from the biased primacy of socioeconomic rather than ethno-cultural status, there are increasing difficulties over time in identifying the relevant target groups. The most widely used criteria, i.e., nationality and country of birth, show strong erosion because of naturalizations and births in the Netherlands (cf. Ankersmit et al., 1990: 12). In Table 2, we present population trends, including migration and birth effects, for the earlier mentioned Mediterranean groups between 1985 and 1989 (cf. Muus, 1989). Table 2 Major Mediterranean groups in the Netherlands Source country Change from 1985-1989 Turkey + 13.5% Morocco + 25.0% Spain -16.0% Italy - 22.2% Yugoslavia - 0.8% Portugal + 2.2% Greece + 6.4% Tunesia - 4.8% Cape Verdian Islands - 24.0% There are significant differences in the legal options for (re)migration between citizens of EC vs. non-EC countries. Since 1983, the migration surplus from the Mediterranean countries mentioned in Table 2, increased from -1.995 to 15.760 in 1988. Tables 1 and 2 show that Turks and Moroccans belong to the largest Mediterranean groups in the Netherlands, who at the same time show the strongest increase over time.

< previous page

page_70

next page >

< previous page

page_71

next page > Page 71

Turks and Moroccans in the Netherlands In the past few years, the number of Turks and Moroccans in the Netherlands has increased by net immigration and natural growth. At this moment, family reunification (including marriage) is almost the only way for Turks and Moroccans to immigrate legally into the Netherlands. For both groups, a significant increase of immigration and birth figures is expected in the nineties, whereas remigration figures will probably remain low. In retrospect, we offer in Table 3 a closer look at changes in birth, migration, and nationality over time, derived from Ankersmit et al. (1990: 19). Table 3 Changes in birth, migration and naturalization saldo Turks

Moroccans

Birth saldo 1986 1987 1988

3.943 4.520 5.048

3.849 4.139 4.337

Migration saldo 1986 1987 1988

3.232 5.808 6.839

5.005 5.807 6.878

- 1.473 -1.398 - 823

- 1.477 -1.437 - 1.187

Naturalization saldo 1986 1987 1988

Table 3 shows that the growth of Turkish and Moroccan populations in the Netherlands is as yet primarily caused by migration effects, and not by birth effects. The observed decrease in naturalizations (Dutch citizenship) in both groups is not an effect of decrease in requests, but in acknowledgements of such requests. Derived from the figures presented in Table 1 (January 1990), both Turks and Moroccans show an underrepresentation of females (45.6% and 44.0% respectively), and an underrepresentation of higher age groups. In 1988, more than 20% of all Turks in the Netherlands were under the age of 10; for Moroccans, this figure was even 30%. Family

< previous page

page_71

next page >

< previous page

page_72

next page > Page 72

size, however, showed large variation. The number of interethnic marriages with Dutch partners remained low. Most ethnic minority groups, including Turks and Moroccans, are concentrated in the four largest cities of the Netherlands (Amsterdam, Rotterdam, The Hague and Utrecht). For Turks, the percentage was 37% in 1988, for Moroccans it amounted to 49%. Although there are no precise statistics on registered unemployment of ethnic minority groups in the Netherlands, there is a continuous decrease in employment of Mediterraneans vs. an increase in employment of Dutchmen. These trends can partly be explained by unfavourable work qualifications and high regional unemployment rates. However, additional factors contribute to these divergent patterns, in particular discrimination on the labour market, both in terms of access and dismissal procedures. In 1989, 27.000 out of 63.000 Turks and 24.000 out of 41.000 Moroccans who belonged to the professional population in the age range of 15-64 years, were unemployed. The influx of ethnic minority children in elementary schools in the four largest cities of the Netherlands is presently about 40% and will increase to more than 50% in the year 2000. However, there is a large variation in percentages between different schools. In the country as a whole, the number of elementary schools with more than 50% of ethnic minority children increased by 70% between 1980 and 1984. It is a widespread phenomenon that ethnic minority students are overrepresented in lower types of secondary or further education and underrepresented in higher types. A nationwide comprehensive secondary school system does not (yet) exist in the Netherlands. Moreover, there is a strong increase of ethnic minority children in special instead of regular education and a strong "drop-out" rate in all types of secondary or higher education. In 1988, almost 60% of the Moroccan students who have participated in any of these types of education, left school without a diploma; the corresponding drop-out rate for Turkish students was 50%. Statistics on education are yearly collected at local schools and centrally disseminated by the Ministry of Education. However, for various school types no reliable data on participation of ethnic minority students have been available since the school year of 1986/1987. In Table 4 we present figures of Turkish and Moroccan full-time students in a variety of educational settings, with figures of Dutch students as a reference group.

< previous page

page_72

next page >

< previous page

page_73

next page > Page 73

Table 4 Participation in education of three ethnic groups Dutch Turkish Moroccan Elementary educ. 1.331.318 33.688 33.753 1988/1989 (BO) Special education 99.894 1.910 2.321 1987/1988 (S(V)O) General sec. educ. 754.732 5.503 3.757 1986/1987 (AVO) Lower voc. educ. 311.485 7.041 6.378 1986/1987 (LBO) Second. voc. educ. 284.345 1.136 705 1986/1987 (MBO) Higher voc. educ. 157.619 235 67 1987/1988 (HBO) University educ. 175.854 239 122 1987/1988 (WO) For the three groups mentioned in Table 4, the proportion of LBO students per 100 AVO students in 1986/1987 was as follows (cf. Ankersmit et al., 1989: 134): 41 Dutch LBO students, 128 Turkish LBO students, and 170 Moroccan LBO students. Explanations of the low AVO representation of Turkish and Moroccan students are attributed to such factors as low socioeconomic status and educational support of the parents, low level of proficiency in Dutch and contacts with Dutchmen, bad housing conditions for homework, maladjustment of schools, teachers, curricula and available funds to the needs of the students, and low teacher expectancies (cf. Roelandt & Veenman, 1990). Most of these factors show an intricate pattern of dependency and interaction. However, in spite of a strong underrepresentation of Turkish and Moroccan students in AVO, MBO, HBO, and WO, there is a steady decrease of this underrepresentation over time. This holds especially for Moroccan students and, for both Turkish and Moroccan students, in secondary vocational education (MBO). Corrected for purely demographic changes, Table 5 contains trend indexes (in %) for degree of participation in three successive school years (cf. Ankersmit et al., 1989: 133):

< previous page

page_73

next page >

< previous page

page_74

next page > Page 74

Table 5 Trend indexes (in %) for degree of participation in three successive school years Turkish students Moroc. students Schooltype 84/85 85/86 86/87 84/85 85/86 86/87 AVO 118 124 128 135 LBO 100 96 83 100 102 88 MBO 100 119 155 100 130 211 HBO 100 109 108 100 124 117 WO 100 100 116 100 100 140 Research on ethnic minorities in the Netherlands Several extensive bibliographies on ethnic minority research in the Netherlands have been published, covering a wide range of topics in different periods of time. Ellemers et al. (1988), Koulen & Smit (1988), and ACOM (1989) focused on the periods of 1945-1986, 1985-1986, and 1984-1989 respectively. The ACOM bibliography for the first time contained a separate section on ''language", with a total number of 84 references. From these bibliographies it becomes clear that there is a fast growing research interest in the field of ethnic minorities. We will pay attention to some general characteristics of ethnic minority research in the Netherlands and to language-related research in particular. General characteristics Research on ethnic minorities in the Netherlands is carried out on a multitude of demographic, legal, economic, social, cultural, linguistic and educational themes. Given the variety of these themes, there are obviously large differences in approach, both from a theoretical and methodological point of view. Research projects are funded by local universities (1), by the Dutch Science Foundation NWO (2), and by national, regional, or local agencies, in particular ministerial departments and large cities (3). However, both the first incentives and the spectacular growth of research occurred in reverse order. As yet, there are few large-scale university-based research programs with a focal interest in the status of ethnic minority groups in the Netherlands. Owing to the large share of governmental contract research, there is a

< previous page

page_74

next page >

< previous page

page_75

next page > Page 75

clear influence of governmental agencies on the type of ethnic minority research. Much of this research is undertaken by social scientists who are successively contracted by various governmental bodies. This contract dependency of research efforts has a number of negative effects, and invariably leads to both an overrepresentation and underrepresentation of certain research questions (cf. Extra & Vallen, 1987). Especially short-term, policy-relevant, or even policy-supportive research has been stimulated. Only in recent years, more independent university-based research has gradually increased. However, long-term research programs, with more fundamental questions and an international orientation still have a modest share. Finally, the Dutch Science Foundation NWO has started to financially support university-based research on ethnic minorities in the Netherlands in the domains of (1) culture, (2) language, (3) law, and (4) religion (in the latter domain with a focus on the status of Islam in the Netherlands). For each of these domains, interuniversity cooperation has been promoted on the basis of matched funding in order to enhance the total research effort. Another characteristic of ethnic minority research in the Netherlands is that most projects are carried out by Dutch researchers. However, both for the conceptualization of research questions and for the collection and analysis of data, in-group expertise can be most valuable or even indispensable. Unfortunately, the number of academically trained in-group researchers is very small in the Netherlands. Ethnic minority students are underrepresented at Dutch universities, both in terms of enrollment and achievement. An improvement of these conditions is only to be expected from special affirmative action programs. However, there are as yet few such initiatives. Language-related research Language-related research on ethnic minorities in the Netherlands has shown a fast increase in the late eighties. Both in the areas of psycholinguistic, sociolinguistic and educational research, the focus has been on second (L2) rather than first (L1) language varieties, on children rather than adults, and on Turkish and Moroccan rather than other ethnic groups. English-based overviews of language-related research have been given by Extra & Vallen (1989, 1988, 1985), Extra (1990a, 1990b), and Extra & Verhoeven (1991). There is a remarkable asymmetry between the available evidence on L2 use of children vs. adults. Whereas the former is commonly

< previous page

page_75

next page >

< previous page

page_76

next page > Page 76

based on what children actually do in spontaneous or elicited tasks, the latter is often limited to self-reports of informants or impressions of researchers, derived from written questionnaires or oral interviews. As a result, most data on adult L2 use are indirect, potentially biased, and difficult to evaluate. Moreover, most studies on adult L2 use did not focus primarily on language, but on more or less related social, cultural or economic issues. Similar comments can be made about presented data on language choice or language preference; these data are commonly based on self-reports of children or adults or on other-reports of, e.g., teachers, parents or researchers, and only rarely on actually observed patterns of language behaviour. Given the social and scientific pre-occupation with issues of second language acquisition, the status and use of ethnic community languages have received far less attention. Moreover, and contrary to traditional English-dominant immigrant countries like the USA, Canada or Australia, there is no tradition of collecting census data on home language use in the Netherlands. Consequently, no representative statistics are available on intergenerational processes of ethnic community language maintenance, shift, or loss. The biased scientific interest in the Netherlands in the use of the dominant second language instead of the dominated first language by ethnic minority groups is first of all an accurate reflection of the vast American literature about bilingualism. The number of American studies on the acquisition and use of Spanish (the most prominent ethnic minority group language in the United States) is extremely limited in comparison with the existing literature on the acquisition and use of L2 English. Another typical factor is the traditional philological orientation of research on languages such as Turkish, Arabic, Chinese or Malay. The fact that in the context of Dutch society these languages are spoken 'here and now' has not resulted, unfortunately, in any substantial modification of research priorities. The philologists' lack of scientific interest stems most commonly from a combination of low esteem for the study of present-day spoken language varieties of minority groups with a low socio-economic status on the one hand, and from ignorance of suitable methods for empirical research on the other. However, the commitment of expert linguists with an in-depth knowledge of the language varieties under consideration is indispensible for a fruitful development of this research area. Within the research program of the Tilburg Research Group on Language and Minorities, a broad versus narrow concept of "language"

< previous page

page_76

next page >

< previous page

page_77

next page > Page 77

and "minorities" is adhered to. Whereas the concept of "minorities" is focused on ethnic minority groups within Dutch society, the concept of "language" relates to both majority and minority language use. From a social perspective, language plays a crucial role in the disadvantaged position of ethnic minority groups in society. L2 proficiency is a necessary (yet insufficient) condition for access to and success within the educational system and on the labour market. At the same time, L1 is subject to a broad spectrum of positive and negative attitudes of both minority and majority groups. The study of the interaction between socio-cultural systems of majority and minority groups is especially relevant in the domain of verbal communication. Major questions within this field are: • Which language varieties are used for between-group and within-group communication? • How do dominant and dominated groups in a multi-ethnic society value dominant and dominated language varieties? • Which changes undergo these varieties and values of language use over time? The research program takes into account both a dynamic and a crosslinguistic perspective on language behaviour of ethnic minority groups in the Netherlands. From a dynamic perspective, the focus is on various processes of L1/L2 change over time, in particular on: • acquisition and use of Dutch as a second language; • acquisition and use of ethnic minority languages; • code-switching, language shift, and language attrition. In this process-oriented approach, patterns of language behaviour are studied at the levels of individuals and social groups. Moreover, structural and temporal determinants of language change are taken into account (cf. Extra, 1990a). Structural determinants, focusing on how language varieties change over time, are derived from target (L2) or source (LI) language principles, and from the interaction between both. Temporal determinants, focusing on how fast language varieties change over time, are derived from the social, economic, legal, cultural, religious, demographic or linguistic distance between speakers of minority languages and speakers of the majority language. From a cross-linguistic perspective, variation in age or generation is taken into account, and variation in source language background. Research projects are derived from one or more of the following research domains (cf. Extra & Vallen, 1985):

< previous page

page_77

next page >

< previous page

page_78

next page > Page 78

(a) Use of Dutch as a second language Within this domain attention is paid to the use of Dutch of, towards and between non-native children, youngsters and adults. Major questions are: • What are the characteristics of varieties of Dutch of non-native speakers in different contact situations with native speakers of Dutch? • Which factors determine the structure or order of the acquisition of L2 Dutch? • Which factors determine the pace or success of the acquisition of L2 Dutch? • How do native speakers of Dutch modify their language use in addressing non-native speakers and in what kind of conditions do such modifications occur? • What are the characteristics of L2 Dutch of non-native speakers in contact situations with other non-natives having different L1 backgrounds? The first question relates to ethnic minority groups with a short or long immigration history in the Netherlands. In the latter case, the focus is on processes of linguistic "fossilization". The fourth and fifth question relate to research into "foreigner talk/register" and "lingua franca" respectively. (b) Use of ethnic minority languages The following questions are addressed: • Which (varieties of) ethnic minority languages are used in the Netherlands in different types of interaction? • How are these varieties acquired in a dominant second language environment? • Which processes of L1-L2 shift and L1 attrition occur over time? • Which intrasentential and intersentential varieties of L1-L2 code-switching and code-mixing occur under which conditions? The first question relates to both spoken and written varieties. With respect to the second question, comparative studies on the acquisition of the same varieties in the home countries are done. The third question makes it possible not only to distinguish between intra- and intergenerational processes, but also and again between structural and temporal determinants of these processes.

< previous page

page_78

next page >

< previous page

page_79

next page > Page 79

(c) Language and education Both macro-objectives (including school career research, with special attention to LI/L2 proficiency of ethnic minority children, youngsters an adults) and micro-objectives (including the structure of verbal interaction in educational settings) can be emphasized. With respect to L1/L2 instruction, research can focus on: • Educational needs and objectives; • Selection and grading of learning units; • Didactic approaches (including the use of audiovisual media); • Process- and product-evaluation (including instrument development); • Educational policy and legislation (also in an international comparative perspective). (d) Language attitudes Within this domain, language attitudes of both native and nonnative speakers of Dutch (including pupils, teachers, parents and governmental institutions) are emphasized: • Which attitudes can be evidenced with respect to the use, acquisition, maintenance, shift, attrition and switching of ethnic minority languages? • Which attitudes can be evidenced with respect to the acquisition and use of Dutch as a second language? • Which attitudes can be evidenced with respect to the teaching in/of Dutch as a second language and the teaching in/of ethnic minority languages? • In which direction do these attitudes change over time? • Which relationship can be evidenced between these attitudes and proficiency in Dutch on the one hand, and proficiency in ethnic minority languages on the other? (e) Language and media In this domain the emphasis is on: • Stereotypes of ethnic minority groups and ethnic minority languages in the mass media (oral and written press, radio and television, educational materials); • Access of ethnic minority groups and ethnic minority languages to these media.

< previous page

page_79

next page >

< previous page

page_80

next page > Page 80

Within these five domains various research projects are currently being carried out by members of the Tilburg research group. Most projects centre on the acquisition of Dutch as a second language. Other projects focus on the acquisition, shift, attrition or switch of ethnic community languages, or they take into account the acquisition of both minority and majority language proficiency. In addition, some projects focus on language acquisition in educational contexts and/or they take into account language attitudes or the use of mass media. Given the earlier mentioned under-representation of research on ethnic community languages, a special extension of the program in this particular area of research was developed in cooperation with the Dutch Science Foundation NWO. This NWO-supported program will be the focus of our next section. Language change among Turks and Moroccans in the Netherlands As part of the research program described in the former section, a significant extension was launched in 1989. The program extension, which will span over five years, is supported by the Dutch Science Foundation (NWO) and is carried out in cooperation with the University of Nijmegen. The focus of research is on processes of language change over time among the two largest Mediterranean groups in the Netherlands, i.e., Turks and Moroccans. We will report on the main objectives of the program and on the aims and design of its constituent research projects. Main objectives of the NWO-supported research program The program aims at crosslinguistic comparison by varying both the source language (Turkish vs Moroccan Arabic) and the age of informants (children vs. adults). Moreover, the program is based on a dynamic, process-oriented approach by taking into account: • first and second language development of children; • language switch, shift and loss of adolescents and adults. The first domain focuses on intragenerational processes of language change over time. Within the first domain, processes of bilingual devel-

< previous page

page_80

next page >

< previous page

page_81

next page > Page 81

opment are studied from an interdisciplinary point of view, combining insights from linguistic theory and developmental theory. From a formal linguistic point of view bilingual development is often defined as an "instantaneous" process in an "ideal" situation in which the child has at his disposal all of the principles and parameters of universal grammar and two sets of data necessary to fix those parameters. Given the obvious fact that languages are not acquired instantaneously, developmental theory must explain the various "delays" which characterize both L1 and L2 development, and the apparent difficulties children encounter in separating the two languages. In this domain two projects have been formulated: one focusing on preschool bilingualism and one focusing on bilingualism at school age. The second domain focuses on intergenerational processes of language change over time. Two projects have been set up in this domain. The first one focuses on structural patterns of code-switching among first and second generation immigrants. The study of code-switching serves a double interest. Firstly, the description and evaluation of code switching in the speech of immigrants in the Netherlands can be expected to yield a crosslinguistic contribution to current theories on constraints of code switching behaviour. Secondly, code-switching constitutes an interlinguistic level of language use that will eventually be a major source for an autonomous course of development that languages will follow in a diaspora context. The other project explores processes of language attrition and language shift in first and second generation immigrants. Attrition is defined as the loss of language skills over time within a generation, while shift is defined as the loss of language skills in the community due to imperfect learning in a second generation. Early bilingual development Background There is a rich tradition of studies on early bilingualism. From reviews by Taeschner (1983: 5-18) and McLaughlin (1985: 7298) it is clear that most of these studies share the following characteristics: • single or multiple case studies, • based on diaries kept by (one of) parents on • the simultaneous acquisition of two related European languages, • one language being spoken by one parent, the other being the dominant language in the wider community, • by children from a high socio-economic background.

< previous page

page_81

next page >

< previous page

page_82

next page > Page 82

The present project will focus on the bilingual development of Turkish children born in the Netherlands. These children form part of a third generation of immigrants who moved from rural sites in Turkey to industrialized areas in the Netherlands. Unlike most of the work that has been done so far, the simultaneous acquisition of two typologically unrelated languages will be studied by outside researchers in children of a low socio-economic background, L1 being the ethnic group language spoken by both parents, and L2 being the dominant language of the environment. Previous studies on early bilingualism give no reason to believe that the process of language acquisition in bilinguals and monolinguals is different in its basic features. The essential difference is that bilingual children are confronted with two sets of input and that they have the additional task of distinguishing two language systems. However, as has also been concluded by Grosjean (1982), the very complex process of learning to separate two different languages is still poorly understood. It remains unclear what sort of operating principles children use while acquiring two languages at the same time. It is also unclear under what conditions processes of language transfer and code mixing occur. Moreover, the studies that have been conducted so far were limited in their scope, given the fact that the languages under consideration were highly related. The analysis of children's speech data in two unrelated languages will offer new perspectives on the role of structural properties of these languages in the process of acquisition. In the present project the following research questions will be addressed: (1) What language is used under which circumstances by and towards the children? (2) How are the first and second language systems of children built up in the course of time? (3) Under what conditions and with what lexical and syntactic rules do processes of transfer and code-mixing occur? (4) In what way does the language input shape the child's grammar? The first question aims at insight into the interactional patterns the children take part in. This insight is needed in order to understand the linguistic background of the children. The second question is concerned with the developmental process itself. It will be investigated how the two language systems get differentiated and developed over time. The third question goes into the conditions under which processes of language transfer and code-mixing occur. Both functional and structural

< previous page

page_82

next page >

< previous page

page_83

next page > Page 83

properties of these processes will be studied. The final question concerns the role of language input. The assumption is made that the input may influence the process of bilingual development in two ways. First, the structural properties of each language may influence the perception and cognitive processing of language elements. Second, the specific input which is selected will determine the range of linguistic parameters to be acquired. Linguistic domains The following linguistic domains will be taken into account: clause structure, lexicon and word formation, reference to entities, reference to space and reference to time. These domains have proved to be highly significant in a large body of cross-linguistic studies on language acquisition in both children (cf. Slobin, 1985) and adults (cf. Perdue, 1984). Cross-linguistic attention will be given to typological differences between Turkish and Dutch in each of these domains. With respect to clause structure a linguistic analysis will be made of the children's spontaneous speech in two languages. The large typological differences in clause structure between Arabic and Turkish as a first language and Dutch as a second language leads us to expect that our data will furnish some new insights into the emergence of two grammars. In our analysis of clause structure a distinction will be made between core properties and peripheral or marked properties of the languages under consideration. As has been shown in work on first language acquisition, children start out with basic sentential phrase structures along with core complements through a principle of canonical mapping (Slobin, 1985; Hyams, 1988). Later on they sort out those aspects of complementation which are peripheral. It is also assumed that the distinction between core and peripheral properties will give better insight into the acquisition of inflectional morphology. On the basis of earlier research (Slobin & Bever, 1982), it can be hypothesized that inflections will be easier learned in Turkish, where inflection is a core property than in Dutch where inflection is a peripheral property. With respect to lexicon and word formation, children must acquire the diversity of options for coining words in the languages under consideration. In a number of studies the acquisition of word formation devices has been investigated. For such diverse languages as English (Clark, 1981; Clark, Hecht & Mulford, 1986) and Hebrew (Berman & Sagi, 1982; Walden, 1982), it was found that children at an early age start coining words in order to fill lexical gaps. Gradually, children give

< previous page

page_83

next page >

< previous page

page_84

next page > Page 84

up some of their early coinages in favor of words already established in the lexicon of the language. Clark has claimed that at least three general principles govern the course of acquisition of word formation rules. The first principle is semantic transparency stating that known elements with one-to-one matches of meaning and form are most transparent for constructing and interpreting new words. The second principle is regularization saying that children will use the same device everywhere to mark the same meaning. The third principle is productivity predicting that those word formation devices used most often by adults in word innovations are preferred in children's language for constructing new word forms. In the present study Clark's principles will be elaborated by investigating the bilingual development of Turkish children. The basic devices for word formation in Turkish and Dutch are very different (cf. Broeder et al., 1988), and the development of lexical variety and word formation devices of Turkish children in L1 and L2 will be explored. Moreover, it will be investigated to what extent the processes of lexical development in the two languages interact. Incorporation of lexemes of one language in word formation processes in the other language will be analyzed. In the domain of reference to entities two phenomena will be studied: pronominal reference and the distinction between definite and indefinite reference. In order to comprehend and use the contrast between definite and indefinite reference children must learn the rules for the establishment and maintenance of connected discourse. In the case a speaker does expect entities to be identifiable, noun phrases will be given the status of definite. In pronominal reference the deictic and anaphoric use of pronouns will be studied. Deictic devices often enter crucially into children's conversational discourse. As is shown by Karmiloff-Smith (1986), the basic referential markers in many languages are plurifunctional. In order to acquire the right distinctions for reference tracking, children must learn to map the right forms and functions. In the present study reference to entities will be studied by analyzing samples of connected discourse in the children's speech. Given the fact that Dutch and Turkish vary considerably in their conceptual notions and linguistic devices for reference to entities, it will be investigated how children learn to use the coding devices for pronominal reference and for the expression of definite vs indefinite reference in their first and second language systems over time. The study of reference to space will focus on the use of spatial concepts which are relevant to the expression of location or motion.

< previous page

page_84

next page >

< previous page

page_85

next page > Page 85

According to Fillmore (1982: 37), the subject of spatial reference has three natural subtopics: systems of demonstratives, pre-or postpositional and adverbial devices for constructing locative expressions and motion verbs. With respect to the acquisition of devices for spatial reference it can be hypothesized that cognitive maturation will play a significant role. However, languagespecific factors can play a role as well. When referring to the location or motion of entities in space, languages may differ both in their use of spatial concepts and in the ways in which spatial concepts are encoded (e.g., Johnston & Slobin, 1979). Dutch and Turkish differ considerably, both in underlying spatial concepts and in the form of the devices they include. In the present study it will be examined how children learn to use the distance features of spatial deixis and the expression of spatial relations in their first and second language over time. In exploring reference to time three basic categories of temporality in language will be distinguished: temporal relations, aspect and internal temporal features. Temporal relations refer to the location of events in relation to a given reference time. Aspect refers to the various perspectives that can be taken towards an event, e.g., perfective vs. imperfective. Internal temporal features refer to quasi-objective time characteristics of an event, such as durativity or transformativity. Given the substantial differences of the temporal systems in the languages under consideration, the simultaneous acquisition of linguistic devices for temporal reference can be thought of as highly relevant. Within the temporal domain it will be investigated what operating principles underlie the synthetic and analytic devices for temporal marking in the children's first and second language in the course of time. Design The present study will have a longitudinal design, based on an intensive data collection with a small number of informants. The main focus will be on the collection and analysis of spontaneous speech data. In-depth analyses of a substantial data base per informant will be conducted within a variety of linguistic domains. Besides, a small number of experimental tasks will be administered to test several aspects of the children's language comprehension. Data collection will take place with informants who are exposed to different conditions of L1/L2 input. Moreover, for reasons of comparison, language data will also be collected with a small group of monolingual children, both in the Netherlands (Dutch) and Turkey (Turkish).

< previous page

page_85

next page >

< previous page

page_86

next page > Page 86

Data collection will start at age 2;0. With 18 consecutive 1-month-intervals spontaneous speech data will be collected. For the Turkish children in the Netherlands, data will be gathered in two languages using audio-tape recordings for one hour in each language. The data collection of the monolingual children will be based on audio recordings in the mother tongue. Half of the data collection will consist of recordings of child-caretaker interaction; half of the time child speech will be elicited from interaction with the researcher or her assistant. In all cases of elicitation of speech, native speakers will collect the data. In addition to the spontaneous speech data, several language comprehension tasks will be constructed and administered to the children in the three conditions at three age levels: at age 3;0, at age 3;3 and at age 3;6. Two of the three tasks are modified versions of tasks developed in the Berkeley cross-linguistic study of language acquisition (Slobin, 1982): a canonical sentence task testing comprehension of subject-object-verb sentences and a spatial reference task testing comprehension of locative expressions. The third task is a modified version of the task on definite and indefinite reference developed by Maratsos (1976). Bilingual development at school age Background Recent studies of first language acquisition have made clear that, by the age of 4, children are in command of many of the grammatical principles and rules governing their native language (cf. Goodluck, 1986). However, several studies have provided evidence that, both at the level of competence rules and performance preferences, language development goes on into the school years (Bowerman, 1979, 1982; Karmiloff-Smith, 1979, 1986). Later language development in children can be characterized by a growing command of discourse. According to Karmiloff-Smith, discourse can be seen as the most significant domain of later language acquisition. Around age 5, developmental shifts take place from intra- to intersentential devices, from basic structures to extended functions and from extra- to intralinguistic abilities. With respect to bilingual development at school age, it is still unclear what sort of operating principles children use. It is also unclear under what conditions processes of language transfer occur. Moreover, the studies that have been conducted so far were limited in their scope, given the fact that the languages under consideration were highly

< previous page

page_86

next page >

< previous page

page_87

next page > Page 87

related (cf. Grosjean, 1982; McLaughlin, 1985). The analysis of children's data in two typologically unrelated languages will offer new perspectives on the role of structural properties of these languages in the process of acquisition. The present project will focus on the bilingual development of Turkish and Moroccan children born in the Netherlands at school age. These children belong to a third generation of immigrants who initially moved from rural sites in Turkey and Morocco to industrialized areas in the Netherlands. The analysis of children's language development will not be confined to isolated sentences, but will also include the children's organizing processes over spans of connected utterances. The narrative, in its broader sense, will be taken as the relevant form of extended discourse. In the project the following main questions will be addressed: (1) How are the grammatical systems of children in first and second language use elaborated in the age range between 4 and 12? (2) How do children in this age range learn to master cohesive devices for anchoring discourse structure in narrative comprehension and production? (3) Under what conditions and with what grammatical constraints do processes of transfer occur? The present study will not be confined to oral language data. Given the fact that, starting at age 7, the school curriculum is highly devoted to reading and writing, written language data will be explored as well. Linguistic domains The following linguistic domains will be taken into account: anaphoric reference, topic continuity, temporality, clause linking and relativization. These domains have proved to be highly significant in a large body of cross-linguistic studies on language acquisition in both children (cf. Roeper & Williams, 1987; Slobin, 1985; 1988; Karmiloff-Smith, 1979) and adults (cf. Hyams, 1988; Givon, 1983). Cross-linguistic attention will be given to typological differences between Turkish, Arabic and Dutch in each of these domains. In the domain of anaphoric reference the developmental patterns of bound and free anaphora in the children's first and second language will be compared. The distinction between the two types of anaphora refers back to different principles in the standard binding theory of Chomsky (1981). In a variety of studies the acquisition of lexical anaphora and

< previous page

page_87

next page >

< previous page

page_88

next page > Page 88

pronouns have been studied in languages such as English and Dutch. With respect to bound anaphora, a fast pattern of acquisition could be evidenced, while the development of free anaphor resolution showed a much more irregular and slow development. However, the outcomes of studies referring to languages that are typologically very different from English seem to challenge Chomsky's claims. More recently, the acquisition of anaphoric reference was explored in a bilingual context. On the basis of empirical data on L2 acquisition of anaphora among Japanese and Spanish learners of English, Flynn (1987) concluded a primacy of the head-initial/head-final parameter's role. In the present study the acquisition of anaphoric reference by Turkish and Moroccan children in L1 and L2 will be studied by means of a series of experiments. The experimental set-up will involve a one sentence/four pictures multiple choice task that is similar to the one used by Deutsch, Koster & Koster (1986). With respect to topic continuity in discourse Hopper (1979) has claimed that narratives tend to divide into two sections: the foreground or sequence of events that define the plot and narrative line, and the background comprising the rest of the narrative which introduces, comments or concludes the plot. From a variety of studies it can be concluded that the thematic status in a narrative is marked (Chafe, 1980; Francik, 1985). Moreover, listeners or readers use thematic markers to organize the narrative (Kintsch & van Dijk, 1978; Morrow, 1985). There is also clear evidence that the way of representation of major characters in a narrative is crucial for its organization. Karmiloff-Smith (1981, 1986) has shown that anaphoric pronominalization is the default case for the thematic subject over a span of sentences. However, with reference to a main protagonist pronominalization can be avoided for a number of reasons (Clancy, 1980; Marslen-Wilson, et al., 1982; Givon, 1983; Redeker, 1987). Several studies addressed the question how topic continuity in narratives is acquired. Bamberg (1985, 1986; Karmiloff-Smith, 1981, 1986; Verhoeven, 1988a; Hickmann, 1982, 1987, 1988). The general conclusion from these studies is that the development of discourse cohesion involves the elaboration of linguistic devices so that independently represented entries in memory form a system, and that a control process must be created constraining the production of discourse markers across spans of related sentences. However, it is not clear to what extent the particular cohesive devices of individual languages affect the acquisition of narrative cohesion. In the present study an attempt will be made to answer this question by means of cross-linguis-

< previous page

page_88

next page >

< previous page

page_89

next page > Page 89

tic comparison. It will be explored how Turkish and Moroccan children learn to use the coding devices for topic continuity in their first and second language systems over time. In exploring reference to time, tense and aspect oppositions in narratives will be investigated. These temporal devices not only function to locate events relative to the moment of speech, but also as an organizer of narrative structure (Schriffrin, 1981; Fleischman, 1985; Wallace, 1982). In the present study, both tense and aspect variation in narratives is dealt with. Given substantial differences between the temporal systems in the languages under consideration, the simultaneous acquisition of linguistic devices for temporal reference can be thought of as highly relevant. Within the temporal domain it will be explored how the temporal organization of children's narratives in L1/L2 develops in the course of time. An attempt will be made to uncover the operating principles which underlie the synthetic and analytic devices for temporal marking in either language. With regard to clause linking the theoretical framework developed by Foley & van Valin (1984) will be taken as a starting point. They propose that in complex sentences, three types of juncture can be distinguished: nuclear juncture (in a core), core juncture (in a clause) and peripheral juncture (in a sentence). Nuclear juncture involves the joining of two nuclei which share all core and peripheral arguments and operators. In core juncture two or more cores are joined sharing the peripheral arguments, whereas in peripheral juncture full clauses are joined. Moreover, Foley and Van Valin consider three types of nexus. On the basis of evidence from a wide variety of languages, they show that the contrast between coordinate and subordinate structures can be expressed in terms of two parameters: dependency and embeddedness. In coordinate structures, the two units are independent and neither one is embedded in the other, while in subordinate structures one unit is embedded in and thus dependent on the other. They argue that a third nexus relation can be posited in which at least some units are dependent on one for the expression of certain categories, but no unit is embedded in any other. From a syntactic point of view, the acquisition of complex sentences involves insight in the layered structure of the clause and in the concepts of embeddedness and dependence. Given the fact that children start out linking clauses before all operators are understood, it can be expected that the linkage of sub-clausal units will precede the linkage of full clauses. Furthermore, it can be hypothesized that the linkage of autonomous clauses is conceptually simpler than the linkage

< previous page

page_89

next page >

< previous page

page_90

next page > Page 90

of clauses with one clause being embedded. Studies conducted by Bowerman (1979), Goodluck & Tavakolian (1982), Aksu-Koc (1988), Slobin (1988) and Verhoeven (1989a) give positive evidence for such claims. In the present study the acquisition of grammatical and pragmatic aspects of clause linking among Turkish and Moroccan children in their first ad second language will be dealt with. It will be determined how Turkish and Moroccan children in the age range from 4 to 12 learn to deal with the syntactic features of dependency and embeddedness in clause linking devices in L1 and L2. With respect to relativization there is a large body of literature on the processing of relative clauses in various unrelated languages. From such reviews as Hakuta (1981), Clancy, Lee & Zoh (1986) and MacWinney & Pleh (1988) it is clear that several intricately interacting factors must determine the processing of relative clauses: (1) the grammatical role played by the head of the relative clause, (2) the use of word order configurations in surface structure, (3) the interruption of processing units, and (4) the use of grammatical markers as cues to processing. From studies across Indo-European languages the general finding is that for children at school age subject-subject sentences are relatively easy, subject-object sentences relatively complex, while object-object and object-subject sentences take an intermediate position. Data on the acquisition of relative clauses in typologically different languages have proved to be very scarce: Hakuta (1981) on Japanese, Slobin (1986) on Turkish and English, and MacWhinney & Pleh (1988) on Hungarian. The attempts so far to relate typological differences to sentence processing difficulties underscore the need for cross-linguistic studies on the acquisition of relative clauses. In the present study the acquisition of relative clauses in Turkish, Arabic and Dutch will be examined in bilingual subjects at school age level. It will be determined in what order the various types of grammatical relations in relative clauses are acquired and what determining factors do account for relative clause difficulty in the three languages under consideration. Design of the study The present study will have a pseudolongitudinal design, based on first and second language data collection with two groups of Turkish, Moroccan and Dutch children, respectively 4 and 8 years old. Within each group, data will be collected over three intervals of one year. The main focus will be on the collection and analysis of narrative telling data. Moreover, a small number of experimental tasks will be admini-

< previous page

page_90

next page >

< previous page

page_91

next page > Page 91

stered. Data will be collected with groups of 50 Turkish, 50 Moroccan and 50 Dutch children of 4- and 8-years-old. All Turkish and Moroccan children will visit a Dutch primary school and have been living in the Netherlands for at least two years. With one-year intervals each group of informants will be tested four times. In order to elicit narrative production, series of pictures will be presented to the children. In these pictures, a sequence of events will be described in which a fixed number of characters is alternately involved. The children will be asked to describe the stories. All the story telling will be recorded and transcribed for further analysis. Apart from oral language data, written samples that are based on series of pictures will be collected among the oldest age group. The analysis of the data will start with a quantitative account of the distribution of referential devices in the children's narratives. In addition, a more qualitative analysis of micro-developmental changes in strategies for narrative cohesion will be done, e.g., by examining spontaneous repairs. Both quantitative and qualitative analyses will be carried out separately for the domains of topic continuity and temporality. In addition to the spontaneous oral and written language data, three experimental tasks will be constructed and administered to the children at the three age levels. First of all, a comprehension task for anaphoric reference will be developed, based on Deutsch, Koster & Koster (1986). The second and third experimental task involve clause linking and relativization. In these tasks children will be asked to act out the action described in a stimulus sentence. Code-switching among adolescents and adults Background Three lines of research have been developed within the field of codeswitching. First, code-switching phenomena are studied from a sociolinguistic point of view, focusing on the question for what reasons people switch codes. Second, code-switching is studied in order to get psycholinguistic insight into the mental organization of language in the bilingual brain. Finally, the linguistic properties of code-switching are studied. In the present study the linguistic dimension of code-switching phenomena will be given central focus. This is not say that socio- and psycholinguistic parameters will be neglected. As is clear from such studies as Pfaff (1979), Poplack (1980) and Nortier (1989), linguistic interpretations of code-switching behaviour are heavily dependent on

< previous page

page_91

next page >

< previous page

page_92

next page > Page 92

the social setting and on the first and second language proficiency of individual speakers. From a linguistic point of view, code-switching can be seen as constrained, rule-governed behaviour. Various kinds of constraints have been formulated by researchers, such as Lehtinen (1966), Hasselmo (1972) and Poplack (1980). The latter proposed two general constraints on the surface structure of sentences containing a switch. The first is the free morpheme constraint, stating that a switch cannot occur between a lexical form and a bound form unless the former has been phonologically integrated into the language of the latter. The other is the equivalence constraint, stating that the surface word order immediately before or after a switch point should be allowed for in both languages. However, the two constraints were actually derived from a specific contact situation (Spanish/English). In subsequent studies counter-evidence could be given, especially in cases of switching in agglutinative languages (cf. Romaine, 1989). Apart from the (re)formulation of surface constraints, there have been a few attempts to describe code-switching data in terms of transformational grammar (Woolford, 1983; Bentahila & Davies, 1983; Di Sciullo et al., 1985). Finally, Muysken (1987) proposed that code-switching theory can be advanced by studying constraint rules in their interaction with so-called neutrality strategies, such as the use of a specific set of morphological patterns for converting lexemes from one language to the other, and the suspension of grammar, i.e., the deletion of function elements. The present project aims at collecting speech samples of first and second generation Turkish and Moroccan immigrants in the Netherlands with a high incidence of code-switches and interpreting the cross-linguistic data in the light of current theories. Design The data will be collected in two stages. During the first stage, one hour audio recordings of informal interaction with a native interviewer will be collected among groups of about 15 Turkish and Moroccan adolescents and adults (total number = 60). Along with the spontaneous speech data, self-assessment of the first and second language proficiency of the informants will be obtained. In the second stage, 4 small groups of 4 informants will be selected from the initial groups of 15 informants. In this stage the data base will consist of an additional hour of audio-recording with an interviewer present, one hour of self-recorded in-group interaction between informants, and personal narra-

< previous page

page_92

next page >

< previous page

page_93

next page > Page 93

tives in L1 and L2 (30 minutes each). From these additional data insight is expected into the ways in which code-switching behaviour varies between informal (in-group) vs formal (interview) settings. The samples of L1 and L2 narratives are collected to establish a basis on which stretches of speech incorporated in code-switched utterances can be compared with analogous structures in utterances without such switches. Finally, samples from the corpus of code-switched material will be submitted to the informants themselves, and to other speakers of L1, in order to obtain judgements on the degree of acceptability of certain code-switched structures. Language attrition and language shift Background During the past decades the study of language loss among immigrants has received scant attention (e.g., Fishman, 1966, 1972; Kloss, 1966; Clyne, 1967; Haugen, 1969; Linguistic Minorities Project, 1985; Boyd, 1986). Saltarelli & Gonzo (1977) describe the pattern of language shift in general terms as follows. The first generation (G1) is bilingual with a strong dominance of the mother tongue, G2 is bilingual with a dominance of one language or with a balanced situation, G3 is bilingual with a dominance of the majority language, and G4 only masters the majority language. Gonzo & Saltarelli (1983) suggest that the combination of processes of language shift and language attrition determines the fate of minority languages. In G1 there is attrition within the individual group members through non-use and interference from the dominant language. The resulting reduced version is then the model for G2 learners of the language, so that only a reduced variety is acquired. Within G2 there is again attrition and consequently reduced transmission of the language to G3, etcetera. An important question is which aspects of the linguistic system are most vulnerable to language loss. It seems that grammatical knowledge of a language is extremely resistant to loss (Jordens, et al., 1989; De Bot & Clyne, 1989; Boyd & Andersson, to appear). Vocabulary, on the other hand, is supposed to be lost quite easily (Andersen, 1986; Cohen, 1986; Olshtain, 1986). Productive vocabulary seems to be more easily lost than receptive vocabulary (Andersen, 1983). However, it seems plausible that non-use of a language leads no so much to loss of voca-

< previous page

page_93

next page >

< previous page

page_94

next page > Page 94

bulary, but rather to reduction of availability of the knowledge of the language (Weltens, 1989). The central question in the present study is to what extent loss of lexical and syntactic skills can be evidenced in first and second generation Turkish and Moroccan immigrants. The loss of lexical skills will be emphasized. Perception data will be gathered using lexical decision techniques, while production data will be collected using a lexical substitution task. Besides, perceptual syntactic skills will be assessed with a judgement task, and productive syntactic skills by means of semi-spontaneous speech data. Design Two experimental techniques will be used to determine a decrease in availability of lexical information: an auditory lexical decision task, in which informants are given (pseudo)words along with the question whether the sequence heard is a word in a specific language, and a lexical substitution task, in which informants are presented (aurally) with a simple sentence in which one word is deleted and they are asked to give within a given time span as many alternatives for the missing word as possible. In order to study the organization of the bilingual lexicon, a repetition task is given in which a target word in language A is repeated in translated form in language B. It will then be determined whether the repeated form, just as in a monolingual repetition task, will be recognized as an existing word faster. The presentation of word forms will be done orally. In order to test the informants' syntactic skills, a cue-experiment referring to the choice of agent/subject will be conducted. Moreover, semispontaneous speech will be collected. In the experiments each group will consist of about 30 informants. For the collection of semi-spontaneous speech groups of 15 informants will be selected. All data collection will be on an individual basis. For the analysis of data multivariate techniques will be used with nationality (Turkish vs Moroccan), language (first vs second) and generation (1st vs 2nd) as the main factors. Conclusions and discussion Research on patterns of bilingual proficiency of Turkish and Moroccan children, youngsters, and adults in the Netherlands has only recently started. From the studies done so far it is clear that among Turkish and

< previous page

page_94

next page >

< previous page

page_95

next page > Page 95

Moroccan children in this diaspora context a gradual process of language shift is taking place. There is clear evidence that second-generation children born in the Netherlands do not reach native-like proficiency levels, due to reduced input conditions. It is an open question to what particular structural varieties and functional distributions the language contact situation will lead. This question is relevant from both a scientific and social perspective. From a scientific perspective, processes of present-day language change of ethnic minority groups in the Netherlands can be compared with the outcomes of similar processes in other countries, both from a crosslinguistic and crosscultural perspective. The features of new varieties in a diaspora context can also be studied from a developmental perspective. Moreover, in the long run, there can be evidence for the emergence of pidgin or creole varieties of the ethnic community languages under consideration. From a social perspective, the use of ethnic community languages in a diaspora context can finally be conceived as a key value of ethnic or cultural identity. It is an open question to what degree this value will be attributed over time to the use of Turkish and Arabic by both majority and minority groups in the Netherlands. Turkish and Moroccan parents, teachers and students show rather favourable attitudes towards maintaining or extending the status of Turkish and Arabic in various types of education. In governmental policy, however, the very existence of ethnic community languages is commonly conceived as a ''temporary problem" rather than a resource that for various reasons might be exploited and extended. As yet, it is unclear how this deficit perspective on ethnic minority groups in the Netherlands will effect the prestige of ethnic community languages, or whether this deficit perspective will shift over time to a policy of more culturally inspired pluralism. References ACOM (Adviescommissie Onderzoek Minderheden) (1989), Onderzoek etnische minderheden. Een geselecteerde bibliografie van onderzoek tussen 1984 en 1989. Den Haag: Ministerie van Binnenlandse Zaken. AKSU-KOC, A. (1988), The acquisition of aspect and modality. Cambridge: Cambridge University Press. ANDERSEN, R. (1983), Pidginization and creolization as language acquisition. Rowley, Mass.: Newbury House.

< previous page

page_95

next page >

< previous page

page_96

next page > Page 96

ANDERSEN, R. (1986), Determining the linguistic attributes of language attrition In: R. Lambert & B. Freed (eds.), The loss of language skills. Rowley, Mass.: Newbury House. ANKERSMIT, T., T. Roelandt & J. Veenman (1989; 1990), Minderheden in Nederland. Statistisch Vademecum 1989; 1990. Den Haag: SDU Uitgeverij. BAMBERG, M. (1985), Form and function in the construction of narratives. Doctoral dissertation, University of California, Berkeley. BAMBERG, M. (1986), A functional approach to the acquisition of anaphoric relationships. In: Linguistics, 24, 227-284. BENTAHILA, A. (1983), Language attitudes among Arabic-French bilinguals in Morocco. Clevedon: Multilingual Matters. BENTAHILA, A. & E. DAVIES (1983), The syntax of Arabic-French code-switching. Lingua, 59, 301-330. BERMAN, R. & Y. Sagi (1982), Word formation processes and lexical innovations of young children. In: Hebrew Computational Linguistics Bulletin, 18, 36-62. BOESCHOTEN, H. & L. Verhoeven (1985), Integration niederländischer lexikalischer Elemente ins Turkische: Sprachmischung bei Immigranten der ersten und zweiten Generation. In: Linguistische Berichte, 98, 347-364. BOESCHOTEN, H. & L. Verhoeven (1986), Turkish language development of Turkish children in the Netherlands. In: A. Aksu & E. Erguvanli (eds.), Proceedings of the second Conference on Turkish Linguistics. Instanbul: University Press. BOESCHOTEN, H. & L. Verhoeven (1987), Language mixing in children's speech: Dutch language use in Turkish discourse. In: Language Learning, 37, 2, 191-215. BOWERMAN, M. (1979), The acquisition of complex sentences. In: P. Fletscher & M. Garman (eds.), Language acquisition. Cambridge: Cambridge University Press. BOWERMAN, M. (1982), Starting to talk worse: Clues to language acquisition from children's late speech errors. In: S. Strauss (ed), U-shaped behaviorial growth, New York: Academic Press. BOYD, S. (1986), Using the present to predict the future in language contact: The case of immigrant minority languages in Sweden. In: B. Weltens, K. de Bot & Th. van Els (eds.), Language attrition in progress, Dordrecht: Foris. BROEDER, P., G. Extra, R. van Hout, P. Strömqvist & K. Voionmaa (1988), Processes in the developing lexicon. Tilburg/Göteborg: ESF-Report. CHAFE, W. (1980), The pear stories: cognitive, cultural and linguistic aspects of narrative production. Norwood, NJ: Ablex. CHOMSKY, N. (1981), Lectures on Government and Binding. Dordrecht: Foris. CLANCY, P. (1980), Referential choice in English and Japanese narrative discourse. In: W. Chafe (ed), The pear stories: cognitive, cultural and linguistic aspects of narrative production. Norwood, NJ: Ablex. CLANCY, P., H. Lee & M. Zoh (1986), Processing strategies in the acquisition of relative clauses. In: Cognition, 24, 225-262. CLARK, E. (1981), Lexical innovations: How children learn to create new words. In: W. Deutsch (ed), The child's construction of language, London: Academic Press. CLARK, E. & R. Berman (1984), Structure and use in the acquisition of word formation. In: Language, 60, 3, 542-590. CLARK, E., B. Hecht & R. Mulford (1986), Coining complex compounds in English: Affixes and word order in acquisition. In: Linguistics, 24, 7-29. CLYNE, M. (1967), Transference and triggering. The Hague: Nijhoff.

< previous page

page_96

next page >

< previous page

page_97

next page > Page 97

COHEN, A. (1986), Forgetting foreign language vocabulary. In: B. Weltens, K. de Bot & Th. van Els (eds.), Language attrition in progress. Dordrecht Foris. DEUTSCH, W., C. Koster & J. Koster (1986), What can we learn from children's errors in understanding anaphora? In: Linguistics, 24, 203-225. DE BOT, K. & M. Clyne (1989), Language reversion revisited. In: Studies in Second Language Acquisition, 11, 167-177. DE RUTTER, J. (1989), Young Moroccans in the Netherlands. An integral approach to their language situation and acquisition of Dutch. Doctoral dissertation. Utrecht: University. DISCIULLO, A., P. Muysken & R. Singh (1986), Government and code-switching. Journal of Linguistics, 22, 1-24. ELLEMERS, J., J. Vermeulen & M. Borghuis (eds.) (1988), Etnische minderheden in Nederland. Een geselecteerde bibliografie van sociaal-wetenschappelijke publicaties 1945-1986. Muiderberg: Coutinho. EXTRA, G. (1990a), Processes of language change over time. A linguistic perspective on ethnic minority research in the Netherlands. In: F. Aarts & Th. van Els (eds.), Contemporary Dutch linguistics. Washington: Georgetown University Press. EXTRA, G. (1990b), Ethnic minorities, language diversity, and educational implications. A case study in the Netherlands. In: T. Bongaerts, K. de Bot & Th. van Els (eds.), Balance and perspective: 25 years of Dutch applied linguistics. Toegepaste Taalwetenschap in Artikelen 36 (1). EXTRA, G. & T. Vallen (eds.) (1985), Ethnic minorities and Dutch as a second language. Dordrecht: Foris. EXTRA, G. & T. Vallen (1987), Onderzoek naar de positie van etnische minderheden in de Nederlandse samenleving: een taalwetenschappelijk perspectief. In: ZWO Jaarboek 1986, Den Haag: ZWO. EXTRA, G. & T. Vallen (1988), Language and ethnic minorities in the Netherlands. In: F. Coulmas & J. Stalpers (eds.), The Sociolinguistics of Dutch. Special Issue of the International Journal of the Sociology of Language 73. EXTRA, G. & T. Vallen (1989), Second language acquisition in elementary schools. A crossnational perspective on the Netherlands, Flanders and the Federal Republic of Germany. In: L. Eldering & J. Kloprogge (eds.), Different cultures, same school. Ethnic minority children in Europe. Amsterdam: Swets & Zeitlinger. EXTRA, G. & L. Verhoeven (1991), The Moroccan community in the Netherlands. Patterns of language choice and language proficiency. In: P. Demas & H. Obdeijn (eds.), Proceedings of the third Moroccan-Dutch Symposium, held at the University of Rabat in 1990. To appear. FILLMORE, C. (1982), Towards a descriptive framework for spatial deixis. In: R. Jarvella & W. Klein (eds.), Speech, place and action, London: Whiley and Sons. FISHMAN, J. (1966), Language Royalty in the United States. The Hague: Mouton. FISHMAN, J. (1977), The social science perspective. In: Bilingual education: current perspectives. Arlington, VA: Center for Applied Linguistics. FLEISCHMANN, S. (1985), Discourse functions of tense-aspect oppositions in narrative: Toward a theory of grounding. In: Linguistics, 23, 851-882. FLYNN, S. (1987), A parameter-setting model of L2 acquisition. Dordrecht: Reidel. FOLEY, W. & R. van Valin (1984), Functional syntax and universal grammar. Cambridge: Cambridge University Press.

< previous page

page_97

next page >

< previous page

page_98

next page > Page 98

FRANCIK, E. (1985). Referential choice and focus of attention in narrative. Doctoral dissertation, Stanford University. GIVÓN, T. (1983), Topic continuity in discourse: An introduction. In: T. Givón (ed), Topic continuity in discourse: a quantitative crosslinguistic study. Amsterdam: Benjamins. GONZO, S. & M. Saltarelli (1983), Pidginization and linguistic change in emigrant languages. In: R. Andersen (ed), Pidginization and creolization as language acquisition. Rowley, Mass.: Newbury House. GOODLUCK, H. (1986), Language acquisition and linguistic theory. In: F. Fletscher & P. Gamran (eds.), Language acquisition. Cambridge: Cambridge University Press. GOODLUCK, H. & S. Tavakolian (1982), Competence and processing in children's grammar of relative clauses. In: Cognition, 11, 1-27. GROSJEAN, F. (1982), Life with two languages. Cambridge: Harvard University Press. HAKUTA, K. (1981), Grammatical description versus configurational arrangement in language acquisition. In: Cognition, 9, 197-236. HASSELMO, N. (1972), Code-switching as ordered selection. In: Studies for Einar Haugen. The Hague: Mouton. HAUGEN, E. (1969), The Norwegian language in America: A study in bilingual behaviour. Bloomington: Indiana University Press. HICKMANN, M. (1982), The development of narrative skills. Doctoral dissertation. University of Chicago. HICKMANN, M. (1987), Clause structure variation in narrative discourse: A functional and cross-linguistic approach (Paper). Nijmegen: Max Planck Institute. HICKMANN, M. (1988), Topics and subjects in discourse: An analysis of French children's narratives. Paper presented at the 3rd European Conference on Developmental Psychology, Budapest. HOPPER, P. (1979), Aspects and foregrounding in discourse. In: T. Giv6n (ed), Syntax and semantics: Vol.12 Discourse and syntax, New York: Academic Press. HYAMS, N. (1986), A principles and parameter approach to the study of child language. Paper presented at the 27th Child Language Research Forum in Stanford. HYAMS, N. (1988), A principles and parameter approach to the study of child language. Paper presented at the 27th Child Language Research Forum in Stanford. JOHANSON, L. (1971), Aspekt im Turkischen. Studia Turcica Upsaliensia 1. JOHNSTON, J. & D. Slobin (1979), The development of locative expressions in English, Italian, Serbo-Croatian and Turkish. In: Journal of Child Language, 6, 529-545. JORDENS, P., K. de Bot & H. Trapman (1989), Linguistic aspects of regression in German case marking. In: Studies in Second Language Acquisition 11 181-206. KARMILOFF-SMITH, A. (1979), A functional approach to child language. Cambridge: Cambridge University Press. KARMILOFF-SMITH, A. (1981), The grammatical marking of thematic structure in the development of language production. In: W. Deutsch (ed), The child's construction of language. New York: Academic Press. KARMILOFF-SMITH, A. (1986), From meta-processes to conscious access: Evidence from children's metalinguistic and repair data. In: Cognition, 23, 95-147. KINTSCH, W. & T. van Dijk (1978), Toward a model of discourse comprehension and discourse production. In: Psychological Review, 85, 363-394. KLOSS, H. (1966), German-American language maintenance efforts. In: J. Fishman (ed), Language loyality in the United States. The Hague: Mouton.

< previous page

page_98

next page >

< previous page

page_99

next page > Page 99

KOULEN, I. & V. Smit (1988). Bibliografie etnische groepen. Een overzicht van in Nederland verschenen publikaties m.b.t. etnische groepen in de periode 1985-1986. Amsterdam: Noord-Hollandsche Uitg. Mij. LETHINEN, M. (1966), An analysis of a Finnish-English bilingual corpus. Indiana University (PhD. Thesis). LINGUISTIC MINORITIES PROJECT (1985), The other languages of England. London: Routledge & Kegan Paul. LUCASSEN, J. & R. Penninx (1985), Nieuwkomers, Immigranten en hun nakomelingen in Nederland 1550-1985. Amsterdam: Meulenhoff. MACWINNEY, B. & C. Pleh (1988), Relative clauses in Hungarian. In: Cognition, 29, 95-141. MARATSOS, M. (1976), The use of definite and indefinite reference in young children. Cambridge: Cambridge University Press. MARSLEN-WILSON, W., E. Levy & L. Tyler (1982). Introduction and maintenance of reference. In: R. Jarvella & W. Klein (eds.), Speech, place and action. Chichester: Wiley. MCLAUGHLIN, B. (1985), Second language acquisition in childhood. Hillsdale, N.J.: LEA. MORROW, D. (1985), Prominent characters and events organize narrative understanding. In: Journal of Memory and Language, 24, 304-319. MUUS, P. (1989; 1990), Migration, minorities and policy in the Netherlands. Recent trends and developments. Amsterdam: Institute for Social Geography. MUYSKEN, P. (1987), Neutrality in code-mixing. In: Eigen en Vreemd Handelingen van het 39e Filologencongres Amsterdam. Amsterdam: Vrije Universiteit. NORTIER, J. (1989), Dutch and Moroccan Arabic in contact: Code-switching among Moroccans in the Netherlands. Dordrecht: Foris. OLSHTAIN, E. (1986), The attrition of English as a second language: A case of Hebrew-speaking children. In: B. Weltens, K. de Bot & Th. van Els (eds.), Language attrition in progress, Dordrecht: Foris. PERDUE, C. (1984), Second language acquisition by adult immigrants: A field manual. Strasbourg: ESF. PFAFF, C. (1979), Functional and structural constraints on syntactic variation in code-switching. In: Language, 55, 291-318. PFAFF, C. (1988), Linguistic and social determinants of Turkish/German bilingualism of migrant children in Berlin. In: Slavica Lundensia, 12, 49-74. POPLACK, S. (1980), Sometimes I'll start a sentence in English y termino en Español. In: Linguistic, 18, 581-616. REDEKER, G. (1987), Language use in informal narratives. Doctoral dissertation, University of California at Berkeley. ROELANDT, T. & J. Veenman (1990), Ethnic minorities in the Netherlands, education and labour market opportunities: Policy and research. In: P. Muus (1990), Migration, minorities and policy in the Netherlands. Recent trends and developments. Amsterdam: Institute for Social Geography, 29-45. ROEPER, T. & E. Williams (1987), Parameter setting. Dordrecht: Reidel. ROMAINE, S. (1989), Bilingualism. Oxford: Blackwell. SALTARELLI, M. & S. Gonzo (1977), Migrant languages: linguistic change in progress. In: H. Giles & B. Saint-Jacques (eds.), Language and ethnic relations. Oxford: Pergamon Press.

< previous page

page_99

next page >

< previous page

page_100

next page > Page 100

SCHIFFRIN, D. (1981), Tense variation in narratives. In: Language, 57, 1, 45-63. SLOBIN, D. (1982), Universal and particular in the acquisition of language. In: E. Wanner & L. Gleitman (eds.), Language acquisition, Cambridge: Cambridge University Press. SLOBIN, D. (1985), The cross-linguistic study of language acquisition. Hillsdale: LEA. SLOBIN, D. (1986), The acquisition and use of relative clauses. In: D. Slobin & K. Zimmer (eds.), Studies in Turkish linguistics. Amsterdam: Benjamins. SLOBIN, D. (1988), The development of clause chaining in Turkish child language. Paper presented at the 4th Conference on Turkish Linguistics, Ankara. SLOBIN, D. & T. Bever (1982), Children use canonical sentence schemas: A crosslinguistic study of word order and inflections. In: Cognition, 12, 229-265. TAESCHNER, T. (1983), The sun is feminine. A study on language acquisition in bilingual children. Berlin: Springer. VERHOEVEN, L. (1988a), Acquisition of discourse cohesion in Turkish. Paper presented at the 4th Conference on Turkish Linguistics, Ankara. VERHOEVEN, L. (1988b), The bilingual child as a word maker: Word formation processes in Turkish and Dutch. Berkeley Linguistic Society. VERHOEVEN, L. (1989a), Acquisition of clause linking in Turkish. In: H. Bennis & A. van Kemenade (eds.), Linguistics in the Netherlands 1989. Dordrecht: Foris. VON STUTTERHEIM, C. (1986), Der Ausdruck der Temporalität im ungesteuerten Spracherwerb. Berlin: De Gruyter. WALDEN, Z. (1982), The root of roots: Children's construction of word formation processes in Hebrew. Doctoral dissertation, Harvard University. WALLACE, S. (1982). Figure and ground: The interrelationship of linguistic categories. In P. Hopper (ed), Tense-aspect. Between semantics and pragmatics. Amsterdam: Benjamins. WELTENS, B. (1989), The attrition of French as a foreign language. Dordrecht: Foris. WOOLFORD, E. (1983), Bilingual code-switching and syntactic theory. In: Linguistic Inquiry, 14, 520-536.

< previous page

page_100

next page >

< previous page

page_101

next page > Page 101

4 Croatian or Serbian as a diaspora language in Western Europe Andrina Pavlinic * In what was until recently Yugoslavia Croatian or Serbian is literally the majority language, the lingua communis, being used by almost 16,5 million people, or four fifths of all inhabitants, residing in four republics: Croatia, Serbia, Bosnia and Herzegovina, and Montenegro (the remaining standard languages are Slovene and Macedonian). This also means that the S/C language derives from several different and distinct civilizations, which have produced several literary standards (written in two scripts, Cyrillic and Latin); the language is used by as many as three main denominations: Orthodox, Roman-Catholic, and Muslim ones. Croatian or Serbian is also the diaspora language used by "Yugoslav"1 communities living in Western Europe. The concept 'diaspora' is used in opposition to that of a 'stabilized minority', that is, one with a historically developed and stabilized minority language (cf. Durovic*, 1988). The children born in a diaspora acquire their mother tongue (MT further in the text) in a linguistically non-homogeneous society unlike those born in a traditional minority group. According to Durovic and his research team in Lund2, the diaspora language is capable of developing as idiolects (that is, on individual levels). The idiolects are conditioned by the "self-regulating capacity" of the language system, which capacity actually represents the authenticity of the language, and is directly linked to the availability of its standard or its norm and of speakers of the standard. In the present chapter the use of S/C language varieties in Western Europe will be dealt with. First of all, the sociolinguistic background of Croatian and Serbian speaking communities living in the former Yugoslavia and in Western Europe will be discussed. In addition, the role of mother tongue education for children in the diaspora will be examined. Moreover, the Zagreb project on the language of S/C speaking children living in European host countries will be introduced. Some contact

< previous page

page_101

next page >

< previous page

page_102

next page > Page 102

phenomena between Croatian, Serbian and Dutch will be explored from compositions written by children attending secondary school in the Netherlands. Finally, the data will be shortly discussed. Sociolinguistic background Standard Croatian or Serbian could in principle be considered as one language but it is clearly realized through four variants, which happen to coincide with the territories of the four republics mentioned. The specificity of the standard leads to a paradox: on the one hand, there is an impressive language community made up of speakers of the four nations and of ethnic minorities co-habitating with them (e.g., Muslims in Bosnia and Herzegovina and in some other regions), which aims at achieving modern communicative efficiency; on the other hand, people in the four regions have constant fears of their distinguishing variant peculiarities being in jeopardy because of their possible loss (cf. Skiljan *, 1989). This situation is further complicated by the fact that each of the two prevailing variants, Croatian and Serbian, has the attributes of a standard language, performing all of its functions in the range from colloquial usage to scientific literary ones, and is as such recognized as a means of public and official communication. This is one of the reasons why even its name is debatable and a cause for dissent. In the source country of today, it definitely has to do with a political choice: it is called Serbo-Croatian in Serbia, Montenegro and some other parts, among other reasons so as to signal that the Serbian variant is referred to. When it is called Croato-Serbian, the Croatian variant is given priority, possibly by Roman-Catholic or Muslim speakers in Bosnia and Herzegovina. In the Republic of Croatia, however, it is officially called the Croatian (literary) language, which co-exists with the Serbian language (in Croatia); the two are referred to as the Croatian or Serbian language (S/C). Anyway, folk appellations have always been Serbian and Croatian, respectively. If one were to use the "Serbo-Croatian" appellation in Croatia today, he/she would be immediately seen as a proponent of the unitarian (federalist) concept of Yugoslavia. The picture gets even further complicated, because S/C has three main dialects: Stokavski* (i.e., Stokavian*), Kajkavski (i.e., Kaikavian), and Cakavski* (i.e., Cakavian*), whose names are derived from the interrogative pronoun "what" pronounced as "sto*", ''kaj" and "ca*" in their

< previous page

page_102

next page >

< previous page

page_103

next page > Page 103

respective regions of use, plus the less-known Torlacki * (i.e., Torlachian) of the Prizren-Timok area, with a multitude of local variations, the degree of mutual comprehension varying with the geographical distance between their respective language settings and the historical and cultural heritage. The literary standard, as it is called in the source country, has two realizations or types, treated on an equal footing: Ijekavski (i.e., Iyekavian) and Ekavski (i.e., Ekavian) (for more details on the language varieties, their status, and patterns of language choice in the Netherlands, see Gvozdanovic*, forthcoming). This introduction may have been necessary for various reasons. First of all, the language situation in "Yugoslavia" is generally reflected among its citizens abroad. Throughout the history of the country or of its individual regions, S/C has been a unifying factor or, as it is at present in some ways, a dividing factor among "Yugoslav" nations and nationalities. Presently diversity is emphasized, which is in the relationship of cause and effect with growing ethnic tensions and the war. An extreme way of achieving this is linguistic regression. This phenomenon may manifest itself as a return to what would have been a "normal'' development of the language had it not been hampered for the sake of the variants' prescribed uniformity. Linguistic regression seems to belong to the sphere of language manipulation, as it is closely related to the phenomenon that Freud used to call the "narcissism of a small difference" (i.e., we are proud to be different from you and expose the difference, to the ear in this case). Secondly, migrant workers' children's language situation is a difficult one; the majority of them, who have spent all or most of their lives in a host country, often have to cope with as many as four language varieties; one would wish them to be plurilingual, but they may be doubly diglossic instead (cf. Pavlinic*-Wolf et al., 1988 - with regard to Wallachian children, speakers of Vlaski* in Copenhagen schools, and Strauch, 1991 Italian children in the schools of Baden-Württenberg, in the same context). The children use the local dialect with their peers and are expected to master the regional variety of the standard (majority) language at school; also, they have acquired a (partial) knowledge of the dialect from their parents and relatives' region of origin, and have to learn standard S/C during supplementary instruction (Yugoslav supplementary education, jugoslavenska dopunska nastava, to use the S for it, has been disintegrating into "supplementary education" and "education in the Croatian language and culture" as a fairly recent develop-

< previous page

page_103

next page >

< previous page

page_104

next page > Page 104

ment), for obvious reasons: the standard is the language of education, literature, media, etc. in the home country or region. Thirdly, language planning in "Yugoslavia" is unsatisfactory and deficient vis-à-vis both its citizens at home and abroad, because a consensus of opinion can hardly ever be reached in this connection, especially as regards the Croatian and Serbian variants; one of the outcomes of such a situation is that dictionaries (especially of S/C for special purposes), orthography books and grammars are lacking for both domestic and external use, as well as textbooks and other didactic materials for mother tongue instruction and maintenance in a foreign socio-cultural milieu; at home, textbooks for learners of S/C as a second language are not sufficiently developed. Finally, the individual variants of S/C are in need of protection at home from the supremacy of just one of these (as the language of the army, administration, legislation, etc.) over the rest of them. One can also claim that S/C is a minority language in the countries of immigration although not always recognized as such (the minority status need not have only negative implications; on the contrary, in some countries it may bring certain benefits to the language concerned, i.e., financial and other kinds of provision). As a minority language, it is in need of protection, however, just like a "minority language proper", and for similar reasons: the language identity of migrant (and former migrant) workers' children has undergone degradation and can hardly be considered as a vehicle of their original culture and communications: it is also in need of protection for reasons originating from the source country, as enumerated earlier. As concerns the former, one way of protecting it is to make mother tongue and culture instruction compulsory by integrating such classes into the curriculum; sometimes this is not done as part of the language policy of the country or region concerned; sometimes, even though S/C classes do exist, "Yugoslav" children may choose not to attend them, one of the reasons being that they find the grammar of their MT too complicated, as has been reported by Vegar (1987) concerning the situation in Vienna; there have also been numerous instances of a variant taught at supplementary MT classes being unacceptable, not so much for the children, who may, in ignorance, call the S/C language "Yugoslav" or "naski *" (i.e., our own), as for their parents. (The latter is now being changed, as stated above). When stressing the importance of such classes, however, I am not short of arguments in their favour; a few randomly selected ones are to be mentioned in the next section.

< previous page

page_104

next page >

< previous page

page_105

next page > Page 105

Mother tongue education The language in question may be a small language spoken by small nations, but it derives from and contributes to a culture or cultures which are old and diversified and have something to offer to Europe's heritage (e.g., from the standpoint of intercultural thinking). One feels inclined to agree with Roosens' thinking (1989: 91) that "we are evolving toward a world that is becoming ever more uniform on the instrumental level, but that, at the same time, is becoming more keenly aware of symbolic expression in areas of life not dominated by the operational sector". On the other hand, there is covert (and overt) thinking that there is a "hierarchy of languages", which reflects the unequal development and the cultural influence of the individual countries (cf. Costa-Lascoux, 1989: 76). The latter may have to do with "one nation - one language" rationale, which has been traditionally upheld by some former nation-states in Europe and has been embedded in their schools (which are, consequently, designed for ensuring school achievement only for children from the favoured classes of society). There have been many instances throughout history even of something so extreme as linguistic self-hatred (to use Sledd's expression) being gradually replaced by a rejuvenated language loyalty among the speakers of many "obsolescent" languages (to use Weinreich's term and observation, 1968); reasons for this may be manifold, and are usually of paramount importance to the speakers; then why not give such languages a chance as a fruit from the human rights basket. Several authors, Stölting-Richert (1988) and Strauch (op.cit.) among them, have reported about the beneficial, even amazing effects of mother tongue instruction, which is much more successful even under difficult conditions than foreign language instruction in regular schools, to say the least of it. There are some indicators that it may even be in positive correlation with school achievement, but this matter should be further examined. The very concept of "mother tongue" has been a subject of much debate and disagreement in the past decade, among scholars and laymen alike, especially in the context of so called third-generation migrants who were born in the countries of their parents' immigration and may, therefore, regard those countries, as well as the majority language, as their own, whereas their mother tongue would lose its availability for them, its usefulness and status, often being degraded to no more than their first second language. Even under such circumstances, it seems to

< previous page

page_105

next page >

< previous page

page_106

next page > Page 106

be quite logical to opt for reinforcing the mother tongue fostering line of thinking, not just as a political declaration, but also as a socio-cultural, professional and pragmatic choice. One tends to agree with the views of Cummins and Skutnabb-Kangas (1984, 1988, and elsewhere) as proponents of mother tongue medium maintenance education for minorities (in the early years of schooling), because they seem to be well-founded and corroborated by a bulk of Scandinavian, American and Canadian research and practical work. The two scholars' line of thinking may find support in Wode's recent work (1988). The German psycholinguist claims that a number of skills that became activated in the process of MT acquisition have remained available throughout one's life, meaning also in the process of L2 acquisition. Their activation in a concrete situation is conditioned by various psychological and social factors, and one of the tasks of psycholinguistic research should be to discover the link between the two. Before Wode, Ringbom (1987: 36-38) pointed out that for an L2 learner to comprehend L2 utterances, which would supposedly precede his/her producing them, a similarity with the already acquired skills, in his/her L1 in the first place, would be a considerable help. To conclude: the acquisition of MT skills is important, not just for a child's linguistic or cultural identity and his/her personality development, but also for majority language acquisition and development. By inference a more successful integration can take place of a person of integrity (in the sense of wholeness, completeness) into either or both of the worlds for which he/she has earned a pass. Therefore, provision for migrant children's mother tongue instruction from an early stage is recommended. With Europe becoming united, there is a growing "need for counter-hegemony", to use Cummins and Skutnabb-Kangas' expression. There is, what can be called, the "political lever effect". To exemplify, let me take the language of Burgenland Croats, who are a traditional (stabilized) Austrian minority with a language (varieties of Cakavski * and Stokavski* dialects) which has survived since the 16th century Turkish war threat in Europe and has had its own development separate from that of the Croatian standard at home. However, it has not been accorded an official status in Austria, or that of a European language, so appropriate legal and other provisions for its protection and preservation have not been made. One tends to agree with Gassner (1990), a Croat from Austria, who claims that even when concessions relating to a mother tongue and culture are gained from a majority

< previous page

page_106

next page >

< previous page

page_107

next page > Page 107

society, there is never any certainty that they will remain as a permanent feature. An obvious conclusion is that minority language speakers, including scholars, should engage in continuous struggle to improve the situation. Some praiseworthy work in this connection has been carried out by The European Bureau for Lesser Used Languages in Ireland. Dutch and Croatian or Serbian in contact The project, carried out in the host countries: Austria, Denmark, France, F.R. Germany, the Netherlands, Sweden, Switzerland and the United Kingdom, has just been finalized in the Institute for Migration and Nationalities of the University of Zagreb. This is a comprehensive sociolinguistic project, one of its purposes being to draw attention to the socio-cultural and linguistic status of Yugoslav labour migrants' children. They are culturally much "less different" than most of the other ethnic groups living in the industrialized countries, and are therefore very seldom focused on in their migration studies. The sample consists of 342 respondents aged 10-18, speakers of Croatian or Serbian, and their L1 competence is reflected through a grammar test, a composition and self-reports, elicited by means of a questionnaire. Information on their L2 competence is obtained only from self-reports. A number of socio-demographic and other data have been collected. The next section is a condensed account of some contact phenomena observed in the compositions ("My Home", "My School", "My Homeland Yugoslavia", or "My New Homeland") of young speakers of the S/C minority language in a Dutch environment. Among the thirty respondents who wrote compositions (there were also those who did not), sixteen were born in the Netherlands, and several of them have by now spent only a couple of years off-and-on in "Yugoslavia". Even a majority of those who were born there have spent most of their lives in the Netherlands. The quality of compositions ranges from very poor (one composition containing one sentence refusing to do the assignment) to very good (5 compositions), with hardly any mistakes. The remaining twenty-four exhibit a number of examples of S/C switch, attrition and loss phenomena. After the compositions of the respondents, mostly attending secondary school in the Netherlands, have been examined, both intra- and interlingual characteristics can be discovered.

< previous page

page_107

next page >

< previous page

page_108

next page > Page 108

Intralingual characteristics Interference with the Dutch language (D) need not be considered as the only source of the respondents' errors. A reduced language system is a characteristic of interlanguage, which has been manifested at the following levels: Phonology (orthography) Q. No. 706 Q. No. 708

Karin je jedno malo mesto de nedolaze Stranci. De ja idem u Jugoslavensku skolu * je lepo.

The correct S/C consonantal cluster, the standard gdje (C)/gde (S) has been reduced to be made easier to pronounce. However, de, di and de* can sometimes be heard in some (spoken) dialects in the regions of origin. Morphology Q. No. 742

... zrak je tamo zdravljiji nego ovde.

This comparative of the adjective zdrav (m.), zdrava (f.), zdravo (n.) is not an example of reduction, but it shows the respondent's uncertainty as to its formation. Zdraviji is the correct suffix in this case, although with some other adjective -lji would be correct, i.e., in grublji, deblji. Morphosyntax Q. No. 735

Svake godine idem na godisnji* odmor u Hrvatskoj. Moja zelja* je da ostanem zivjeti* u Hrvatsku.

These are examples (found in the same Questionnaire) of misused (displaced) case endings (Hrvatsku would have been correct in the first instance, Hrvatskoj in the second), which really abound in the compositions. There are several instances of case ending omission (loss), which can be attributed to: (a) interlanguage reduction, interference with D (which has no case system), the children's or their parents' regional dialect.

< previous page

page_108

next page >

< previous page

page_109

next page > Page 109

Syntax Q. No. 730

Zato se ponosim mojom domovinom.

The possessive adjective mojom - the Instrumental case of moja (f.); moj (m.), moje (n.) - is out of place here. The referential possessive svojom - Nominative svoj (m.), svoja (f.), svoje (n.) - should be used instead. It is a particular syntactic device of S/C denoting that something or somebody pertaining to the subject in the 1st person singular (sg.) is being referred to by the subject. The same kind or error is also frequently made in the source country, sometimes even by most educated speakers (out of carelessness rather than ignorance). Q. No. 743

. . . Onda mi prodavamo jedni drugima knjige od zivotinjama *. . . .

The correct preposition in this context (with Dative sg. & pl. such as Dative plural zivotinjama) should be o, whereas od should require a Genitive sg./pl. At this linguistic level there are numerous examples of different types of reduction, such as: • usage of only one or two tenses (e.g., present and perfect, or present and future) when three or four may have been used; • reduction of a number of prepositions, conjunctions, etc.; • reduction of sentence structure and types of sentences; • reduction/avoidance of compound clauses, etc. Text (discourse) Q. No. 708

Ja imam isto prijatelicu zove se: Violeta.

This is one of the numerous examples of cohesive devices omission; the sentence should have read: Ja takoder* imam prijateljicu koja se zove Violeta. Extreme cases of this phenomenon look like "stream of consciousness" type of sentences (i.e., pouring out one's mind without a halt). Semantics (vocabulary) An interesting combination of interlanguage and interference phenomena is the following example:

< previous page

page_109

next page >

< previous page

page_110

next page > Page 110

Q. No. 272

Zove se "Vladimir Nazor" i predaje se samo subotom nastava.

The entire sentence construction is taken over from the Dutch (interference): "er wordt alleen op zaterdag les gegeven", or "er is alleen op zaterdag onderwijs"; the so called approximation strategy (a feature pertaining to interlanguage) is applied in predaje se ... nastava, so that a new lexical unit predaje se has been formed, one which is semantically very close to the correct odrzava * se. Orthography (phonology) There are errors exhibiting an ignorance of e-reflex*, c/c*, distinction, etc., i.e., in: Q. No. 708

... hocu* pijevacica* da budem ...

(instead of hocu and pjevacica*). The example is also a mixture of S and C variants; the C should read Ja zelim* biti pjevacica. Interlingual characteristics Most of the remaining examples seem to have to do only with contact/ interference phenomena. Some of them are going to be presented below. Phonology (orthography) Q. No. 708

Moja ucitelica* je lepa i dobra.

Instead of the I there should have been lj/pronounced something like ly/; however, there is no such letter (and phenomena) in D. Morphology Q. No. 708

I moj uctelica zivo* pokaze*.

It should have been moja ucteljica because of S/C congruence: the possessive adjective has to be in agreement with its noun (feminine, in this instance). In D, however, possessives such as this have only one form irrespective of the gender of the noun.

< previous page

page_110

next page >

< previous page

page_111

next page > Page 111

Examples at this linguistic level are few and far between. Syntax Q. No. 717

Ja volim mnogo moju domovinu.

This is a typical example of D word-order interference, with an adverb following the verb: "ik hou veel van ...". In S/C the adverb would precede the verb. Q. No. 732

Tako ce * biti svake godine tako sve dok se jednog dana vratimo.

The dependent temporal clause (starting with sve), which serves the function of interrupting the action in the main clause, should contain a negative: dok se jednog dana ne vratimo. The negative has been left out because of D interference (no negative in such a sentence-type). Interference at the level of syntax is the most frequent of all contact phenomena observed, so it may deserve special attention. Word-order changes abound. While the use of a personal pronoun in the subject-role is necessary in D, it may be (and often is) redundant in the same S/C context: Q. No. 741

Subotom ja idem u Jugoslavensku skolu*.

Furthermore, following the example of the D language, with a determinant (an article, possessive, etc. preceding the noun), possessive adjectives which are redundant in the S/C language are frequently used by S/C speakers in the Netherlands: Q. No. 740

U mojoj skoli* imam puno drugova.

Prepositions with nouns, instead of case-inflection, are recurrent examples. The preposition od in the sentence below is redundant (it is reminiscent of D usage, though): Q. No. 702

< previous page

Imala sam pet godina kad sam otlisla* iz Jugoslavije tako da se od tih godina mnogo ne sjecam*.

page_111

next page >

< previous page

page_112

next page > Page 112

Text A reduced MT system is most frequently observed through reduced strategies of linking sentences in a text; for instance, a series of sentences may always begin in the same way: in ja (=I), or in mi (= we, members of the family, or we, "Yugoslavs"); subjectrole personal nouns are much more frequently used in D than in S/C (where they may be "hidden"). The argumentation strategy is also reduced, with a very limited number of conjunctions at the users' disposal (or, conjunctions may be misused). One example is provided below for each of the phenomena, the first one also exhibiting verb repetition (the same verb and verbal tense idemo). Q. No. 706 Ja se mnogo radujem odmoru. Jer kad je odmor idemo mi, ja i moja porodica u Jugoslaviju. Mi idemo S.R. Hrvatsku na more u mesto Karin kod Obrovca i Biograda. Onda mi idemo jos * dve nedelje u S.R. Srbiju u selo Kamenica. Standard S would have instead (one possibility): Mnogo se radujem odmoru, jer tada ja i moja porodica idemo u Jugoslaviju. Obicno* odlazimo na more, u mesto Karin kod Obrovca i Biograda, u S.R. Hrvatskoj. Zatim provodimo jos dve nedelje u selu Kamenica u S.R. Srbiji. Q. No. 708 odgovoris*.

Jugoslovenski jezik trebas* dobro da znas* ili kad te neko pita dobardan ti drebas* isto da znas da

The sentence, like most of the examples above too, exhibits several contact phenomena, but to concentrate on the misused conjunction: instead of the disjunctive ili, the adjunctive pa should have been used, and the closest standard S should read: Jugoslovenski (srpskohrvatski) jezik treba dobro da znas, pa kad te netko pozdravi sa "dobar dan", treba da na isti nacin* odgovoris. Semantics (vocabulary) The examples, which are quite numerous, can roughly be divided into two types: S/C adaptation (mostly translation) of entire D phrases, which is the more common of the two, and S/C adaption of individual D words; see an example of the latter in the following sentence:

< previous page

page_112

next page >

< previous page

page_113

next page > Page 113

Q. No. 731

Ja se nadam da ce * doci* dan da cu* se vratiti u svoju domovinu.

In S/C kad should be used in the temporal clause of the above type, instead of da under the influence of the D ''dat". An example of the former type is the use of the D construction "laten maken" in the following sentence: Q. No. 718

Na nasu* skolu* su pustili da graffiti crteze* napravu na jedan zid.

The sentence can hardly be understood (but it still can). In standard S/C it should read: Dozvolili/dopustili su da se na jedan zid nase* skole * nacrtaju grafiti. The underlined sentence below is a literal translation from the D: "Van mamma moet ik . . . ": Q. No. 739 od mame moram.

Ponekad mi se nenada ici* zato sto* me druge drugarice pozovu da sa njima idem se vozati. A

The closest S/C would be: A zbog mame moram (ici). Orthography The common mistakes are S/C adjectives in -ski, -ski*, and * endings, derived from proper nouns, spelled in capital letters (like comparable adjectives in D), against the strict S/C rule of small letters usage: Q. No. 734

Jugoslavija je moja domovina iako ja bolje pricam* Holandski.

There is one example of an almost precise transfer from D: Q. No. 742

. . . u Jugoslaviji nema discriminatija;

the D would have "discriminatie", and the S/C diskriminacije.

< previous page

page_113

next page >

< previous page

page_114

next page > Page 114

Conclusions and discussion The present study shows that the sociolinguistic situation of S/C speaking communities living in the diaspora is quite complex. Not only is a great number of varieties being used, but even the educational support of these varieties is highly different in various locations. Moreover, the study on the compositions written by Croatian and Serbian adolescents living in the Netherlands has shown that their native language varieties are substantially influenced by the second language submersion context. Both intra- and interlingual characteristics of language change could be evidenced at all levels of the linguistic system. From an intralingual point of view, it was found that a restricted input leads to a reduced language variety. Most reductions were counted at the level of syntax. From an interlingual point of view, there was evidence that contact between Croatian and Serbian varieties with Dutch may also lead to interference. For Croatian or Serbian language varieties in Western Europe to survive educational support from the authorities seems required. However, as has been noted by Strauch (1991) the content of mother tongue education is also in question. In the wake of the post-1968 reforms the school curricula for MT teaching have abandoned a good deal of the formal teaching of language structure and concentrated on the communicative and creative use of language. According to Strauch the systematic teaching of the formal and stylistic basis for this communicative ability has become too sporadic. As a consequence, the acquisition of basic language structures, which is possible for the majority of indigenous (in his case, German) children, is inaccessible to migrant children in his home country. Here is his warning: Since the correct use of the standard language nevertheless strongly influences teacher judgements, a vicious circle sets in: the children are, involuntarily, penalized for not knowing something that they have not been taught properly.

< previous page

page_114

next page >

< previous page

page_115

next page > Page 115

Notes 1. When this paper was submitted for publication (early in 1991), Yugoslavia/Yugoslav had a different meaning because the country by that name did exist, it was still kept together, and the appellation was, therefore, justified. A year and a war later (early 1992), the country no longer exists, because two of its constituent republics, Croatia and Slovenia, have gone their separate ways in pursuit of democracy, after free democratic elections and referenda; they are now recognized as independent and sovereign states. The process of democratization continues in the former Yugoslavia. In Croatia, the majority language is Croatian, but "ethnic and national communities or minorities within the Republic of Croatia are entitled to cultural autonomy" (Article 15 of the 1991 Constitution). "Members of all national or ethnic communities of minorities in the Republic of Croatia are entitled to the free use of their language and alphabet, both publicly and privately. In those municipalities where members of a national or ethnic community or a minority represent the majority of the total population, the alphabet and language of that national or ethnic community or minority will be officially used together with the Croatian language and the Latin alphabet" (Article 7 of the Constitutional Law of Human Rights and Freedoms and the Rights of National and Ethnic Communities or Minorities in the Republic of Croatia, 1992). 2. They worked on a longitudinal study of S/C speaking Yugoslav children in Sweden, a comprehensive linguistic research project called JUBA, in the 1980-1983 period. Their findings were mostly published in two Slavica Lundensia books (edited by L. Durovic *): • No. 9 - Lingua in Diaspora (Studies in the Language of the Second Generation of Yugoslav Immigrant Children in Sweden), Slaviska Institutionen vid Lunds Universitet, Lund, 1983, 233 pp.; • No. 11 - Child Language in Diaspora (Serbo-Croatian in West European Countries), the same publisher, Lund, 1987, pp. 731 (M. Stankovski's contribution), and pp. 175-194 (M. Tomasevic*' contribution). The rest of the book contains an abundance of information on Yugoslav migrants' linguistic situation in F.R. Germany (W. Stölting J. Matesic*), Sweden (A. Pavlinic*-Wolf et al.; K. Magnusson), and Norway (S. Mønnesland). In Slavica Lundensia No. 12 - Diaspora Languages in Western Europe (The Workshop on "Language Minorities - Children's Language in Diaspora"), the same publisher, Lund 1988, there is a contribution by W. Stölting-Richert: "Linguistic, Sociological and Pedagogical Aspects of Serbo-Croatian in West-European Diaspora" (pp. 11-21), with a useful bibliography including, among others, some authors who have been studying phenomena of S/C speakers' bilingualism in Europe. For Stölting's classical work in the field (plus Bibliography), see References, 1980. For some recent bibliographies, see: P. Mrazovic* and W. Stölting-Richert, 1989; S. Savic* (ed.), 1989; A. Necak*-Lük, 1988. Among what I consider as the most valuable primary sources, see: R. Katicic*, 1986; D. Brozovic* and P. Ivic*, 1988. A list of further reading can be provided on request.

< previous page

page_115

next page >

< previous page

page_116

next page > Page 116

References COSTA-LASCOUX, J. (1989), Immigrant children in French schools: Equality or discrimination. In: L. Eldering & J. Kloprogge (eds.), Different cultures same school (Ethnic Minority Children in Europe). Amsterdam/Lisse: Swets & Zeitlinger, 61-84. CUMMINS, J. (1988), From multicultural to anti-racist education. In: T. Skutnabb-Kangas & J. Cummins (eds.), Minority Education: From shame to struggle. Clevedon/ Philadelphia: Multilingual Matters, 127-157. DUROVIC *, L. (1988), The concept of diaspora language. In: Slavica Lundensia 12. Lund, 7-9. GASSNER, V. (1990), Contact Bulletin 7, 6, Atha Cliath. Ireland: The European Bureau for Lesser Used Languages. GVOZDANOVIC*, J. (forthcoming), Serbo-Croatian. In: G. Extra & L. Verhoeven (eds.), Ethnic Minorities in the Netherlands. Tilburg: Tilburg University. PAVLINIC*-WOLF, A., K. Brcic* & N. Jeftic* (1988). Supplementary mother-tongue education and the linguistic development of Yugoslav children in Denmark. In: Journal of Multilingual and Multicultural Development IX, 1-2. Clevedon, Avon, 151-167. RINGBOM, H. (1987), The role of the first language in foreign language learning. Clevedon/Philadelphia: Multilingual Matters. ROOSENS, E. (1989), Cultural ecology and achievement motivation: Ethnic minority youngsters in the Belgian system. In: L. Eldering & J. Kloprogge (eds.), Different cultures same school. Amsterdam/Lisse: Swets & Zeitlinger, 85-106. SKUTNABB-KANGAS, T. (1984), Bilingualism or not: The education of minorities. Clevedon, Avon: Multilingual Matters. SKUTNABB-KANGAS, T. (1988), Multilingualism and the education of minority children. In: T. Skutnabb-Kangas & J. Cummins (eds.), From shame to struggle. Clevedon/Philadelphia: Multilingual Matters, 9-44. STÖLTING-RICHERT, W. (1988), see note 1. STRAUCH, R. (forthcoming), Southern European migrants in German schools: A challenge for the educational system. In: Proceedings of the International Conference on "School and Migration in the Europe of the 1990's" (Sitges-Barcelona, November 22-24, 1990). Barcelona: Divisio Ciències Educació, Barcelona University. SKILJAN*, D. (1989), Percepcija varijantnosti hrvatskog ili srpskog jezika. In: Kulturni radnik XLII, 6, Zagreb, 76-79. VEGAR, B. (1987), Neki aspekti u njegovanju govornog i pismenog izraza bilingvalnog ucenika. In: Migracijske teme 3, 1, Zagreb, 93-101. WEINREICH, U. (1968), Languages in contact. The Hague: Mouton. WODE, H. (1988), Einführung in die Psycholinguistik. Ismaning: Max Hueber Verlag.

< previous page

page_116

next page >

< previous page

page_117

next page > Page 117

PART 2 FIRST LANGUAGE ACQUISITION IN A SECOND LANGUAGE CONTEXT

< previous page

page_117

next page >

< previous page

page_ii

next page >

< previous page

page_119

next page > Page 119

5 Turkish language development in Germany 1 Carol W. Pfaff Viewed historically, the development of Turkish and other minority languages in Europe is the outcome of the acquisition and use of these languages by individual members of the minority speech communities. These developments in langue, which typically arise in non-standard varieties, often in the process of attrition, can, in fact, only be inferred from development in paroletaking this to include not only production and comprehension of the minority language, but choice among the available varieties in the verbal repertoire of the speakers. It is clear that social and political factors such as generation, degree of contact with speakers of each language, educational policy, and so on play important roles in combination with psycholinguistic factors to determine the course of acquisition. However, the specific processes by means of which these social and psycholinguistic factors interact are still far from adequately determined. The goal of this chapter is to examine developing varieties of Turkish in Germany in a framework which attempts to integrate some of the most salient linguistic, social, and psychological factors which influence the language development of migrant children. Berlin is a good setting for such an enterprise because it offers a wide range of sociolinguistic settings and, in particular educational environments which result in a wide range of individual variation in children's verbal repertoires, in particular a decided range in language proficiency and preference in Turkish and German. The focus in the present chapter is the nature of the variation between Turkish-dominant and German-dominant children. In particular, I will present aspects of the nominal and verbal morphosyntax and discourse strategies in Turkish from two studies of Turkish / German bilingual children in Berlin. On the basis of data from four children, two from each of the studies, we begin to relate the variation in morphosyntax to strategies in discourse.

< previous page

page_119

next page >

< previous page

page_120

next page > Page 120

The remainder of the chapter is organized as follows: First, the theoretical framework and social parameters of the studies is presented in a schematic model of factors in bilingual language acquisition and use. The next section provides an overview of the two studies, followed by a summary of the major findings regarding the development of both Turkish and German. The body of the chapter is devoted to the quantitative and qualitative analysis of the Turkish of the four children, focusing on some details of prodrop, case marking and verbal morphology, and on the use of these features in narrative and conversational discourse from the KITA study. The final section takes up the issue of the relationship between morphosyntactic development and communicative competence and indicates some directions for further research. Modelling bilingual development The attempt to capture the underlying competence of code-switching and other language-contact phenomena, whether in terms of the (development of) psycholinguistic competence of individual bilinguals, or the pragmatic and socio-historical (development) of varieties of language in use in bilingual communities, is a dauntingly complex undertaking. The sketch of a "model" which I offer here, is surely incomplete (for instance in not explicitly including the dimension of time) and makes no pretentions to being more than an attempt to express some of the interrelationships between a number of external and internal, social and linguistic factors which I believe must be taken into account by any attempt to interpret the messy and complex, yet necessarily incomplete data which any empirical study of language use by bilinguals provides. The model presented here is frankly eclectic. I have borrowed freely from a number of earlier models (Schumann, 1978; Felix, 1982; Dulay, Burt & Krashen, 1982, and Sanchez, 1983) which schematically represent some of the interrelationships among some of the relevant factors in second language acquisition and bilingualism. Likewise, I have adopted terminology from a number of approaches to first language acquisition in the generative framework and from the explicitly crosslinguistic work of Slobin et al. Aspects of conversational interaction, individual psychology and sociopolitical setting are represented informally in the larger framework.

< previous page

page_120

next page >

< previous page

page_121

next page > Page 121

Figure 1 Psychological Cognitive and Social Factors in Bilingual Language Acquisition of Migrant Children

< previous page

page_121

next page >

< previous page

page_122

next page > Page 122

Some relevant exemplifications of those factors which have emerged from research on language acquisition and use by migrant children, in particular from our research on Turkish children in Berlin, are specified in Figure 1. Rather than attempt to go into detail about each component of this model here (see Pfaff, 1988b), I will limit myself to just a few general explanatory remarks. The basic input - output schema is familiar from work in the generative framework and incorporates the assumption that some kind of species-specific and language-specific universals play a role in both first and second language acquisition, though possibly to different degrees, and constrain language contact-phenomena and language change as well. These are the familiar essentials of "the language faculty", shared by all humans and by all (normal) members of any speech community which give rise to the individuals "linguistic competence" in particular varieties of their first, second, etc. languages. The present model, however, explicitly does not abstract away from all performance variation and interaction, as many "purely theoretical" models do. Rather, some of these factors are taken as essential for the appropriate interpretation of the observable data on performance which are to be used as evidence for conclusions about individual competence and the underlying language faculty - if not for an understanding of the linguistic universals in principle. Although the components are assumed to have considerable independence from each other, they are not assumed to be totally independent. Some of the linkages between them are represented in the diagram by feedback arrows (wavy lines) and permeable boundaries (dotted lines). For instance, the interplay between interlocutors' and the child's discourse strategies in interaction is an important non-structural factor which determines not only the specific input available for linguistic processing by the child but appears to be highly significant globally in the development of L1 and L2 competence, as well as in LI/L2 mixing. In the remainder of the section, we will examine some results of Turkish-dominant and German-dominant children from the EKMAUS and KITA studies of Turkish / German bilingual children in Berlin in the light of this model.

< previous page

page_122

next page >

< previous page

page_123

next page > Page 123

The EKMAUS and KITA studies Background and Goals Bilingual issues are of particular importance in Berlin, one of the centres of heavy immigration from southern Europe and the Mediterranean countries over the past 20 years. Further, with recent migration within the European Community and from Eastern Europe, bilingual issues will remain crucial. Obviously, different phases of migration focus attention on different aspects of bilingualism. Among the issues most relevant for the second and third generation children and adolescents of families from Turkey who live in Berlin are those which centre around education, the question of whether, when and how mother tongue instruction is to be provided, the implementation of German as a second language instruction, possible modifications in the instructional methods or materials for instruction to accommodate to the linguistic needs of the pupils. For all of these purposes, a foundation of basic information about the language abilities of the children is essential, but very little empirical evidence on language development of the bilingual children in Berlin has been available. These two projects are an attempt to partially fill this gap as well as to address theoretical issues in bilingualism. 2 In addition to the practical problems associated with bilingualism here, many important theoretical linguistic issues are thrown into sharp focus against the background of this natural language contact situation in which immigrant children spend their formative years. Prominent among these are questions of language development and language use, including: (1) The role of purely linguistic factors in processes of first and second language acquisition, considering cross-linguistic acquisition of typological diverse languages focusing on language-specific vs. general cognitive strategies. (2) The nature and extent of mutual influence of bilinguals' languages on each other, taking into account psycholinguistic separation vs. fusion of linguistic systems, sociolinguistic and linguistic constraints on language mixing and historical developments in the speech community, development of ethnic varieties, convergence, and so on. (3) The effect of social factors, such as the extent and quality of discourse with native speakers of both languages on processes of

< previous page

page_123

next page >

< previous page

page_124

next page > Page 124

language acquisition and processes of language attrition and language loss. Ultimately, our goal is to link all these aspects in an integrated theory of bilingual language acquisition, as sketched out in the model presented in the preceding section. Steps in this direction were taken in the two studies investigating the Turkish and German of bilingual children described below. Sample and Methodology The EKMAUS Study The EKMAUS study was conducted 1983-86 in afternoon day-care centres attended by primary school children from two Berlin districts, Wedding and Kreuzberg. The design was cross-sectional, investigating the speech of bilingual children between the ages of 5-12 with varying degrees of contact to German and differing in the age at which contact with the second language began; monolingual control groups were also included. 3 This study was experimental and employed games with toys, pictures and narratives based on verbal stimuli to elicit oral production, comprehension and metalinguistic judgments of various aspects of nominal reference. It also provided opportunities for free conversation and personal narratives linked to the games. One interview in each language was conducted with the children in their after-school day-care centres, which provided a casual rather than school-like setting for our interviews. (See Pfaff et al., 1985 for a detailed description.) The KITA study The KITA study, in progress since May 1987, is a five-year longitudinal investigation of the speech of preschool and early school-age children who attend(ed) a bilingual day-care centre (''Kindertagesstätte, or "Kita") in Berlin-Kreuzberg, a district with a high proportion of migrant children, especially of families from Turkey. About 90% of the children in the Kita speak Turkish at home. During the time of our investigation, the caretaker staff was composed of equal numbers of native speakers of Turkish and of German who generally followed the policy of speaking their mother tongue to all children. Thus all children were exposed to native adult varieties of both Turkish and German from a very early age, as young as six months in some cases. In addition, they were exposed to a number of

< previous page

page_124

next page >

< previous page

page_125

next page > Page 125

different native and non-native, child and adult varieties of both these (and other) languages. Our sample covered the age range from 2-8 and included Turkish children (n=22), German children (n=5), and children of mixed marriages (n=4), coded T, D and TD, respectively. The focus of the investigation was the development of morphosyntax, language mixing and discourse strategies. Children were recorded, individually or with another child, by teams of two Turkish or two German investigators, who used a variety of toys, books, pictures, etc. and who engaged the child in conversation about these as well as about events and topics which beyond the "here and now" of shared experience as much as possible. Recordings averaging half an hour, longer for the older children, were made separately for each language every monthat the Kita for those preschool children and, for those who had started school, at home, after school. (See Pfaff et al., 1987 for details.) In the following, I take up some of the specific features of the Turkish of the four children below (cf. Pfaff & Savas, 1988 for details on their background). 4 EKMAUS Study: KITA Study:

Turkish-dominant A S2 11 f 9 T16 f 3;11 - 5;8

German-dominant B S 6 04 f 8 T05 m 4;4 - 6;0

Before turning to the features of Turkish for these specific children, the following section summarizes the linguistic development of a larger sample from the longitudinal KITA study. This provides an overview of the development of Turkish / German bilingual children's grammatical and communicative competence in both languages. Early development of Turkish and German Language dominance Although their early exposure to both languages initially led us to expect that the children would display the characteristics of "simultaneous acquisition of two first languages", our analysis of their developing competence to date reveals that their language acquisition patterns differ significantly from those "Type 1" bilingual children of mixed marriages most frequently reported in the literature on early child bi-

< previous page

page_125

next page >

< previous page

page_126

next page > Page 126

lingualism. (See Romaine, 1989.) Instead, we find a pattern of successive first and second language acquisition with clear patterns of language dominance. The majority of the Turkish children are clearly Turkish-dominant. Their acquisition of Turkish proceeds essentially like that reported for monolinguals (Aksu-Koç, 1988, Aksu-Koç & Slobin, 1985, Ekmekçi, 1979, 1986) while their acquisition of German differs strikingly from that reported for monolinguals (Mills, 1985, Clahsen, 1988) and is in some respects similar to the patterns characteristic of natural second language acquisition of German by adults and older children (Klein & Dittmar, 1979, Clahsen, Meisel & Pienemann, 1983, von Stutterheim, 1986, Kuhberg, 1987, and others). For Turkish children who are German-dominant, the acquisitional patterns for Turkish morphosyntax are like those of monolinguals only in some respects, while some structures do not develop to the same extent, if at all, a pattern termed "stagnation" by Verhoeven and Boeschoten, 1986. This stagnation goes along with a tendency to avoid or limit their interaction in Turkish in favor of German, as we will discuss below. The few German children in the Kita, with one exception, do not develop productive competence in Turkish, despite their constant exposure to input from the Turkish caretakers and the predominant use of that language by the other Kita children. Language Mixing The children rapidly develop toward distinct norms in Turkish and German, approximating the requirements of the "monolingual mode" in German, with decreasing admixture of Turkish elements in German but approaching a "community norm" which permits the incorporation of lexical items from German into Turkish. (See Pfaff et al., 1988, and Pfaff, 1990b.) Separation of Systems Early separation of grammatical systems is supported by our study despite the occurrence of a considerable amount of lexical mixing. Systematic typological differences in nominal and verbal inflectional morphology, prodrop and word order develop independently in both languages essentially as they do for monolinguals.

< previous page

page_126

next page >

< previous page

page_127

next page > Page 127

The Turkish-dominant children's inflectional morphology is virtually indistinguishable from that of monolinguals and even the German-dominant children in EKMAUS and KITA studies make fewer errors in morphology in their Turkish than in their German, in subsystems such as case marking, subject-verb agreement and tense-mood-aspect marking. This is not surprising, in light of the fact that Turkish is morphologically regular and transparent, while German is notoriously opaque. The Turkish-dominant children make virtually no errors in case marking in Turkish, while well over half their German NPs are nonstandard in this respect. The weakest Turkish speakers among the German-dominant children produce a maximum of 1015% nonstandard casemarked forms in Turkish, while producing approximately 30% nonstandard casemarked NPs in German. Similarly, for subject-verb agreement, the strongest Turkish speakers make only occasional errors (about 2%) in their Turkish, but up to 80% errors in German. The weakest Turkish speakers make only about 5% errors in Turkish agreement marking, and their proportion of agreement errors in German is also low. For the two children from the KITA study, discussed in detail below, the figures for nonstandard case marking are as follows: T16: Turkish: 5% T05: Turkish: 4%

German: 58% German: 28%

Relating these differences to the differences in the use of case categories will be discussed below. Conversational strategies Children develop notably different conversational strategies in their dominant vs. weaker language. While we are only beginning to analyze this aspect of our results, some preliminary results include differences in conversational features such as topic initiation, reliance on scaffolding, and willingness / ability to converse on topics beyond the "here-and-now". Developments in Turkish morphosyntax In the following sections, I will present more detailed material on some of the features sketched above in the speech of four children drawn from the ends of the language-dominance continuum in both the EKMAUS and KITA studies. EKMAUS A S2 11 and KITA T16 are

< previous page

page_127

next page >

< previous page

page_128

next page > Page 128

clearly Turkish-dominant and EKMAUS B S6 04 and KITA T05 are clearly German-dominant. Prodrop Turkish and German differ typologically with respect to the setting or implementation of the prodrop parameter (PD). Turkish is a prodrop language which permits null objects as well as subjects (see Özsoy, 1987; Kornfilt, 1987; Kerslake, 1987 etc.), and German is a non-prodrop language which permits some null topic in appropriate discourse conditions. (Weissenborn, 1990). As noted before, the patterns of pronominal usage vs. null anaphora develop essentially independently in each of the bilingual child's languages with systematic differences apparent from the outset. The proportion of null realizations for all children at all ages is always significantly greater for Turkish than for German, casting doubt on the hypothesis proposed by Hyams and others that the sentences without subjects in early child language can be simply accounted for in terms of a postulated "initial setting" of the prodrop parameter at +PD. (See Pfaff, 1988c, and Pfaff, in prep.) However, this does not mean that no differences between Turkish-dominant and German-dominant children can be observed. In German, the frequency of nonstandard utterances without overt pronouns is noticeably greater for the Turkish-dominant children. Similarly, in Turkish, the German-dominant children tend to use higher frequencies of overt pronouns where monolinguals and Turkish-dominant bilinguals use null anaphora, as illustrated in (1) and (2), (with overt pronouns in boldface, and personal endings underlined). While (1), produced by the Turkish-dominant A S2 11, is perfectly clear without an overt pronoun, the use of these in (2), produced by the German-dominant B S6 04, is redundant and unusual in this context, a series of requests for specific toys that the target child makes to the doll being addressed in the PLAYROOMS game. (1) A S2 11 f 9 Orhan, san topu alabilir miyim? 'Orhan, may I take the yellow ball?' (2) B S6 04 f 8 Peter, sen bana topu verir misin? 'Peter, will you give me the ball?' Some quantitative results for selected Turkish-dominant and German-dominant EKMAUS and KITA children are given in Table 1.

< previous page

page_128

next page >

< previous page

page_129

next page > Page 129

Table 1 Null anaphora in Turkish for selected children in comparable contexts Turkish-dominant German-dominant Context n % null 5 n % null Ekmaus A S2 11 (19) B S6 04 (f8) 1s, 2s playrooms 15 100% 18 7% subject & indirect (14%)6 object 3s in actions 70 100% 50 93% subject 43 96% 37 73% direct object 3s, 3p animates 8 88% 16 56% in snowman Kita T16(f 3;11-5;7) T05 (m 4;4-6;0) subjects 1s, 2s 96 77% 71 59% 3s 362 81% 214 42% Nominal and verbal morphology As noted before, the bilingual children make relatively few errors in Turkish morphology. This is particularly striking for the suffixes which express case, subject-verb agreement and tense-mood-aspect. Those errors which do occur are characteristically different for Turkish-dominant and German-dominant children. Further distinctions are to be found when one examines the frequency and distribution of the suffixes quantitatively or in context. The following sections illustrate these points with material from our studies. Case marking Quantitative results and examples of the development of case marking in Turkish have been discussed in detail elsewhere (Pfaff, 1988c; Pfaff & Savas, 1988; Pfaff et al., 1988). Here, I will note only that the type, as well as the quantity of the few errors made by the Turkish-dominant and German-dominant bilinguals differ characteristically.

< previous page

page_129

next page >

< previous page

page_130

next page > Page 130

While the errors of Turkish-dominant children tend to involve over-marking of the possessive on nouns as in (3), the errors of German-dominant children involve lack of possessive inflection as in (4) and substitutions of one suffix for another as in (5). (3) T16 4;6 sonra bebegin * kolusunu (=kolunu) koparmak istiyo 'then he wants to take the doll's arm off' Forms such as kolusunu are not "errors" in case marking, per se, but represent overgeneralization of the regular rules to certain lexical items. Such forms are also reported in monolingual pre-schoolers' speech (Ekmekçi, 1986). In (4), T05 produces a genitive/possessive marked by a possessive pronoun but no possessive suffix on the possessed noun: (4) T05 4;10

benim para (=param) var 'I have money'

While examples such as (4) could be claimed to be evidence of influence from German, or of a tendency toward analyticity, this generalization is not supported quantitatively. We find no general trend toward analyticity in other cases. Whether or not the substitution of dative for locative in (5) represents transfer from German is questionable.7 (5) T05 5;0 dügüne(=dügünde)* Eis var [DAT -e for std LOC -de] 'there's ice cream at the wedding! In (6), however, the substitution made by the German-dominant EKMAUS child involves suffixes which have no German equivalents. (6) B S6 04 f 8

(Home language)

Ben annemden babamdan Almanca konusuyom ama, annem babam benlen Türkçe konusuyo*. ... bazenleri anlamiyom.8 (std. = annemle babamla [ablative -den/-dan for std. comitative -le/-la] 'I speak German from (=with) my mother and father but my mother and father speak Turkish with me. ... sometimes I don't understand.'

< previous page

page_130

next page >

< previous page

page_131

next page > Page 131

Figure 2 Case Function of NPs in Turkish and German for T16 & T05

< previous page

page_131

next page >

< previous page

page_132

next page > Page 132

Figure 2 summarizes the use of NPs in both Turkish and German by two children over a period of nearly a year, T16 (3;10-4;9) and T05 (4;4-5;2). The overall length of the bars in the graph corresponds to the number of instances of NPs with lexical heads produced by each child in each language and the size of the patterned segments within each bar represents the percentage of the indicated semantic / case functions. 9 Language dominance is reflected in the proportion of overt NPs in Turkish and German. T16 has more NPs in Turkish than in German, while T05 has more NPs in German than in Turkish. The typological difference in the prodrop parameter is reflected by the relatively smaller proportion of subjects to objects for Turkish than for German. This difference is much more prominent in the speech of Turkish T16 than in the German-dominant T05, reflecting convergence, but not merger, of the two linguistic systems. As noted above, case marking errors in Turkish are rare for all children. However, as shown in Figure 2, the distribution of overt NPs over case function differs systematically for Turkish-dominant and German-dominant children. The proportion of naming (NP-essives and cop-essives) is higher in the non-dominant language. The fact that nouns in these functions are uninflected in Turkish partially accounts for the low proportion of errors in case marking in Turkish. These observations about the distribution of overt NPs actually reflect differences in the discourse strategies employed by the children in their dominant and non-dominant languages. We will return to this topic in the next sections. Verb morphology In Pfaff et al., 1989, we discuss the development of verbal morphology in both Turkish and German in detail for a subsample of ten children. In this section selected aspects of the results for finite and non-finite forms are presented. With respect to finite forms, the use of -yor, -DI and -mIs* was analyzed.10 Briefly, -yor refers to ongoing action while the two suffixes DI and -mIs refer to the past from different perspectives. -DI is used to report witnessed events while -mIs is evidential, used to report states or inferences from past events not directly witnessed. (See Slobin & Aksu,

< previous page

page_132

next page >

< previous page

page_133

next page > Page 133

1982; Aksu-Koç, 1988 for more precise characterization of their functions in adult and child speech). The use of non-finite forms reflects one of the most striking typological differences between the grammatical systems of Turkish and German (or other Indo-European languages). Turkish employs the device of non-finite verb suffixes to express grammatical relations between clauses expressed by subordinating conjunctions or adverbials in German or English. There are several non-finite verb forms, variously termed "deverbal adverbs", "gerunds" or ''converbs" in the literature on Turkish. In their treatments of the development of these structures in Turkish child language, Slobin (1988), Aksu-Koç (1988), and Verhoeven (1988) use "gerunds" while Boeschoten (1990a) uses "converbs". Slobin (1988: 28) summarizes the structures and meanings of the four most common gerund suffixes as follows: X-Ince Y X-Erken Y X-ErEk Y X-Ip Y

'when X, Y; as soon as X, Y' 'while X-ing, Y' '(in, by) X-ing, Y' 'X and (then) Y; having X-ed, Y'

As far as acquisition of these finite and non-finite forms is concerned, studies of monolinguals show that they acquire finite before non-finite forms and these are acquired in a sequential order, related to the cognitive and linguistic complexity. The first forms, -IncE and -Erken are reported quite early. Slobin (1988: 30) reports these in spontaneous speech samples of 2-year-olds and Aksu-Koç (1988: 67) first in narratives at age 3. Bilinguals in the Netherlands show extremely restricted use of gerunds of any type. Verhoeven (1988: 448-9) reports only three instances of the use of participles and gerunds in narratives by eight children, interviewed twice at the ages of 7 and 8. There were no instances at age 7, and only three instances at age 8. Instead, the children made extensive use of simpler adverbial cohesive devices, typically used by 5-year-old monolinguals in Turkey. Verhoeven suggests that this "stagnation" is related to restricted input in the second language environment and that the use of adverbials is a "compensation strategy" used to bring about text cohesion. Boeschoten's results for twelve younger children, interviewed three times at ages 4, 5 and 6, show considerable individual variation (Boeschoten 1990a: 100-120). Only five of the children used any

< previous page

page_133

next page >

< previous page

page_134

next page > Page 134

gerund forms, while all used adverbials, usually in combination with non-finite gerunds or with complex verb finite forms. He also reports semantic overextension of -Erken in some of the children who used it Table 2a Finite and non-finite Verbs in Turkish (T16) imp opt er yor DI mIs * EcEk sE deriv nonfin AGE 3 0 0 5 4 2 2 0 l(p) 0 3;11 4;3 4 0 0 29 43 7 9 0 2(p) 4(ken, 1(m) IncE) 4;5 0 0 0 16 9 3 2 1 3(p) 5(ken) 4;6a 2 0 0 1 5 24 0 0 1(m) 3(ken, IncE) 4;6b 10 3 1 51 24 29 1 0 1(p) 5(ken, 2(m) IncE) 4;9 3 2 0 59 49 1 25 0 1(m) 6(Ip. IncE) 5;1 1 3 1 51 19 4 12 0 6(p) 12(ken, IncE, Ip) 5;2 0 0 0 21 4 9 1 0 2(r) 2(Ip) 5;3 0 0 0 19 3 9 0 0 0 1 (ErEk) 5,4 0 1 0 69 11 15 0 0 4(p) 4(ken, 1(c) Ip) 5;5 7 5 3 52 39 51 6 0 3(m) 3(Ip) 1(r) 3 1 0 44 15 85 2 0 1(r) 4(ken, 5;6 l(c) Ip) 3 5 3 17 3 28 11 l 1p 3(ken) 5;7 l(c) 4 3 0 13 27 16 0 0 1(p) 1(IgI)* 5;8 [aldigi* zaman]

< previous page

page_134

next page >

< previous page

page_135

next page > Page 135

Table 2b Finite and non-finite Verbs in Turkish (T05) AGE

imp 2

opt er yor DI 3 10 4 i

mIs * 0

EcEk 0

sE 0

deriv 1(p)

4;4

nonfin 0

5

5

0

25

9

1

12

2

3(p) 3(c)

4;8

0 10

1

1

4

2

0

5

0

2(m)

4;10

0 17

3

1

62

15

1

8

3

2(p) 2(c)

5;0

0 0

2

5

9

17

0

1

0

2(m)

5;2

0 2

2

0

15

11

0

4

0

2(m)

5;5

0 8

1

0

27

25

2

6

0

3(p) 1(c)

5;7

0 4

2

3

61

35

3

11

1

1(p) 1(m)

5;9

0 4

3

0

26

8

0

3

0

2(c) 1(m)

5;10

0 15

2

1

46

29

3

4

2

1(m)

5;11

0 6

2

1

6;0

27

9

2

1

0

0 0

In our studies, we find the gerund forms present in the speech of Turkish-dominant bilinguals while German-dominant bilinguals do not use these forms at all. Our results indicate that the Turkish-dominant children in Berlin more nearly approximate the language behaviour of monolinguals than do Turkish children in the Netherlands, whose Turkish is more like that of our German-dominant children. Tables 2a and 2b give an overview of the development of verb morphology over a two-year period for two Kita children, the Turkish-dominant T16 and the German-dominant T05. The table represents inventories of all the verb forms used in the Turkish interviews. Turning first to the relative frequency of the finite forms -DI, -mIs, and -yor, note that -mIs forms are notably less frequent for the German-dominant T05 than for the Turkish-dominant T16. However, it is not the case that T05 has substituted the wrong suffix where -mIs forms would be expected. Nor has he simply omitted all indication of tense-aspect. Instead, he has described the scene from a different perspective,

< previous page

page_135

next page >

< previous page

page_136

next page > Page 136

using -DI past or -yor for ongoing activity, or -EcEk for future, all of which are appropriate (and well-formed) in this context. Second, and even more striking is the fact that while the Turkish-dominant T16 uses non-finite forms throughout this period, the German-dominant T05 uses no finite forms at all. As with the results for case marking, these results for verb morphology are linked to the children's conversational strategies, as we will see in the next section. Conversational strategies In this section we examine two dimensions of conversational strategies employed by the children: (1) local strategies related to the perspective taken by the children in producing narratives, and (2) global strategies related to the children's interaction with the interlocutors. Local strategies: narrative perspective To illustrate the narrative perspectives and strategies taken by the two children, consider the following parallel examples from the picture book, Lady and the Tramp: 11 Jim gives his wife a Christmas present (7) T16: burda Weihnachtsmannbaum yapmislar* 'here 0 (= Jim Dear and Darling) made a Santa Claus tree' (8) T05: bu Weihnachten diye 'this (= Christmas tree) (is) for Christmas' In (7) and (8) both children focus on the Christmas tree which is in the background. T16 refers to the presumed prior actions of the couple decorating the tree. She uses a -mIs form denoting non-witnessed action. T05 in contrast simply identifies the tree's function in a sentence without an overt verb. Lady and Tramp chase chickens (9) T16: burda da köpek xxx istiyomus, bunlari 'and then here the dog xxx wanted these'

< previous page

page_136

next page >

< previous page

page_137

next page > Page 137

(10) T05: o zaman bu böyle , bak! bu bunu yicek diye. 'then this (=Tramp) did like this, look! because this (=Tramp) is going to eat this (=chicken). In (9) and (10), the picture represents an action. T16's version expresses the motivation for the action while T05's version 'did' is combined with emphasis on witnessing the event, with focuses on the actions itself. Note that T05's use of -DI form the imperative bak 'look!' directed to the interlocutor. His version of the motivation is expressed by use of the future form, yicek 'he is going to eat'. In the dog pound (11) T16: burda da kafeste 'and here 0 (= the dogs) stay in the cage' köpekler kafeste , 'the dogs were in the cage' (12) T05: o zaman bu ko- bunlar korkuyola. 'then this(=Lady), these(=Lady and Tramp) are afraid.' In (11) and (12) the picture represents the dogs behind bars in the pound. Here, T16 refers to the resultant stage again using mIs * forms while T05 refers to the ongoing emotional state of the dogs. Summarizing the effects that these differences in narrative perspective have on the children's use of grammatical forms, we note the following: 1. prodrop: T16 uses either null subject and plural marked verb or plural subject and unmarked verb; T05 uses pronouns, here, deictic bu, bunlar. 2. case marking: T16 uses a locative, T05 doesn't mention the location. 3. verb morphology: T16 uses -mIs, T05 uses -yor or -DI, -Ecek. The differences are not of grammaticality, but of narrative perspective. 4. subordination marked on verb: Used by T16, not by T05.

< previous page

page_137

next page >

< previous page

page_138

next page > Page 138

5. mixing: both T16 and T05 use a German loan for the culturally-specific Christmas tree. T16's version shows that she has not analysed (or has reanalysed) the German lexical item, while T05's is appropriate. 6. temporal connective: T16 uses da, T05 uses a nonstandard extension o zaman. 7. causal connective: T05 diye Global strategies: participation in discourse We turn now from the choice of forms within a given language to the children's choice between languages and their interactional styles in discourse. From our observations of the KITA children, from informal observations reported by parents and caretakers and from self-reports of language choice by the EKMAUS children, it is clear that linguistic dominance correlates with preferred language used in everyday interactions in their families, with friends, etc. In contrast, in the conversations we recorded with them, language choice was limited; it was understood by the children that they were expected to use one language in so far as possible, Turkish with the Turkish interviewers, German with the German interviewers. Our preference for the "monolingual mode", as Grosjean terms it, in our play sessions enabled us to investigate the relationship between everyday interaction and the development of the children's linguistic competence in both their languages. We approached this question by examining the types of discourse structure characteristic of the conversations we have recorded with the children in their stronger and weaker languages in different elicitation contexts, focused on the "here and now" in structured interactions, games with toys or looking at picture books, or in conversation drawing on the child's own experience, fantasies or opinions on a variety of topics beyond the "here and now". Not surprisingly, the child's willingness and ability to engage in such conversations was more likely to take a larger proportion of the interview time in the interactions in the child's preferred, or dominant, language. Even in the more structured "here-and-now" oriented interactions, children were more able to direct the conversation themselves in their dominant language, while in the

< previous page

page_138

next page >

< previous page

page_139

next page > Page 139

other language they tended to rely much more on scaffolding provided by the interviewers. Table 3 gives a schematic overview of our preliminary findings. Table 3 Conversational strategies in Turkish and German Turkish-dominant German-dominant T16 (3;11-5;8) T05 (4;4-6;0) Turkish German Turkish German Reliance on much much little scaffolding little Detail frequent rare rare frequent some few Questions few few (for vocab) (for info) Topic few very few rare some initiations many (to many (back many (to many (to Digressions convers.) to activity diff. act.) convers.) Ease of elic. easy difficult difficult easy free conv. whenever possible Delegating to interfrequently viewer or turn to rare rare to sister another mother Although exemplification of all of these strategies would go beyond the scope of the present chapter, the excerpt from the German-dominant T05's Turkish version of the story of Lady and the Tramp given below contains illustrative instances of several of the interactional features summarized in Table 3 as well as of the linguistic features discussed in the preceding section. T05 = m 6;0 TSA = interviewer: Scene: Aunt Sara chases Lady. *T05: 1

(a)ma bu köpek, köpek+\

< previous page

page_139

next page >

< previous page

page_140

next page > Page 140

*TSA: ++mhm, köpek de o odada duruyor, bak üstünde, di+mi, burda da var, çantasi falan. *T05: 2

bu kim?

? bak simdi *, adamla ähm o +koca gezmeye . çocuk diye evde *TSA: kim bu gelmis*. bu da kimmis* biliyor+musun, bu teyzesiymis*. teyze gelmis çocuga*, küçük bebege* bakmak için. bu evde bir köpek oldugunu* bilmiyormus* bu teyze. ne olmus* simdi*? ama Lady'i *T05: 3

dövüyo.

*TSA:

aa, neden dövüyo?

*T05: 4

&bu &bu bunu hiç &g görmedi diye.

*TSA: ?

hiç görmedi , zannediyor ki bir köpek geldi ve de bebegi* di+mi? onu korumak istiyor köpek. Lady. neyle dövmek istiyor bak?.

. halbuki Lady bebegi

*T05: 5

Besen@s [:=t süpürge] [*].

*TSA:

süpürge

*T05: 6

Lady? bu da korkmus*.

*T05 1

but this dog, dog+\

*TSA

mhm and the dog is standing there, look on the bed, right? and there are the woman's things, her bag and stuff.

*T05 2

who's this?

eline,

havaya #

Lady?

*TSA who's this woman? look now, the man and woman uh the married couple are going out. this woman has come so the child won't be at home alone, and this woman, do you know who this is?, she's this man's aunt. the aunt has come to look after the child, the little baby. but she doesn't know Lady, she doesn't know there's a dog at home, this aunt. look, what's happening now? *T05 3

0(=she)'s hitting 0(=dog).

*TSA

aa, why is 0(=she) hitting 0(=her)?

1999*T05 4

because this - this hasn't ever s-seen (= 'met') this.

*TSA

0(=she)'s never seen 0(=her), 0(=she) doesn't know 0(=her) at all. 0(=she) thinks a dog came from outside

< previous page

page_140

next page >

< previous page

page_141

next page > Page 141

and will bite the baby. but does Lady bite the baby? 0(=she) won't bite 0(=it), will she? 0(=she) wants to protect it(=baby), the dog. Lady. what does O(=aunt) want to hit 0(=dog) with? look! *T05 5 Besen (German 'broom') *TSA 0(=aunt) took a broom in her hand, raised it in the air # what did Lady do? *T05 6 Lady? and this(=Lady) is scared. In order to really appreciate its significance, it should be contrasted to T05's very different style in the German interviews and to the style used by Turkish-dominant children such as T16. (See Pfaff et al., 1991 for such a comparison). It would also be important to contrast the interviewers' speech, as they interact with Turkish-dominant and with German-dominant children. However, such a complete analysis is beyond the scope of this chapter. I limit myself here to listing a few characteristics of each of T05's turns in the passage above to illustrate the linguistic and interactional features we have been considering: 1. T05 uses deictic reference, breaks off his turn and lets the interviewer complete the statement. 2. Again, T05 delegates his turn to the interviewer by means of a question, leading to a new sequence rather than continuing the topic which the interviewer has suggested. 3. T05 responds to interviewer's use of -mIs * forms with -yor, referring to the ongoing action in the picture rather than the state of events in the story. 4. Again, deictic reference. The use of -DI past in görmedi 'she didn't see her' seems inappropriate here, since the statement is an inferential use of -mIs. The lexical item, görmek 'see' actually an inference, paraphrasing the interviewer's use of 'know, be acquainted with'. Note that T05 does not follow the is also less appropriate than the interviewer's model, interviewer's model explicitly, though he does incorporate the content she provides. 5. T05 uses a German lexical item, Besen, to identify the broom. The structure, an NP essive, does not occur with the German indefinite

< previous page

page_141

next page >

< previous page

page_142

next page > Page 142

article, ein Besen, which would be typical of his German production. He has not code switched into German here but is still continuing to comply with the task, using Turkish in the Turkish interview, even though he is aware of the fact that the interviewer is also bilingual. 6. This instance of korkmus *, one of T05's two -mIs* forms, (the other is the same lexical item, a few lines later) occurs in direct response to the interviewer's use of mIs forms in her question. Conclusions The type of analysis illustrated above offers a promising approach to the issue of interrelating discourse interactional features with the acquisition and use of linguistic forms, structures and systems. Similar strategies surely are characteristic of all types of language acquisition, but such an integrated approach is particularly appropriate to a study of acquisition and attrition in very heterogeneous bilingual communities such as the Turkish population in Germany. In the preceding examination of the Turkish speech of Turkish and German-dominant bilinguals, we have focused on several structural aspects of their grammars: structural inventory, frequency of alternative realizations, and errors. Of these, we have seen that actual errors are relatively infrequent even in German-dominant children's speech, but that there are clear differences in the inventory of structures used and in the frequencies with which the various alternatives are employed. Further, as we have seen, there is a relationship between the quantity and distribution of the actual linguistic forms used by the children and the overall discourse strategies they employ. Underlying both are the joint aspects of competence; grammatical competence and more global communicative competence. Notes 1. Portions of this paper were presented at the Workshop on Ethnic Minority Languages in Europe. Tilburg, December 14, 1990, and at the Third Colloquium on Language Acquisition "Peer communications: functions and forms" Akademie der Wissenschaften, Berlin, 22.1.1991.

< previous page

page_142

next page >

< previous page

page_143

next page > Page 143

2. The EKMAUS study, Linguistische und kognitive Entwicklung: Die Beziehung zwischen Erst- und Zweitspracherwerb, was undertaken in connection with a group of projects funded by the Freie Universität Berlin during the years 1983-1986 under its Forschungsschwerpunkt EKMAUS (an acronym for the Entwicklung von Konzepten und Materialien für die Förderung ausländischer Kinder und Jugendlichen im schulischen und außerschulischen Bereich 'Development of concepts and materials for support of foreign children and adolescents in school and non-school contexts'). The KITA study, Natürlicher bilingualer Spracherwerb von Kita-Kindern: Vom Krippenalter bis zu den ersten Grundschuljahren, was funded by the Deutsche Forschungsgemeinschaft (DFG), under its ''Language Acquisition" research program, 1987-1992. 3. EKMAUS study informant groups are as follows: A = early childhood in Germany, little German peer contact (n=20) B = early childhood in Germany, more German peer contact (n=20) C = migrated after two or more years of school in Turkey (n=6) D = Turkish monolinguals in Ankara (n=15) E = German monolinguals in Berlin (n=13) 4. The codes convey some significant information about the children's sociolinguistic background: the amount of contact to German peers (Groups A or B) and the particular day-care centre they attended at the time of the interviews. S2 was attended by all Turkish children, but had both Turkish-speaking and German-speaking caretakers. S6 was ethnically mixed and had only German-speaking caretakers. 5. Percentages are of null/null + pronominal realization (excluding NPs with lexical heads for 3s and 3p). 6. 4 (of 5) sentences with null subjects were repetitions of the interviewer's model. 7. The standard German equivalent sentence requires a preposition denoting locality, plus the dative: "auf/bei der Hochzeit gibt's Eis". 8. benlen (=std. benle) is colloquial, bazenleri (=std. bazen) is nonstandard. 9. Semantic / case categories used are defined in Pfaff et al. 1988. For Turkish, in which the case markings more nearly correspond to underlying semantic functions, we have used the traditional case labels for the marked cases. For German, where the correspondences are more complex, we used semantic labels. In both languages, we distinguish naming from other uses of the nominative such as agent, experiencer, and so on. NP essives are isolated nouns, possibly with indefinite articles, while Cop(ular) essives are nouns used as predicate nominals. The latter are realized without overt copula forms in Turkish and, frequently, though nonstandardly, in German. (See Pfaff et al. 1988 for examples and discussion.)

< previous page

page_143

next page >

< previous page

page_144

next page > Page 144

10. conventions of Turkish morphophonemics vowel harmony:I = i, , u, ü E =e,a consonant harmony (voicing assimilation): D = d, | 11. Walt Disney's Lady and the Tramp Ladybird Books. This is a picture book (with text in English - which none of the children in the KITA study could read). The protagonist is the dog, Lady, who lives with a young couple, "Jim dear" and "Darling". When they go away, Aunt Sara comes to stay with their baby and chases Lady out. Lady meets another dog, Tramp, with whom she has some adventures before returning home with him to a happy ending. References AKSU-KOÇ, A. (1988a), The acquisition of aspect and modality. The case of past reference in Turkish. Cambridge: Cambridge University Press. AKSU-KOÇ, A. (1988b), Simultaneity in children's narratives: The development of cohesion in discourse. In: S. Koç (ed.), Studies on Turkish linguistics. Ankara: Middle East Technical University. AKSU-KOÇ, A. & D. Slobin (1985), The acquisition of Turkish. In: D. Slobin (ed.), The crosslinguistic study of language acquisition, Vol. 1: The data. Hillsdale, N.J.: Erlbaum. BOESCHOTEN, H. (1990a), Acquisition of Turkish by immigrant children: A multiple case study of Turkish children in the Netherlands aged 4 to 6. Tilburg: Tilburg University (PhD. Thesis). CLAHSEN, H. (1988), Normale und gestörte Kindersprache: linguistische Untersuchungen zum Erwerb von Syntax und Morphologie. Amsterdam: Benjamins. DORIAN, N. (1981), Language Death. The Life Cycle of a Scottish Gaelic Dialect. Philadelphia: University of Pennsylvania Press. EKMEKÇI, Ö. (1979), Acquisition of Turkish: A longitudinal study on the early language development of a Turkish child. Unpublished doctoral dissertation, Austin: Texas University. EKMEKÇI, Ö. (1986), The developmental errors in the preschool Turkish children's speech. In: Proceedings of the Turkish Linguistics Conference. Istanbul: Bogaziçi Univ. Publ. EXTRA, G. & L. Verhoeven (1990), Ethnic minority research in the Netherlands. Cross-linguistic perspectives on Turkish and Moroccan communities abroad. Paper presented at the Workshop on Ethnic Minority Languages in Europe. Gilze-Rijen, December 1990. FELIX, S. (1982), Psycholinguistische Aspekte des Zweitspracherwerbs. Tübingen: Narr. JASPAERT, K., S. Kroon & R. van Hout (1986). Points of reference in first-language loss research. In: B. Weltens, K. de Bot & T. van Els (eds), Language Attrition in Progress. Dordrecht: Foris, 37-49. JOHANSON, L. (1971), Aspekt im Türkischen. Uppsala.

< previous page

page_144

next page >

< previous page

page_145

next page > Page 145

KARDAM, F., I. Temuçin, S. Cinar & C.W. Pfaff (1988b), Migrants from Turkey in West Berlin. Status. Legislation. Language Rights and Approaches to Bilingual Education. Presented at the Conference Minority Rights and Minority Education, Cornell University, May 7-9, 1988. KERSLAKE, C. (1987), NP-deletion and pronominalization in Turkish. In: H. Boeschoten & L. Verhoeven (eds.), Studies on Modern Turkish. Proceedings of the Third Conference on Turkish Linguistics. Tilburg: Tilburg University Press. KORNFILT, J. (1987), Turkish and the Turkic languages. In: B. Comrie (ed.), The world's major languages. London: Croom Helm. LAMBERT, R. & B. Freed (eds). The Loss of Language Skills. Rowley, MA: Newbury House. ÖZSOY, S. (1987), Null subject parameter and Turkish. In: H. Boeschoten & L. Verhoeven (eds.), Studies on Modern Turkish. Proceedings of the Third Conference on Turkish Linguistics. Tilburg: Tilburg University Press. PFAFF, C.W. (1987), Processes of First and Second Language Acquisition. Cambridge, MA: Newbury House. PFAFF, C.W. (1988a), Linguistic and social determinants of Turkish/German bilingualism of migrant children in Berlin. In: L. Durovic (ed.), Diaspora languages in Western Europe. Lund. PFAFF, C.W. (1988b), Zweitsprachenerwerb aus sozio- und psycholinguistischer Sicht am Beispiel türkischer Kinder in Deutschland, Beitrag zum Symposium: Modelle und Entwicklungs-förderung der Zweisprachigkeit. Bad Homburg 11-13 July. PFAFF, C.W. (1988c), The development of nominal reference in Turkish/German bilingual children. Paper presented at the Symposium on Dimensions of Bilingual Development. Hamburg: Hamburg University. Sept. 8-10, 1988. PFAFF, C.W. (1989), Use of Turkish by migrant children in Berlin (West). Paper presented at the Second Workshop on Turkish in North-Western Europe. Tilburg, January 23-24, 1989. PFAFF, C. W. (1990a). Connatural and abnatural change in an unnatural system: the case of case-marking in German. In P. Mülhäusler, et al. (eds.) Development and Diversity: Linguistic Variation across Time and Space. Summer Institute of Linguistics and The University of Texas at Arlington. PFAFF, C. W. (1990b). Mixing and Linguistic Convergence in Migrant Speech Communities: linguistic constraints, social conditions and models of acquisition. Paper presented at the European Science Foundation Workshop on Code-switching and Language Contact: Constraints, conditions and models. London, Sept. 1990. PFAFF, C. W. (1991). Turkish in contact with German: Language maintenance and loss among immigrant children in West Berlin. International Journal of the Sociology of Language, 90, 97-129. PFAFF, C. W. & O. Çakarcan (1989). Nominalreferenz in Erzählungen. Türkisch-Deutsches Sprachwissenschaftliches Symposion. Ankara. PFAFF, C. W., F. Kardam, J. Voß, H. Freesemann, T. Savas *, and M. Thiel (1987). Bilingualer Spracherwerb. DFG Projekt Arbeitsbericht I. PFAFF, C. W., F. Kardam, J. Voß, H. Freesemann, T. Savas, and M. Thiel (1988). Bilingualer Spracherwerb. DFG Projekt Arbeitsbericht II. PFAFF, C. W., F. Kardam, J. Voß, O. Çakarcan, H. MacKerron, T. Savas, and M. Thiel (1989). Bilingualer Spracherwerb. DFG Projekt Arbeitsbericht III.

< previous page

page_145

next page >

< previous page

page_146

next page > Page 146

PFAFF, C. W., F. Kardam, T. Savas *, M. Thiel and O. Çakarcan (1991). Conversational strategies of Turkish/German bilingual children. Paper presented at the 3. Kolloquium zum Spracherwerb Peer communications: functions and forms Akademie der Wissenschaften, Berlin, 22.1.1991. PFAFF, C. W., O. Çakarcan, F. Kardam. T. Savas, and M. Thiel (1990). Bilingualer Spracherwerb. DFG Projekt Arbeitsbericht IV. PFAFF, C. W. & T. Savas (1988). Language development in a bilingual setting: the acquisition of Turkish in Germany. In S. Koç (ed.) Studies on Turkish Linguistics. Ankara: Middle East Technical University. REHBEIN, J. Sprachloyalität in der BRD? Ausländische Kinder zwischen Sprachverlust und zweisprachiger Erziehung. OBST ROMAINE, S. (1989). Bilingualism. Oxford: Basil Blackwell. SANCHEZ, R. (1983). Chicano Discourse: Socio-historic Perspectives. Rowley, MA: Newbury House Publishers, Inc. SCHUMANN, J. (1978). The Pidginization Process. A Model for Second Language Acquisition. Rowley, MA: Newbury House Publishers, Inc. SHARWOOD SMITH, M. (1989). Crosslinguistic influence in language loss. In Hyltenstam & Obler (eds.), Bilingualism across the Lifespan: Aspects of Acquisition, Maturity and Loss. Cambridge: Cambridge University Press. pp. 185-201. SLOBIN, D. I. (1977). Language change in childhood and history. In J. Macnamara (ed.) Language Learning and Thought. New York: Academic Press. SLOBIN, D. I. (ed.) (1985). The Crosslinguistic Study of Language Acquisition, Vol. 1 & 2. Hillsdale, NJ: Lawrence Erlbaum. SLOBIN, D. I. (1988). The development of clause chaining in Turkish child language. In S. Koç (ed.) Studies on Turkish Linguistics. Ankara: Middle East Technical University. SLOBIN, D. I. (1990). The development from child speaker to native speaker. In Stigler, Shweder & Herdt (eds.) Cultural Psychology: Essays on Comparative Human Development. Cambridge: Cambridge Univ. Press. pp. 233-256. SLOBIN, D. I. & A. Aksu (1982). Tense, aspect and modality in the use of the Turkish evidential. In P. Hopper (ed.) Tense Aspect: Between Semantics and Pragmatics. Typological Studies in Language, Vol. 1. Amsterdam: John Benjamins. SLOBIN, D. I. & A. Talay (1986). Development of pragmatic use of subject pronouns in Turkish child language. In A. AksuTaylan (eds.) Proceedings of the Turkish Linguistics Conference. Istanbul: Bogaziçi* University. Koç & E. TEKINAY, A. (1987). Sprachvergleich Turkisch-Deutsch. Möglichkeiten und Grenzen einer kontrastiven Analyse. Wiesbaden: Verlag Dr. Ludwig Reichert. VERHOEVEN, L. (1988). Acquisition of discourse cohesion in Turkish. In S. Koç (ed.) Studies on Turkish Linguistics. Ankara: Middle East Technical University. VERHOEVEN, L. & H. Boeschoten (1986). First language acquisition in a second language submersion environment Applied Psycholinguistics 7:241-55. WEISSENBORN, J. (1990). Functional categories and verb movement: the acquisition of German syntax reconsidered Linguistische Berichte Spracherwerb und Grammatik Sonderheft 3:190-224. YAVAS*, F. (1980). On the meaning of the tense and aspect markers in Turkish. Unpubl. Ph.D. dissertation. University of Kansas.

< previous page

page_146

next page >

< previous page

page_147

next page > Page 147

6 Turkish language development in The Netherlands Anneli Schaufeli In the sixties, Turkish migrant workers started to come to the Netherlands. They intended to stay only temporarily. The jobs they had did not require much proficiency in Dutch and the government did not take too much trouble to organize or facilitate second-language courses. Consequently, the majority of this first generation of Turkish immigrants is predominantly Turkishspeaking. No data are as yet available on language loss within this group but, if this appears to be the case at all, it is expected to be only modest (Jaspaert & Kroon, 1990). The children of these immigrants are generally born in the Netherlands. Considering their limited first-language (L1) input, these children of the second generation will probably not acquire their L1 in all its aspects. Their home language is generally Turkish. Driessen (1990) reports that approximately 93% of several hundreds of Turkish children participating in his research, predominantly speak Turkish with their parents and 43% also speak Turkish with their brothers and sisters. Apart from this and, in most cases, 2.5 hours of first-language education at primary school, almost no L1 input is available. From earlier research (Verhoeven & Boeschoten, 1986; Verhoeven, 1988, 1989) it appeared that the first-language development of Turkish children in the Netherlands is indeed stagnating. Up until the age of five their L1 seems to be on almost the same level compared to their monolingual peers in Turkey. However, from the age of five onwards, when they have entered primary school, their L1 developmental curve levels off much stronger than that of their peers in Turkey. This suggests that certain linguistic devices occurring in the later stages of language development may never be acquired in this specific situation. Therefore, when looking at Turkish in the Netherlands from a language attrition or change perspective, one is led to the assumption that the major cause for such a possible attrition/change process must be found in an incomplete transmission of the language from the first to

< previous page

page_147

next page >

< previous page

page_148

next page > Page 148

the second generation. Such an attrition process between generations will be called language erosion, whereas loss within individuals - as might be the case in first-generation immigrants - will be called language attrition. In the present chapter, some findings will be presented of a research project which aimed at finding the particular aspects or areas in the L1 of Turkish children living in the Netherlands which suffer from such language erosion (Schaufeli, 1991b). In this research the language of eighteen Turkish children with an average age of 11;5 was examined. Most of these children were born in the Netherlands, a few of them arrived there at the age of two or three. They all had their whole primary education, starting at the age of four, in the Netherlands and are Dutch-Turkish bilinguals. Eleven-year-olds are assumed to have completed their language acquisition, at least as far as the formal linguistic aspects are concerned. Therefore their language may be tentatively regarded as the product of the assumed incomplete transmission of the L1 from the first generation to the next. This group of children participating in the research will be referred to as the N-group. As a reference point, a monolingual group in Turkey with an average age of 11;3 and similar socio-economic and dialect backgrounds was incorporated into the research design. This group - henceforth the T-group - theoretically represents the level of language proficiency of the first-generation parents at this age. Several linguistic levels were incorporated in the study, i.e., the lexical, the morphosyntactic and the discourse level. Of course no exhaustive description could be given, but several aspects of these various levels were studied in order to cover a wide range of areas. The main issue addressed in the present chapter is the degree to which the various linguistic levels are affected by language erosion. Therefore the proficiency of both groups of children on each linguistic level will be discussed on the basis of the scores on two general measures, and in order to find out which linguistic level is most affected, scores on the various linguistic levels will be compared. Design of the study The research findings presented in this chapter are mainly based on semi-spontaneous speech data. These consist of eleven stories told on

< previous page

page_148

next page >

< previous page

page_149

next page > Page 149

the basis of picture series. As the basic unit for the analysis of these semi-spontaneous speech samples, the T-unit (Hunt, 1970) was chosen: a main clause and all its subordinate clauses. Regarding the specific requirements of the measures at issue, different parts of these speech samples were used for conducting the various analyses. This issue will be dealt with while discussing the measures. Apart from these speech data, the scores of an oral productive vocabulary test and the outcome of a composition task were examined. The former being a measure for vocabulary size, the latter being used for an analysis on the discourse level. The two measures I would like to discuss on the lexical level are (1) the scores on an oral productive vocabulary test consisting of 62 items (Schaufeli, 1991a) indicating the size of the vocabulary, and (2) a measure for lexical richness conducted on the speech samples. The calculation of a measure for lexical richness must be based on the entire samples. As the separate stories differed considerably in length and each story requires different lexical items, a fixed number of, for example, 100 T-units of the individual samples would not have comprised an equal number of stories and therefore would not have been comparable with respect to this lexical measure. Therefore all the stories were incorporated into the lexical analysis. The individual samples varied widely in length, but the measure used - the Index of Guiraud - corrects for this factor (van Hout & Vermeer, 1988). This Index is calculated as follows (Guiraud, 1959):

The morphosyntactic proficiency was measured by means of the mean length of utterance (MLU). It reflects the average number of morphemes per T-unit:

This is considered to be a measure for general morphosyntactic complexity.

< previous page

page_149

next page >

< previous page

page_150

next page > Page 150

The second measure on this same level is the Agglutination Index (A-index) reflecting the average number of bound morphemes attached to a stem. It is calculated according to the following formula:

This measure is considered to be reflecting the degree to which language use is analytic vs. synthetic. This index shows close similarity to the synthetic index as proposed by Greenberg (1960). For calculating measures on the morphosyntactic level such as these, samples of 100 T-units are considered to be sufficient. The first 100 of the individual samples were used for this purpose. The last two variables on the discourse level may need a more detailed explanation. In both oral and written texts, narrative cohesion was measured. It was defined following Halliday & Hasan (1976), as establishing cohesive relationships (cohesive ties) between elements in a text. These ties are established by formal linguistic means such as pronominal reference, ellipsis and lexical substitution. Only relationships between sentences where taken into account. Per T-unit the number of correctly realized cohesive elements, i.e., elements establishing inter-sentential ties, were counted in both oral and written narratives, in order to calculate the Cohesiveness Index (Scinto, 1983):

For this analysis of oral speech on the discourse level, the first story of each semi-spontaneous sample has been taken into account. Because almost all the stories were about the same two persons (a father and his son), only in the first one the characters were introduced as new by all the children. In the subsequent stories some relied on the preceding stories and others did not. As this may have caused variation with respect to the use of reference, only the first story was taken into consideration for the calculation of the oral Cohesiveness Index. The N-sample consisted of 227 T-units, the T-sample of 167. Additionally, the children wrote a composition about a 'walking tree'. This resulted in a written N-sample of 439 T-units and a T-

< previous page

page_150

next page >

< previous page

page_151

next page > Page 151

sample of 274 T-units, which were used for calculating the second written Cohesiveness Index. Results The scores on all the measures discussed above can be found in Table 1 below. In this chapter only quantitative results are discussed in relation to each other. Details about the (qualitative) interpretation of the separate measures can be found in Schaufeli (1991b). It will be clear that the difference between the groups is always in favour of the monolinguals, i.e., on all measures the monolingual T-group scores higher than the bilingual N-group. As far as the vocabulary test is concerned there is a large difference in score levels between the two groups, but the standard deviations are also very high. Nevertheless, the difference is statistically significant (t-test for independent samples, p<.05). In Table 1, between brackets the scores are given in percentages: whereas the T-children reply correctly to 3/4 of the elicited items, the N-children do not know even half of them. Being the control group, the T-group establishes the standard of comparison. Therefore the differences between the groups are expressed in standard deviations (SD) of the T-group. For the vocabulary test, the difference score appeared to be as much as 3.9 SDs. This seems to be a dramatic difference. With respect to the Index of Guiraud again statistically significant differences between the two groups were found. However, the difference seems to be considerably small (0.9 SD), compared to the difference on the vocabulary test. This implies that the narrative task is relatively easy for the bilinguals, at least as far as vocabulary is concerned. An important general characteristic of a test is that specific items are elicited. However, spontaneous speech provides the possibility of avoiding or hiding lexical gaps by means of substitutions or paraphrases. Moreover, the semi-spontaneous task used for this study did not require the use of specific vocabulary and this may have caused a ceiling effect for the monolingual T-group. These children simply did not get the opportunity to use a whole range of relatively specific lexical items, while telling the stories. The vocabulary test, however, did not imply such a ceiling effect.

< previous page

page_151

next page >

< previous page

page_152

next page > Page 152

On the morphosyntactic level both the MLU and the A-index show statistically significant differences between the groups (t-test independent samples, p<.05). Expressed in standard deviations of the T-group, the differences are almost equal: 1.3 SD and 1.4 SD respectively. This means that the language use of the bilingual children in the Netherlands is to the same degree morphosyntactically less complex and less synthetic, compared to the language use of their monolingual peers in Turkey. Table 1 Scores on the various measures reflecting different aspects of language proficiency N-group T-group Difference (N=18) (N=18) (SDT) Lexicon M 27.4 (44.2) 46.7 (75.0) 3.9 Vocabulary Test: SD 7.0 4.9 M 7.6 0.9 Index of Guiraud: 8.2 SD 0.7 0.7 Morphosyntax M 7.1 8.6 1.3 MLU: SD 0.8 1.2 M 1.4 A-index: .77 .88 SD .10 .08 Discourse M 1.8 2.1 0.54 Oral Cohesiveness: SD 0.46 0.56 M 0 2.4 2.4 Written Cohesiveness: SD 0.5 0.5 It must be noted that the A-index of the N-group may have been increased by sentences such as the following, an example of one of the most basic sentence types in Turkish: (1) Gör-mü-yor-lar. see-NEG-PROGR-3PL They don't see (it). Such sentences obtain a very high A-index, because of the absence of other arguments (e.g. object, adverbials) rather than the high number of suffixes. They occur quite frequently in the N-sample whereas they seem to occur only occasionally in the Tsample. With respect to narrative cohesion a difference was found in the oral narratives, but it was not statistically significant. It also appeared to be smaller (0.54 SD) than the differences concerning the measures on

< previous page

page_152

next page >

< previous page

page_153

next page > Page 153

the other linguistic levels. In the written narratives no difference at all was found. This is all the more surprising considering the very limited writing experience of the bilingual children in the Netherlands. Summarizing the findings presented in Table 1: on the discourse level the differences between the monolingual and the bilingual group seem to be negligible. These variables should probably be regarded as relatively meta-linguistic devices. Once the formal means have been acquired, their appropriate application does not seem to be language specific. Therefore narrative experience in Dutch - the bilinguals' second language - may be useful for the development of narrative skills in Turkish as well. However, both on the lexical and the morphosyntactic level, statistically significant differences were found between the groups. The largest differences seem to exist with respect to lexicon. Although in a spontaneous language task which does not require specific vocabulary (and therefore possibly involves a ceiling effect for the T-children), the differences seem to be of moderate importance. In a task where specific lexical items do have to be used (the vocabulary test), the bilinguals seem to lag behind dramatically. On the morphosyntactic level two measures were conducted, one reflecting syntactic complexity, a second reflecting the degree to which the language use is synthetic. Both measures show a significant difference between the two groups under study. The bilinguals produce less complex and less synthetic sentences than the monolinguals. In Schaufeli (1991b) it is shown that the various measures on one and the same linguistic level do not correlate. Without discussing them in detail, the Spearman rank order correlation coefficients for the measures presented in this chapter are given in Table 2. Table 2 Spearman rank order correlation coefficients for the measures within the various linguistic levels. Lexicon Morphosyntax Discourse N Voc. N MLU AGuiraud index N Oral Written T T T Voc. --MLU --.50* .01 Oral --.29 Guir. .00 - A-in .27 --- Writ -.23 --*p<.05

< previous page

page_153

next page >

< previous page

page_154

next page > Page 154

Within both groups correlations between the various measures appear to be very weak, with the exception of the correlation within the N-group on the lexical level. This general trend suggests that different aspects of the same construct (a certain linguistic level) are measured, which do not always correlate on the variable 'language proficiency'. Therefore on each level only one of the measures discussed in this chapter was chosen in order to compare the outcomes on the various linguistic levels by means of calculating Spearman rank order correlation coefficients. On the lexical level the test scores were preferred because of the assumed ceiling effect in the semi-spontaneous speech task. On the morphosyntactic level the MLU was selected because for the N-group it seemed to predict general level of morphosyntactic proficiency better than the A-index, as was mentioned earlier. Finally, to rule out differences originating from a possible difference in imagination, the Cohesiveness Index on the oral narratives (which are picture-based) was preferred to the one on written narratives. The correlation coefficients are shown in Table 3. In general, the correlation within the N-group appears to be stronger than within the T-group. Table 3 Spearman rank order correlation coefficients between the various linguistic levels N T Vocabulary MLU Oral Cohesiveness Voc. --.45 .59* MLU .26 --.52* O. Coh. .22 -.07 --*p<.05 Within the T-group the correlations between linguistic levels seem to be weak or even completely absent. The correlations within the N-group are moderate. Although the correlation coefficients between groups do not show statistically significant differences (Hays, 1981: 464), this outcome should not be given too much weight. In a population of 2x18 the coefficients must differ around .50 or even .60 for statistically significant differences to be obtained. Therefore I consider it legitimate to

< previous page

page_154

next page >

< previous page

page_155

next page > Page 155

conclude that the correlations within the T-group are generally weaker than within the N-group and considering the fact that the T-group always scores higher than the N-group, it may tentatively be concluded that scores on the variables on the various levels seem to diverge more as language proficiency increases. Conclusions and discussion From the findings presented above it can be concluded that the bilingual children in the Netherlands are generally less proficient in Turkish compared to their monolingual peers in Turkey. Their vocabulary seems to be quite limited, although in spontaneous speech which does not require specific vocabulary items, this is not immediately clear. Morphosyntactically, their language use is less complex and less synthetic. However, compared to the lexical level, the overall differences between the two groups seem to be moderate. With respect to establishing narrative cohesion on the discourse level, almost no differences could be found. This is all the more surprising because Verhoeven (1988) did find a considerable difference between monolingual and bilingual eight-year-olds, on a similar discourse variable. But as was mentioned earlier, narrative experience in Dutch may have a positive effect on narrative skills in Turkish. Finally, I would like to relate the research findings with respect to the correlations between the various linguistic levels as presented in this chapter to earlier research findings concerning younger children in the same situation. Verhoeven and Boeschoten (1986) reported several correlation coefficients for Turkish children in the Netherlands and in Turkey. Contrary to the findings presented here, they found stronger correlations within the monolingual group than within the bilingual group. For example, the correlation coefficient between MLU and the scores on a productive vocabulary test was .39 for their N-group and .63 for their T-group. Generally speaking, the stronger correlations may be due to the fact that Verhoeven and Boeschoten combined several age levels. Furthermore, the measures used in both studies are similar, but not the same - MLU was calculated differently and different vocabulary tests were used. However, as was already mentioned, another explanation for the contradictory results may be offered: correlations between several language skills may vary according to the language proficiency level. The acquisition of vocabulary in particular is a process that never

< previous page

page_155

next page >

< previous page

page_156

next page > Page 156

really ceases, while morphosyntactic complexity has its limitations, i.e., not all children may have acquired an equal number of vocabulary items whereas on the morphosyntactic level they will all have reached more or less their proficiency ceiling. Intrinsic attributes of the variables themselves may also cause the absence of any correlation. The case of the strongly differing correlations between MLU and Oral Cohesiveness as shown in Table 3 may illustrate this. In short sentences only few cohesive items can be realized. As children start to produce longer sentences the cohesiveness between sentences will increase. However, as soon as complex sentence structures are acquired, intra-sentential relationships will offer a second possibility for establishing cohesion. Consequently, inter-sentential cohesiveness may decrease. On the intermediate levels, correlations between several skills may be relatively strong: the basic skills (simple sentences, basic vocabulary) have been acquired, while the extras still have to be learned (complex sentences, specific vocabulary). If this is the case, it can be argued that the eleven-year-old children in the Netherlands, lagging behind in their language acquisition, may have reached such an intermediate level comparable to the level of eight-year-old monolinguals. However, this rather strong claim should be regarded as highly tentative and needs further investigation. References DRIESSEN, G. (1990), De onderwijspositie van allochtone leerlingen. De rol van sociaal-economische en etnisch-culturele factoren, met speciale aandacht voor bet Onderwijs in Eigen Taal en Cultuur. Nijmegen: ITS, Nijmegen University (PhD. Thesis). GREENBERG, J.H. (1960), A quantitative approach to the morphological typology of language. In: Int. Journal of American Linguistics, Vol. 26, 178-194. GUIRAUD, P. (1959), Problèmes et méthodes de la statistique linguistique. Dordrecht: Reidel. HALLIDAY, M. & R. Hasan (1976), Cohesion in English. London: Longman. HAYS, W. (1981), Statistics. New York: Holt-Saunders. HOUT, R. VAN & A. Vermeer (1988), Spontane taaldata en het meten van lexicale rijkdom in tweede taalverwerving. In: TTiA 3, 67-80. HUNT, K. (1970), Syntactic maturity in school children and adults. In: Monograph of Social Research on Child Development 134 Vol. 35, 187-201. JASPAERT, K. & S. Kroon (1990), Eindverslag project: 'Taalverschuiving en taalverlies bij Turken en Italianen in Nederland en Vlaanderen'. 's-Gravenhage: Stichting Taalwetenschap.

< previous page

page_156

next page >

< previous page

page_157

next page > Page 157

SCHAUFELI, A. (1991a), A domain approach to the first language vocabulary of Turkish children in the Netherlands. In: W. Fase, K. Jaspaert & S. Kroon (eds.), Maintenance and loss of minority languages. Amsterdam: Benjamins. SCHAUFELI, A. (1991b), Turkish in an immigrant setting, a comparative study of the language of monolingual and bilingual Turkish children. Amsterdam: Universiteit van Amsterdam (PhD. Thesis). SCINTO, L. (1983), The development of text production. In: J. Fine & R. Freedle (eds.), Developmental issues in discourse. Norwood, N.J.: Ablex, 225-268. VERHOEVEN, L. (1988), Acquisition of discourse cohesion in Turkish. In: S. Koç (ed.), Studies on Turkish linguistics, Proceedings of the fourth International Conference on Turkish Linguistics. Ankara: Middle East Technical University, 437-452. VERHOEVEN, L. (1989), Acquisition of clause linking in Turkish. In: H. Bennis & A. van Kemenade (eds.), Linguistics in the Netherlands. Dordrecht: Foris, 153-162. VERHOEVEN, L. & H. Boeschoten (1986), First language acquisition in a second language submersion environment. In: Applied Linguistics 7, 241-256.

< previous page

page_157

next page >

< previous page

page_ii

next page >

< previous page

page_159

next page > Page 159

7 Summative assessment of ethnic group language proficiency Jeroen Aarssen, Jan Jaap de Ruiter and Ludo Verhoeven The rapid increase in the number and size of ethnic minorities in the Netherlands has had a profound effect on Dutch education. As a result of immigration and a high birth rate, the number of ethnic minority children within the Dutch educational system increased to ten percent of the school population in 1990. In the early seventies the educational policy of the Dutch authorities was ambiguous. On the one hand, it was aimed at the integration of immigrant children into the Dutch educational system. On the other hand, their potential remigration was taken into account. During the seventies it was gradually realized that the majority of the immigrant families had definitely settled in the Netherlands. School teams realized that in order to do justice to the development and maintenance of the immigrant child's social and cultural background, it was necessary to pay attention to his or her own language and culture in a more structural way. The Primary Education Act of 1980 dealt with the position of the languages of ethnic minorities originating from Mediterranean countries. According to this act, mother tongue instruction in, amongst others, Turkish and Arabic, may be given if the children's parents apply for it and if the authorities deem it necessary. The frequency of these lessons is limited to a maximum of 2.5 hours a week within the school curriculum. It is clear that this legislation offers no genuine opportunities for ethnic group language maintenance. With respect to the development of ethnic group languages in a second language submersion context it becomes highly relevant to have insight into the proficiency levels of minority children at different age levels. Earlier studies have shown that the proficiency levels of children at kindergarten level tend to be lower than that of their monolingual peers. Verhoeven & Boeschoten (1986) found that the first language progress of Turkish children in the Netherlands in the age range of 4-8

< previous page

page_159

next page >

< previous page

page_160

next page > Page 160

years was slower than that of their peers living in Turkey. Teunissen (1986) administered bilingual measures for listening comprehension and vocabulary for Turkish and Moroccan children at school age. The results showed a trend of both listening comprehension and vocabulary being slightly better in L2 Dutch. Van de Wetering (1990) examined the efficiency of mother tongue instruction for Moroccan children in the age range of 6-9. She found that the children's proficiency in Arabic was positively related to the numbers of years of first language instruction enjoyed either in Morocco or in the Netherlands. An important question is what levels of first language proficiency ethnic minority children attain at the end of primary school. In his recent dissertation study Driessen (1990) investigated the ethnic group language proficiency of Spanish, Turkish and Moroccan children by the end of primary school. Both a general (written) proficiency test and a self-evaluation test were administered. In the self-evaluation test the children had to rate their oral and written language proficiency along 6 'can-doscales' ranging from 1 ('I am unable') to 5 ('I find it very easy'). The mean scores for oral language proficiency was 4.0 for the Spanish group, 4.2 for the Turkish group and 3.5 for the Moroccan group. For written language proficiency the mean scores for the Spanish group and the Turkish group turned out to be reasonable (72.6 and 57.9 percent correct, respectively). For the Moroccan group the children's scores on the general proficiency test were extremely low: less than 25% of the items were scored correctly. No less than 42% of the Moroccan children failed to do any item correctly, owing to the fact that they were unable to read Arabic. However, the content validity of the tests used can be seriously questioned. In the present study an attempt will be made to assess the ethnic group language proficiency of Moroccan and Turkish children at the end of primary school in a diagnostic way. The notion of language proficiency will not be conceived of as a monolithic ability. The assumption is that using a language involves distinct subskills that can be acquired in different patternings. Besides oral measures for vocabulary and listening comprehension, written measures on the levels of graphemes, lexicon, syntax and text will be distinguished. In order to test the validity of the measures taken, the interrelationship between language skills and background variables will be computed.

< previous page

page_160

next page >

< previous page

page_161

next page > Page 161

Design of the study Informants 81 Moroccan children participated in the research, 40 of whom were boys and 41 girls. 24 of the subjects were so-called 'higher grade entrants' (HGE), i.e., children with educational experience in Morocco who entered Dutch elementary school later; the others may be called first grade entrants (FGE). The children attended schools in the fairly big cities in the central and southern part of the Netherlands. As a rule, two Moroccan assistants (and in some cases Moroccan teachers) administered the tasks. Most schools were visited twice. Also 69 Turkish children participated in the research. There were 35 girls and 34 boys. Five were higher grade entrants. In most cases two Turkish assistants administered the tasks. Instruments In developing similar instruments for both groups we were confronted with the fact that the Moroccan group is characterized by the use of a standard language, i.e., Standard Arabic together with the Moroccan Arabic dialect, serving the role of lingua franca, and various Berber languages (see Abbassi, 1977; de Ruiter, 1989). In view of the fact that both formal, primarily written Standard Arabic and informal, basically spoken, dialectal Moroccan Arabic are important to the young Moroccans from different points of view, we decided to test the children's skills in both. Therefore the tasks were divided into basically oral and written ones in which the former focusses on skills in Moroccan Arabic dialect and the latter on those in Standard Arabic. In view of the similarity desired in the tasks, we decided to develop similar oral instruments for the Turkish group as well, although written standard and (non-standard) Turkish spoken at home do not diverge as much as Standard Arabic and Moroccan Arabic do. Two tasks were developed for oral proficiency: the Oral Vocabulary (OV) task and the Instruction (INS) task. Besides, five tasks were developed for written proficiency: the Word Decoding (WD) task, the Spelling (SPEL) task, the Written Vocabulary (WV) task, the Syntax (SYN) task and the Reading Comprehension (RC) task. The following sections describe the development of the various tasks for the two groups. In Table 1, the separate tasks are summarized.

< previous page

page_161

next page >

< previous page

page_162

next page > Page 162

Table 1 Overview of separate tasks Oral Vocabulary (OV) Oral Language Instruction Comprehension Proficiency (INS)

Written Language Proficiency

word level text level grapheme Word Decoding (WD) level Spelling (SPEL) word level Written Vocabulary (WV) word level Syntax (SYN) sentence level Reading Comprehension (RC) text level

Arabic Language Proficiency Tasks Oral Vocabulary. This task is based on a list of 7.000 predominantly concrete words, which six-year-olds are supposed to know (Verhoeven & Vermeer, 1989). We took this list because we expected that the Moroccan children are somewhat retarded in their lexical development (cf. de Ruiter, 1989). Moreover, Moroccan Arabic is an informal language in which predominantly concrete words are used. For this task, which is identical in structure for both groups, we made use of pictures and items from the receptive vocabulary task that forms part of the (Dutch) Language Test for Ethnic Minority Children (Taaltoets Allochtone Kinderen (TAK) by Verhoeven & Vermeer, 1987). Target words were translated into Moroccan Arabic and Turkish. Note that the procedure differs from the original Dutch task, in which children are tested individually. In our version, the task assistant reads the target words aloud. The children are asked to circle one picture out of four, i.e., the one that matches the word they have just heard. The task consists of 60 items. Instruction Comprehension. The basic idea behind this task is that giving and carrying out instructions constitutes a typical classroom behaviour. The instruction task consists of five drawings. We invented an imaginary country 'Botanya'. Five pictures show Botanya's flag, its national dress, its paper money, its capital's plan and a square in a major city. Every picture was accompanied by six instructions, read aloud by the researcher. With the aid of coloured pencils children are asked to carry out the instructions, such as 'Colour the left triangle on the middle bar of the flag red'. Focus is given to directions, left/right, colours and geometrical figures. The children were also asked to follow

< previous page

page_162

next page >

< previous page

page_163

next page > Page 163

the correct routes to certain objects (post office, restaurant) on the city plan, described by the researcher ('take the third street on the right'). Every correctly executed instruction is rewarded one point. The task contains 30 items. Word Decoding. In the Word Decoding Task the children were given items with four words; the task assistant read out one word and the children were instructed to mark the word they thought they had heard. The items were ordered according to their morphological complexity. The choice of words used in this task had to meet two criteria: they should not be unknown to the children and they had to reflect the basic word models of Standard Arabic. In order to reach a random series of words, 385 words were picked from the vocabulary of Kohnstamm et al. (1981), which consists of words six-year-old native Dutch children should understand. According to the two criteria just mentioned, a further selection was made of 46 words that were used for the task. Distractors were developed, primarily based on minimal differences in consonants and vowels. An example of a sequence is maalun maylun naarun maarrun. Spelling. The Spelling Task consisted of 42 items, in which three sentences (A, B, C) with underlined words were presented. One of these underlined words may contain a spelling error. The children were asked to mark the sentence with the possible error or to mark a fourth possibility (D), viz. 'no error'. The selection of the underlined words was made on the basis of a survey on errors Moroccan children and students in Arabic made in spelling Arabic. On the basis of this survey the 42 items were created The most important categories of errors were: • interchanging consonants (casalun•,asalun) • interchanging long and short vowels ('awlaadun•,awladun) • deletion of the '1' of the article 'al' ('al-saacatu,a-saacatu) Written Vocabulary. The Written Vocabulary task is based on a list of 25,000 words children are supposed to have at their command at the end of elementary school. From Van Dale's Basic Dictionary of Dutch (Huijgens & Verburg, 1987) a random list of 108 words was selected and translated into Arabic. After omission of words for which no adequate translation could be given, a list of 54 words was left. These words were presented in items to the children in a Multiple Choice format, in which they had to mark the synonym of a target word. Three

< previous page

page_163

next page >

< previous page

page_164

next page > Page 164

distractors were created on the basis of phonological or semantic similarity with the target word. Syntax. The Syntax Task has a format similar to the spelling task except for the fact that complete sentences had to be judged. Children were asked to mark an incorrect sentence from four alternatives ('no error' as a final option included). The errors were based on a survey on syntactic errors, made by Moroccan children in the Netherlands. The task consists of 42 items. Major error categories were: • agreement in gender (haadhihi d-dafaatiru *haadhaa d-dafaatiru) • agreement between subject and verb (dakhala t-talaamiidhu *dakhaluu t-talaamiidhu) • construct (baabu d-daari *baabu d-daaru) • conjunction ('an 'adhhaba*'an 'adhhabu) Reading Comprehension. The Reading Comprehension task consists of three short stories each of which is followed by a number of multiple choice questions. In view of the fact that Moroccan children in the last grade of elementary school do not have much experience in reading stories in Standard Arabic, the stories chosen were relatively easy (grade 6 equivalent). Questions were asked about explicit and implicit information in the texts. The task consists of 18 items. Turkish Language Proficiency Tasks Oral Vocabulary. The oral vocabulary task is based on the same list of 7000 words that was used for Arabic. As, compared with Standard Arabic and Moroccan Arabic, written standard and (non-standard) spoken Turkish are relatively similar, no particular dialect words were used. The Turkish oral vocabulary task consists of 60 items. Instruction Comprehension. The format of the instruction task has been described above. The Turkish version is equivalent to the Arabic one. Word Decoding. This task consists of 46 items. Each item consists of 4 words that have almost identical orthographical patterns, i.e., they are more or less, but not exactly, similar to one another. An example is the sequence kus * kos* kas*. One word is read aloud by the researcher, after which the children have to circle the word they hear. The list of

< previous page

page_164

next page >

< previous page

page_165

next page > Page 165

words has an increasing level of difficulty (from monosyllabic to polysyllabic words; from frequent to less frequent words). Spelling. After an analysis of a corpus of Turkish children's essays and Turkish adults' letters we categorized spelling errors as follows: • reduction of double consonants (mahalle *mahale) • doubling of consonant (ezan *ezzan) • deletion of vowel or consonant (yagmur *yamur) • phoneme change (bisiklet *pisiklet) • addition of vowel or consonant (saat *sahat) On the basis of these categories we selected misspelled words and put these into sentences in which the misspelled word was underlined. Every item contains two other sentences in which one word has been underlined. These underlined words serve as distractors. A fourth possibility (D, no error) completed each item. The test consists of 42 items. Written Vocabulary. As was the case with the Arabic version, this task was also based on a sample of 25,000 words. As described above, a list of 54 words was selected and these words were translated into Turkish. We tried to avoid paraphrases, using synonyms instead. The target words were put into sentences. Four options (A, B, C, D) were given; one synonym and three distractors. Children were asked to choose the synonym of the target word. Syntax After an analysis of syntactic errors in essays by Turkish children living in the Netherlands and letters by Turkish adults, living in Turkey or the Netherlands, we categorized these syntactic errors as follows: • causative suffix (kaçan * ) • possessive constructions (videomuz var  *video bizim var) • incorrect or incomplete marking (yatak odam *yatak odasim) • clause linking (çok erken kalkip gittik *çok erken kalkarak gittik) • conditional suffix (-irse *-se) • morpheme order (zengin degildi * *zengindi degil*) On the basis of these categories we selected 42 incorrect constructions and put them into sentences. Two distractors were also chosen along with a fourth possibility (D, no error). The children were asked to find

< previous page

page_165

next page >

< previous page

page_166

next page > Page 166

the wrong sentence or to choose answer (D) if they could not find any error at all. Reading Comprehension. Four Turkish stories of different text length were chosen. We edited these texts and constructed 20 multiple choice items in which questions about the content were asked. There were four basic categories of questions: explicit and implicit information in the text, main line of thought and anaphora. Procedure In order to get a general impression of the children's abilities in the ethnic group language, means and standard deviations on each measure were computed. In addition, several statistical analyses were conducted to evaluate the reliability and validity of the measures taken. First of all, the internal consistency of the tasks was examined by calculating Cronbach's coefficient alpha. Second, in order to check the content validity of the language proficiency tasks, a fit measure of the one parameter item response model of Verhelst & Eggen (1989) was computed. This model offers a psychometrical account of the testing behaviour of the child, and claims to yield sample free item analyses. A fit of the model would also lend empirical support to the assumption that the items on the task can be scaled along one and the same dimension. Third, the interrelationship between language skills was examined. Correlations between the scores on the measures for each language were computed. Moreover, factor analyses were conducted clustering the scores on the tasks that had been assessed (own values greater than 1.0; varimax rotation of factor patterns). Finally, the relationship between language skills and background variables was explored by computing correlations between language proficiency scores and the period of L1 instruction, the use of L1 at home and in the peer group. At the same time, a t-test was conducted to test the differences in first language proficiency between children who entered primary school in the Netherlands in the first grade and those who entered Dutch primary school in later grades.

< previous page

page_166

next page >

< previous page

page_167

next page > Page 167

Results for Arabic Oral language proficiency Scores. Table 2 presents the basic statistics for oral proficiency tasks in Arabic. Table 2 Basic statistics for Oral Vocabulary (OV) and Instruction Comprehension (INS) in Arabic standard n of items mean % correct deviation OV 60 44.68 74 9.94 INS 30 17.69 59 7.96 The scores on the oral vocabulary task are good. The scores on the instruction task are somewhat lower. One may draw the conclusion that the children have reasonable skills in Moroccan Arabic. Reliability. We calculated Cronbach's alpha and item-test correlations to test the internal consistency of the oral language tasks. Table 3 gives an overview of the alphas for the oral tasks. Table 3 Cronbach's alpha for the oral tasks in Arabic alpha .92 OV .92 INS The internal consistency of the two tasks is good. However, some items of the Oral Vocabulary Task were removed either because they had a 100 % correct score or because their p-value was extremely high. After the deletion of 22 items alpha remained .92. There was no need to delete items in the instruction task. Content validity. In order to check the content validity of the tasks, we analysed them with the 'one parameter item response model' (Verhelst & Eggen, 1989). Within this model the chance of a correct response is

< previous page

page_167

next page >

< previous page

page_168

next page > Page 168

determined by the child's proficiency and by the relative difficulty of the item. A chi square test has been used to test the goodness of fit of the model. If the value of chi square is not significant, there is no need to reject the model. Table 4 presents the chi square statistics for the oral tasks. It can be seen that there is no need to reject the model for either task. Table 4 Chi square statistics of the oral language tasks in Arabic chi square df P OV 17.6 35 .97 INS 43.2 29 .06 Written language proficiency Scores. Table 5 presents the scores on the written tasks. Table 5 Scores of the written language tasks in Arabic n of itemsmean % correct standard deviation WD 46 37.25 81 8.93 SPEL 42 10.98 26 4.27 WV 54 17.05 32 7.31 SYN 42 10.12 24 3.46 RC 18 10.81 60 4.67 It can be seen that the scores on Word Decoding and Reading Comprehension are high, while the scores on the other tasks are extremely low. The children are quite proficient in word decoding and do not face major problems in understanding short stories. In fact, this is the first time that research on such a large scale shows that Moroccan children have some basic skills in Standard Arabic (cf. van de Wetering, 1990; Driessen, 1990). On the other hand, it is clear that they have not as yet developed a basic command of spelling and syntax. Furthermore their vocabulary scores are slightly above chance level.

< previous page

page_168

next page >

< previous page

page_169

next page > Page 169

Reliability. The values of Cronbach's alpha were .93 (WD), .58 (SPEL), .81 (WV), .38 (SYN) and .85 (RC). 11 items of the Word Decoding Task were deleted because of ceiling effects. The adapted form of the task yielded a high reliability. In both spelling and syntax tasks items (18 and 20 in numbers) were deleted in order to attain a higher alpha. The written vocabulary task, after deleting 16 items, reached a good reliability and the same can be said of the reading comprehension task, in which no items had to be deleted. Table 6 gives an overview of the alphas for the written tasks. Table 6 Cronbach's alphas for the written tasks in Arabic alpha WD .94 SPEL .76 WV .90 SYN .76 RC .85 Content validity. Table 7 contains the values of chi square for the written tasks. Table 7 Chi square statistics for written language tasks in Arabic df P chi square WD 47.16 34 .07 SPEL 13.69 23 .96 WV 26.36 37 .89 SYN 10.79 21 .94 RC 6.31 17 .98 The figures show that there is no need to reject the model for the written tasks. Interrelationship between language skills Table 8 shows the correlations between the adapted versions of the various tasks.

< previous page

page_169

next page >

< previous page

page_170

next page > Page 170

Table 8 Correlations between separate tasks; 0.01; *** = p < 0.001 INS WD SPEL WV VO INS WD .58*** SPEL .29* -.25* WV .25* .29** .23* SYN .26* .22* .39** .38**

* = p < 0.05; ** p < SYN -.17* .29 -.04 .20 .29

RC .08 .06* .61*** .24** .55 .57***

The interrelationship between the tasks appears quite evident. A further factor analysis (see Table 9) shows that two factors explain 71 % of the variance (factor 1: 42.9 % and factor 2: 28.1 %; the spelling and syntax tasks were not taken into account in this analysis due to the fact that their scores were on chance level). The first factor may be labeled an oral factor (OV and INS have high loadings on this factor), the second seems connected to reading skills (high loadings for WD and RC). Table 9 Factor loadings of test scores for Arabic after varimax rotation Factor 1 Factor 2 OV .88 -.00 INS .87 .07 WD -.06 .82 WV .48 .50 RC .13 .91 Relation to background variables Table 10 presents the correlations for the Arabic proficiency tasks with the variables 'number of years of education in native language (ENL)', 'home language' and 'peer language'. The spelling task and the syntax task were excluded from this analysis because of the low mean scores. It can be seen that there is a significant relationship between the amount of instruction in Arabic and the scores on WD and WV. Moreover, the measures of OV and WD appear significantly related to the use of Ara-

< previous page

page_170

next page >

< previous page

page_171

next page > Page 171

bic at home. The measures of INS, WV and RC tend to be significantly related to the use of Arabic in peer contacts. Table 10 Correlations between test scores and years of Education in Native Language (ENL), home language and peer language; * p < 0.05 **= p < 0.01 OV INS WD WV RC ENL .17 -.13 .22* .30* .09 home language .38** .03 .29* .10 .04 peer language -.16 .21* .07 .22* .32* Table 11 Differences between first grade entrants (FGE) and higher grade entrants (HGE) in test scores; *= p< 0.05 **= p< 0.01 task group n mean sd t df 48 24.79 8.99 -1.18 66 OV FGE 20 27.25 5.51 HGE 43 16.95 8.40 59 INS -1.12 FGE 18 19.44 6.69 HGE 52 29.56 7.13 WD 0.99 69 FGE 19 27.63 7.88 HGE 41 6.88 4.15 SPEL -0.61 54 FGE 15 7.67 4.70 HGE 42 4.53 3.03 SYN -1.51 56 FGE 15 6.20 5.13 HGE 44 13.66 8.89 WV 0.33 60 FGE 18 12.89 6.37 HGE 24 10.58 3.94 RC -0.40 36 FGE 14 11.21 5.86 HGE

< previous page

page_171

next page >

< previous page

page_172

next page > Page 172

Finally a t-test was executed in order to find out whether there are differences in scores between first grade entrants (FGE) and higher grade entrants (HGE). Table 11 shows the results. The HGE's have somewhat higher scores than the FGE. In no case, however, are the differences significant. Evidently, the educational experience of these pupils in Morocco did not influence their performance favourably. Results for Turkish Oral Language Proficiency Scores. In Table 12 the basic statistics for the two oral tasks in Turkish are presented. Table 12 Basic statistics for the oral proficiency tasks in Turkish standard n of items mean correct deviation OV 60 48.44 81 4.54 INS 30 26.12 87 2.56 The high scores justify the conclusion that in general the Turkish children have no difficulty at all with oral language skills. Reliability. The values of alpha were .71 (OV) and .57 (WD) respectively. After removing items with extremely high or low means, or with low item-test correlation both alphas reached more acceptable values: Table 13 Cronbach's alphas for the oral tasks in Turkish alpha OV .75 INS .67

< previous page

page_172

next page >

< previous page

page_173

next page > Page 173

Content Validity. In order to check the content validity of the oral proficiency tasks in Turkish a fit measure of the one parameter item response model was computed. Table 14 gives an overview of the resulting chi square values. Table 14 Chi square statistics of the oral tasks in Turkish chi square df p OV 23.3 39 .97 INS 17.3 18 .55 As the chi square values of both tasks are not significant, no grounds exist for rejecting the model. Written Language Proficiency Scores. The scores on the five written language tasks are as follows: Table 15 Basic statistics for the written proficiency tasks in Turkish standard n of items mean % correct deviation WD 46 42.97 93 5.23 SPEL 42 22.11 52 8.06 WV 54 30.98 57 6.28 SYN 42 24.71 59 5.99 RC 20 12.82 64 3.70 In general, the children seem to have no difficulty with the written language tasks, although there is much more variation than on the oral proficiency tasks. Scores on WD are extremely high. All children performed well on this task. Reliability. The values of alpha were .92 (WD), .90 (SPEL), .78 (WV), .80 (SYN) and .74 (RC). Table 16 gives the alphas for the written proficiency tasks after items with extremely high or low means, or with low item-test correlation have been removed.

< previous page

page_173

next page >

< previous page

page_174

next page > Page 174

Table 16 Cronbach's alphas for the written tasks in Turkish alpha WD .93 SPEL .91 WV .82 SYN .86 RC .74 Content Validity. Again, chi square values have been calculated in order to test the fit of the one parameter item response model. The values are given in Table 17. Table 17 Chi square statistics for the written language tasks in Turkish chi square df p WD 18.75 34 .90 SPEL 19.45 34 .95 WV 21.85 37 .95 SYN 19.95 29 .95 RC 15.65 19 .65 Given the fact that all chi square values are not significant, there is no ground to reject the model for the tasks under consideration. Interrelationship between language skills Correlations between the separate test scores were calculated, using the reduced version of the tasks. The values of the correlation coefficients are presented in Table 18. It can be seen that in general there are significant positive correlations between the separate Turkish proficiency tasks.

< previous page

page_174

next page >

< previous page

page_175

next page > Page 175

Table 18 Correlations between separate tasks; *= p < 0.05 **= p < 0.01 ***= p < 0.001 INS WD SPEL WV SYN RC OV .42*** INS .40*** .26* WD .49*** .15 .38*** SPEL .21* .30*** .31** .49*** WV -.39*** .15 .45** .57*** .63*** SYN .52*** .32 .40*** .61*** .60*** .57*** In order to reduce the original seven variables, a factor analysis was carried out. It appears that two factors account for 66.9% of the variance. The rotated factor matrix is shown in Table 19. Table 19 Factor loadings of test scores for Turkish after varimax rotation Factor I Factor 2 OV .31 .78 INS .01 .88 WD .44 .51 SPEL .83 .06 WV .73 .34 SYN .82 .12 RC .74 .29 The rotated factors are amenable to interpretation. Inspection of the factor matrix suggests that the first factor is an oral factor (INS and OV have high loadings on this factor). The second factor may be labeled as a written factor (high loadings for SPEL, SYN, WV and RC). WD has a loading > .40 on both factors. This is the only written task that also has an oral component in the procedure, in that the researcher reads the target words aloud. The results of the factor analysis show a strong correspondence with the results of the similar analysis of the Arabic tasks (see above). Relation to background variables No significant correlations were found between the scores and age, and between scores and home language. This seems to be due to the fact

< previous page

page_175

next page >

< previous page

page_176

next page > Page 176

that there was almost no variation in the variables age and home language. Correlations between the test scores and the background variables 'number of years of Education in Native Language' (ENL), and peer language are presented in Table 20. Table 20 Correlations between test scores and years of Education in Native Language (ENL) and peer language; (0= Dutch, 1=both Turkish and Dutch, 3=Turkish) *= p < 0.05 **= p < 0.01 OV INS WD SPEL WV SYN RC ENL .27* .31** .23* .12 .18 -.01 .06 peer language .27* -.16 .02 .02 .17 .03 .28** Table 21 Differences between first grade entrants (FGE) and higher grade entrants (HGE) in test scores; *= p< 0.05 **= p< 0.01 task group n mean sd t df 61 26.92 4.48 -1.72 64 OV FGE 5 30.40 1.34 HGE INS 61 17.61 2.38 -0.37 64 FGE 5 18.00 0.71 HGE WD 63 32.63 4.64 -0.90 65 FGE 4 34.75 0.50 HGE SPEL 63 18.37 7.85 -2.86** 65 FGE 4 29.75 4.57 HGE SYN 62 20.10 5.80 -1.24 64 FGE 4 23.75 2.63 HGE WV 63 24.37 5.91 -1.80 65 FGE 4 29.75 2.75 HGE RC 61 12.82 3.70 -2.12* 64 FGE 5 16.40 2.41 HGE

< previous page

page_176

next page >

< previous page

page_177

next page > Page 177

There is a significant correlation between ENL and the scores on OV, INS and WD. Peer language and the scores on OV and RC correlate significantly. The significance of differences between first grade entrants (FGE) and higher grade entrants (HGE) were calculated by means of a t-test. We must bear in mind, however, that the last group is very small. The results of the t-test are given in Table 21. As this Table shows, higher grade entrants in general have higher scores. For SPEL and RC the difference is significant. Conclusions and discussion From the present study several conclusions can be drawn. First of all, the reliability and validity of the tasks under consideration seem quite reasonable. In all cases the measures turned out to be internally consistent. Moreover, for none of the measures did the one parameter item respons model of Verhelst & Eggen (1989) have to be rejected. Thus, it can be concluded that the chosen tests are homogeneous. Furthermore, it can be concluded that the first language proficiency of the Turkish children turns out to be reasonably good. An oral and a written language factor appear to underlie the children's results on the measures taken. Except for some individual cases, the Turkish children have no problems with the comprehension of basic instructions. They also have a reasonable oral vocabulary. With respect to the written language tasks, there is more individual variation in scores. The two oral measures and the measure on word decoding turn out to be significantly related to the period of L1 instruction the children have had. The scores on the measures for oral vocabulary and reading comprehension are also significantly related to the use of L1 in the peer group. Turkish children who had entered Dutch primary school in later grades obtained generally higher scores than first grade entrants. However, the differences in scores were not significant. Compared with the Turkish group, the Moroccan children obtained lower scores on all measures under consideration. Factor analysis gives support to a conception of Arabic proficiency in which two dimensions can be distinguished: an oral factor and a factor connected to reading skills. The scores on the oral tasks turn out to be reasonably good. With regard to written language proficiency, the children obtained moderate

< previous page

page_177

next page >

< previous page

page_178

next page > Page 178

scores on the measures for word decoding and reading comprehension. The scores on written vocabulary turn out to be low, those on spelling and syntax extremely low. The period of L1 instruction appeared positively related to the children's scores on word decoding and written vocabulary. The language used at home showed a strong relationship to the children's oral vocabulary and their word decoding ability, the language in the peer group with the children's written vocabulary and their reading comprehension ability. It can be concluded that the children have great difficulty in reflecting on written Standard Arabic. However, it is interesting to note that there are no substantial differences in first language proficiency between children entering Dutch primary school in the first grade and those entering Dutch primary school in later grades. Apparently, the influence of former instruction in Arabic in Morocco is not strong. The present study makes it clear that the concept of ethnic language proficiency as a pluralithic ability does make sense. Relative patterns of first language development in different ethnic groups can be demonstrated. The present study also provides some evidence that the first language ability of ethnic minority children can be predicted from background factors that are related to the language socialization at home and the language instruction at school. In a follow-up study the prediction of language abilities from sociolinguistic and educational factors will be examined in more detail. In that study it will also be determined to what extent the ethnic group language proficiency of children by the end of primary school is a predictor of secondary school success. References ABBASSI, A. (1977), A sociolinguistic analysis of multilingualism in Morocco. Ann Arbor. DRIESSEN, G. (1990), De onderwijspositie van allochtone leerlingen. De rol van sociaal-economische en einisch-culturele factoren, met speciale aandacht voor het onderwijs in eigen taal en cultuur. Nijmegen: Nijmegen University, Instituut voor toegepaste sociale wetenschappen (PhD. Thesis). HUIJGENS, M. & M. Verburg (1987), Van Dale Basiswoordenboek van de Nederlandse Taal. Utrecht: Van Dale Lexicografie. KOHNSTAMM, G., A. Schaerlaekens, A. De Vries, G. Akkerhuis & M. Froonincksx (1981), Nieuwe streeflijst woordenschat voor 6-jarigen. Lisse: Swets & Zeitlinger. RUITER, J.J. DE (1989), Young Moroccans in the Netherlands: An integral approach to their language situation and acquisition of Dutch. Utrecht: Utrecht University (PhD. Thesis).

< previous page

page_178

next page >

< previous page

page_179

next page > Page 179

TEUNISSEN, F. (1986), Een school, twee talen. Utrecht: Utrecht University (PhD. Thesis). VERHELST, N. & T. Eggen (1989), Psychometrische en statistische aspecten van peilingsonderzoek, PPON-rapport 4. Amhem: Cito. VERHOEVEN, L. & H. Boeschoten (1986), First language acquisition in a second language environment. In: Applied Psycholinguistics 7, 241-255. VERHOEVEN, L. & A. Vermeer (1987), Taaltoets allochtone kinderen (TAK). Tilburg: Zwijsen. VERHOEVEN, L. & A. Vermeer (1989), Diagnose van kindertaal. Tilburg: Zwijsen. WETERING, S. VAN DE (1990), Onderwijs in eigen taal en cultuur aan Marokkaanse kinderen in Nederland. Het OETC als resultante van een maatschappelijk krachtenspel. Utrecht: Utrecht University (PhD. Thesis).

< previous page

page_179

next page >

< previous page

page_ii

next page >

< previous page

page_181

next page > Page 181

8 Parental attitudes towards child bilingualism in the Nordic countries Sirkku Latomaa The Nordic countries are becoming to a greater extent a multilingual and multicultural part of the world. In what ways will the post-World War II immigration to and within the Nordic countries influence the language situation in the long run? Will multilingualism be a short intermediate stage, where the new minorities are quickly assimilated, both linguistically and culturally? Or will several minority languages and cultures endure over a longer period of time? Which linguistic groups in each Nordic country have the best possibilities for maintaining their languages? Which of these groups have chosen to and have been able to utilize these possibilities? Questions within this field of problems are being studied by a group of researchers in four Nordic countries through an ongoing project 'The Development of Immigrant Languages in the Nordic Counries' 1. In this project, we are mainly interested in different minorities' social organisations. In the first part of our project, we try to find out about the number and distribution of a large number of minorities in the Nordic countries. In the second part, we investigate how the geographic and demographic conditions are utilized by certain selected groups in different localities. We study the role that the socio-cultural background has in the development of the minorities' social relationships within the group and within the family. When doing this we are relying partly on Fishman's (1972) theory on domain distribution's relationship to language shift and partly on Milroy's (1980) application of network theory to theories of linguistic change. We try to find out which spheres of activity are dominated by each language and which minority groups form a base for dense multiplex social networks. In order to form this kind of network there have to be enough immigrants living and working in close contact with one another. Our first hypothesis concerns the relationship of different kind of networks and language maintenance: close, multiplex networks within

< previous page

page_181

next page >

< previous page

page_182

next page > Page 182

the group promote language survival while sparse, uniplex networks and networks which cross over language boundaries promote language shift. Another condition for immigrant languages' survival is that the languages are passed on from the first to the second generation. In this context, it is interesting to study the attitudes that the members of different minorities have concerning their language and its role in the children's and the group's future in the Nordic countries. This chapter will primarily address this last question. Informants and data The minority groups and the localities selected are presented in Table 1. Table 1 Overview of minority groups and localities. Køge, Denmark Gothenburg, Turkish-speaking Turks Sweden Eastern Finnmark, Finnmark, Norway Gothenburg, Finnish-speaking Finns Sweden Vietnamese-speaking Helsinki, Finland Vietnamese Bergen, Norway Gothenburg, Sweden Copenhagen, English-speaking North Denmark Americans Helsinki, Finland The groups are very different from one another. They vary, e.g., in size; Finns in Sweden and Turks in Denmark are the largest immigrant groups in these countries, whereas North Americans form a fairly small minority in all three countries. They result from different types of immigration: while Turks and Finns have emigrated for reasons of employment, Vietnamese have the status of refugees. As to North Americans, most of them have settled in the Nordic region to marry or co-habit with a spouse from the majority population. This group consists mainly of mixed-language families, whereas some of the other groups show only minor tendencies in this direction. There are also differences in the majority population's attitudes towards the different groups and their respective languages. Compared to the other minority

< previous page

page_182

next page >

< previous page

page_183

next page > Page 183

languages in our study, English has a very high status and is widely used in the official life in the Nordic countries. In our survey, Vietnamese families in Norway have the largest number of children, while the smallest families can be found among the Finns in Sweden. Except for among North Americans, working class occupations dominate. The groups vary as to the average length of stay: from about 17 years (Finns in Sweden) to 5 years on the average (Vietnamese in Finland). Except for North Americans in all three countries and Vietnamese in Finland, the other groups show notable tendencies to concentrate in certain areas in the municipality investigated. In our project, data is collected from a sample of 30 informants from each informant group via a personal interview of about 3-4 hours, in most cases conducted by a fellow countryman at the informant's home. Each informant is a parent of one or more school-age children. At the moment, we have finished interviewing four of our groups, namely Turks in Denmark, North Americans in Sweden, and Vietnamese in both Norway and Finland. In addition to these groups, I will use the data from the Finnish group in Gothenburg where 2/3 of the interviews have been completed. Altogether, data for this preliminary report consists of interviews with 140 minority parents. Parental attitudes towards children's bilingualism Attitude is often taken to comprise three components: cognitive, affective and behavioral. One knows or believes something, has some emotional reaction to it, and therefore may be assumed to act on this basis (Edwards, 1985: 139). Instead of using indirect assessment methods, attitudes are approached in a naturalistic way in this study. Informants were asked directly about their attitudes towards various aspects of their children's bilingualism. In addition to these direct open questions, I have used selfreports of language usage as an indication of parental attitudes. Below I shall try to answer questions like: How important is it according to the parents that their children learn L2 2? How important is it that they maintain L1? What role do the parents relinquish to the school? What are the educational needs of minority children according to the parents? What role do the families take themselves; what methods do they use when passing on the minority language to their children?

< previous page

page_183

next page >

< previous page

page_184

next page > Page 184

Importance of L1 and L2 The general attitude towards bilingualism may be said to be positive in all groups. Vietnamese parents in Finland point out that 'when it comes to young people, bilingualism can only be an advantage'. Many of the Turkish parents in Denmark stress that they want their children to learn both languages equally well. North Americans see their children's bilingualism as an individual advantage, something their particular child can use in the present and future for various purposes. This is true of the other groups as well, but, in addition, there are voices among the Vietnamese groups that stress the importance of the children's bilingualism for the family and maybe even for the minority community. On the whole, parents emphasize the fact that children have a great capacity for learning and that bilingualism is something positive. Why do the children have to learn L2? It can be expected that the majority language is seen in a utilitarian manner (cf. Edwards, 1983: 226). It is useful; the children need it to get along in society, to get a good education and/or a good job in the future. It is also necessary; 'it is the only language that can be used when you go outside of your home' as a Turk in Denmark said. 'We live here and it is the language of the country' say many of the informants. Turks and Vietnamese also mention that children should learn the majority language to adapt themselves in the new society and to avoid isolation. On the other hand, children should maintain the minority language, too, because 'it is the mother tongue' or 'our language' and 'we are Vietnamese/Turks even if we live abroad'. In other words, children should learn something about their origin. There are even some practical reasons: particularly Finns and Turks mention going back to their home countries for vacations or even permanently. If the family returns to its home country, the children should be able to go to a Finnish or a Turkish school. In addition, even some of the Vietnamese mention their home country in this connection. They have visited Vietnam on a vacation, and have noticed how important it is for the children to be able to communicate with the relatives and fellow countrymen back home. For the North Americans, whose L1 is an international language, bilingualism is of highly practical value, too. The Vietnamese have some additional reasons not mentioned by the other groups. The language seems to be important for the Vietnamese family for a wide range of reasons. According to some parents, it is a carrier of cultural identity: 'When language is maintained, traditional habits and culture are maintained'. It should also be maintained in order

< previous page

page_184

next page >

< previous page

page_185

next page > Page 185

to avoid conflicts within the family, to maintain the emotional ties between the parents and the children. Parents stress that they want to bring up their children in Vietnamese. It is extremely important that the children learn to behave properly within the Vietnamese community, i.e., to be polite, to show respect in Vietnamese (cf. Romaine, 1989: 271, 272). For example, children's skills in Vietnamese and parents' success in bringing up their children are measured by the children's use of the right kinship terms and pronouns of address. To summarize the parental viewpoints above: L2 is associated with the future, with utility and necessity and never anything affectionate, while L1 has the meaning of keeping the past, of future concerns and of affection. Another aspect of interest is the extent to which parents are satisfied with their children's skills in both L1 and L2. There seem to be interesting differences between the groups studied. Turks tend to be quite dissatisfied with their children's skills in L2; in contrast, they seem to be generally satisfied with their children's skills in L1. The same kind of tendency applies to Finns in Gothenburg. However, the concern for the children's L2 is relatively small. Vietnamese in both countries seem to be mostly satisfied with their offsprings's skills in both languages, but more dissatisfied with L1. The same applies to North American parents in Gothenburg. Role of formal L1 instruction What role does the school play? Is there a place for the minority language in the curriculum? The range of provision of institutional support varies greatly in the Nordic countries. The minority language is usually taught as an optional subject for two hours per week. Moreover, there are some possibilities of getting Vietnamese-medium instruction in Bergen and Turkishmedium instruction in Køge. For North Americans, there are English-medium schools in all the localities chosen. As to the Finns in Gothenburg, they may choose between bilingual public education and regular Swedish classes for their children. How many of the informants choose to utilize the possibility of children having home language instruction? In practice, virtually all Turks and Vietnamese do. The ones that do not participate are either too young or the instruction is not available. These groups seem to rely heavily on the support given by the school. At the same time, the high participation rate may also serve as an expression of language loyalty. One third of the North American children do not receive home language

< previous page

page_185

next page >

< previous page

page_186

next page > Page 186

instruction. In six of these ten cases, this is deliberate, reasons being the following: 'the children didn't want to be different', 'they are not interested', 'they don't know enough English to do it', 'there are other, more important subjects', 'mother takes care of it'. On the basis of this data, the indications are that Americans do not set all that much store by the school's support, compared to the other groups. However, 50% of the children in the group do participate in home language instruction, and some of the children go to the English-medium schools. More than half of the Finnish children 3 participate in bilingual education, the rest go to Swedish classes. Children attending regular Swedish classes may also receive two hours of Finnish instruction per week either instead of other lessons or after regular school hours. Three of the families have not chosen to utilize this possibility. They report that the children did not want to study the subject any more or that it was an unnecessary extension to already long school days. Why did the parents choose to have home language instruction? Why is formal schooling needed? Why is it not enough to pick up the spoken form of the language at home? The reasons discussed above for maintaining the language apply here, too, but they may be supplemented with some comments on the contents of school instruction. The school seems to be given the responsibility to teach the children to read and write. Some of the Turkish and Vietnamese parents report to have done that themselves, but for the majority it seems to be the school's duty. There are also many other things that the parents expect their children to learn about during these two weekly hours: the history, culture, tradition and geography of the respective native countries. In addition, according to two Turkish and two Vietnamese parents, home language instruction gives a basis for the learning of the majority language. There are even some less committed reasons for participating in the home language instruction: 'because it is there'. This comment was made by two Vietnamese and one American informant. Likewise, for two representatives of the same groups, it seems to be something that complements the parents' efforts to teach L1 to their children. Finnish parents have chosen bilingual education for their children, e.g., because of their own minimal competence in Swedish; the parents feel they would not be able to follow up their children's education in a regular Swedish class. Other reasons mentioned by the Finns are the following: 'there is better discipline in the Finnish classroom', 'children didn't know enough Swedish to participate in Swedish-medium instruction',

< previous page

page_186

next page >

< previous page

page_187

next page > Page 187

and the same as mentioned above 'it was there in the neighbourhood school'. Are the parents satisfied with the kind of public support they get for the minority language? Do the groups make different demands? Is it for example the case that Vietnamese parents in general have a passive attitude toward their childrens's education (cf. Chung, 1988: 283-284). Do the Finns demand a lot (cf. the Finnish parents' school strikes, described, e.g., in SkutnabbKangas & Cummins, 1988)? Is it correct that the Americans do not set all that much store by the school's support? What kind of role do the Turks take in relation to their children's education? In the case of Vietnamese parents, it is interesting to note that they show a lot of interest in their children's education: in Finland 2/3 and in Norway 1/4 of the parents demand more hours for Vietnamese in the school curriculum. With two hours a week they feel that 'Vietnamese is treated like a foreign language rather than a mother tongue' right now. Parents also worry about the future of this instruction and criticize the content of it A big problem in Finland seems to be the fact that teachers do not always come from the same dialectal area in Vietnam as the pupils. The former division of the country is reflected in antagonism within the group and it shows up even in the parents' mistrust towards home language instruction. Parents complain that children do not understand what the teacher says or that teachers teach in the wrong way - they make mistakes in writing and they pronounce incorrectly - and parents have to correct their mistakes at home. Moreover, 'teachers lack education', 'they don't have proper methods or motivation', 'they don't have any books' or 'they have the wrong books'. Political disputes are hidden behind comments such as: 'I don't want my child to mix the dialects'. Finns, the largest and oldest immigrant group in our study, have had time and opportunity to organize relatively good institutional support for the Finnish language in Sweden. The parents interviewed so far use the possibilities that are available but do not seem to demand that much more. There is a central organization of Finns in Sweden the activists of which seem to create a kind of 'official' attitude 4. In practice, at least according to our data, it seems to be a vocal minority who makes these demands, and whose attitudes are not representative of those of the whole group (cf. Edwards, 1985: 142). Ordinary group members may feel under pressure to express the 'official' attitude: 'I don't have the right opinion in this matter', said one informant. On the one hand, it may be that North Americans do not rely so much on educational support for English, because it has a lot of

< previous page

page_187

next page >

< previous page

page_188

next page > Page 188

other kinds of support and also because they often have the means of providing these other types of support. On the other hand, half of the parents in our data would like to have more English home language instruction or even English-medium instruction for their children. One parent criticizes the content of the instruction and two parents are dissatisfied because home language instruction has not been organized for their children. Interestingly, seven of the Turkish parents have considered sending their children to Turkey to be educated there. One of the parents points out that, in Denmark, Turkish is always taught in the afternoon when the children are tired. Many parents would like to have bilingual education for their children: 'I would like to have a school where Turkish and Danish education would be side by side'. Methods used by parents to supplement formal education in L1 What do parents do at home to promote the children's bilingualism? In addition to speaking their own language many of the parents report having read or told stories to children in their native language. The Turks in Denmark watch Turkish programs and videotapes on TV and listen to Turkish radio programs. For Finns in Sweden there are also Finnish children's programs available on TV that the informants have taken advantage of. Some Americans and Finns report that children have participated in the activities of a play group organized by the American Women's Club and a Finnish social club, respectively. Vietnamese seem to try to supplement what is done in school by providing more formal instruction in the home as very many report teaching L1 at home, helping with homework and explaining what their children did not understand in home language instruction. One important way of activating the children's L1 are long trips made to their respective home countries. Finns travel to Finland at least once a year and they may get Finnish visitors even more often. Americans seem to travel to or to get visitors from the U.S. almost annually. Most of the Turks travel to Turkey at least every fourth year. For the Vietnamese, who are refugees, the situation is different. Even though it has been possible for them to travel to Vietnam for a vacation for the past couple of years, only a few families have done so. On the other hand, the Vietnamese have a lot of relatives in other countries whom they may visit and who may come to visit them. Face-to-face contacts with the relatives abroad are not yet as common as in the other groups, however. As to the Vietnamese in Finland, half of the group have visited another country or

< previous page

page_188

next page >

< previous page

page_189

next page > Page 189

have got visitors from abroad. On the average, such contacts have taken place about every fourth year. On the whole, it seems that Finns, Turks and North Americans rely on less formal methods of encouraging language acquisition at home, while the Vietnamese, partly due to circumstances, rely primarily on more formal methods. Attitudes and behaviour Earlier I pointed out that some Finns and especially Turks are dissatisfied with their children's L2 whereas Vietnamese and Americans seem to be more critical of their children's L1. One possible explanation for this may lie in the practice of language use in the families. In Table 2, I have used a very rough but demanding measure of language use in the family: L1 is always spoken by all the members in question. Table 2 Language use in the families: ''always L1" (n = 20 in the Finnish group, n =30 in the other groups). parent  children  children « children parent children 5 % % % Turks6 90.0 86.7 65.5 Finns 70.0 55.0 46.7 Vietnamese a. Bergen7 b. Helsinki 80.6 70.9 17.2 90.0 63.3 25.0 North Americans 46.6 26.6 20.8 In all five groups, there seem to be similar tendencies. As expected, parents speak more often L1 to their children than children speak L1 to their parents. Likewise, children use more L1 when speaking to their parents than to their brothers and sisters. There are also some interesting differences between the groups. The two Vietnamese groups differ from the other groups in one respect: the difference in percentage between the two latter categories (children -> parents, children <-> children) is twice as large as in the other groups. According to these tentative results, it is no wonder why Turks and Finns may feel more

< previous page

page_189

next page >

< previous page

page_190

next page > Page 190

secure with their children's L1. Compared with the other groups in this survey, the minority language is widely used in the Turkish families. In the Finnish families, Finnish is still used a lot especially among the second generation. Equally well it can be understood why Vietnamese and especially North Americans feel somewhat threatened in this respect. There is much less place for English in American homes compared to the other groups. Changes are also happening in the most recent groups of our survey; Vietnamese children speak more and more of the majority language among themselves. On the basis of these results it seems that language shift has taken place most rapidly in the Vietnamese groups. However, to their parents children still speak a lot of Vietnamese. This may be due to the fact that Vietnamese children are expected to show a higher degree of respect to their parents than e.g. Finnish children are. In Table 2, a very demanding measure 'always L1' - was used. However, if another definition 'mostly L1' were used, it would be even more evident that in only a few cases have the parents started using equally much or even more L2 than L1 with their children. These exceptional cases are also of interest. What prompts people to stop speaking L1 to their children; why is the language not passed on to the children? One of the three Turkish parents using half Turkish and half Danish when speaking to their children gives this as a reason: 'We want him to be better at Danish.' In other cases parents have not made a conscious decision or so it seems at least. Those North American parents who use mostly or only Swedish either blame the circumstances ('circumstances have made it impossible for the children to learn the language') or let the children decide ('initiatives must come from the children', 'if the children lack motivation, it is no use trying to push them', 'when the children need the language, they will learn it'). More than attitudes it is, in the end, a question of practical difficulties. It is very seldom that parents have negative attitudes toward their own language 7 or their children's bilingualism. It is rather the case that the families change their practices under pressure as time goes by. Pressures in the families are different In all-minority families pressure comes from the surrounding majority community, in mixed-language families the competition between languages is already there in the beginning8. When practices change, it can be an agonizing process. One Vietnamese family described a crisis where the two generations could not understand each other any more. According to the parents, the only way out of this crisis was the obviously very painful decision they made: to learn more Finnish in

< previous page

page_190

next page >

< previous page

page_191

next page > Page 191

order to communicate better with their children. Their affection for 'their own language' was still there, coupled with an admission that it was not doing much for their children compared to the language that was spoken in the country. Conclusions In this chapter, an attempt has been made to describe the tendencies found concerning parental attitudes towards bilingualism as found in the interview material of a joint Nordic project. The parents' perceptions of their children's bilingualism have been described from various aspects. Bilingualism was generally seen as something positive. L1 and L2 were valued for different reasons: L1 primarily for its social and affective values, L2 primarily for its utility. Parents expressed a strong desire for public support of minority languages by utilizing the existing possibilities or by demanding more of them. In addition to speaking mostly L1 at home, parents tried to maintain L1 in the family with the help of several methods e. g. by reading and telling stories to their children in their native language. The results presented above are to be considered tentative as all the data for the project have not been gathered yet. It will be interesting to see whether the tentative finding of the rapid language shift among the Vietnamese will still hold true when the results for the project as a whole are completed. The role of social networks, which has not been discussed here, will be a subject of further study also in terms of attitudes. Acknowledgements Due to the nature of the project, I am indebted to several people. I have benefitted much from working together with Paula Andersson, Harald Berggreen, Sally Boyd, Eric De Geer, Jørgen Gimbel, Jorid Hjulstad Junttila, Anne Holmen, Jens Normann Jørgensen and Pirkko Nuolijärvi. Data and parts of the analysis in this study were provided by Paula Andersson, Harald Berggreen, Sally Boyd and Anne Holmen. I would also like to thank Paula Andersson, Harald Berggreen, Sally Boyd, Eric De Geer, Anne Holmen and Jens Normann Jørgensen for encouragement and critical comments on an earlier version of this chapter.

< previous page

page_191

next page >

< previous page

page_192

next page > Page 192

Notes 1. The research project "The Development of Immigrant Languages in the Nordic Countries", led by Sally Boyd, is funded by the Joint Committee of the Nordic Humanistic Research Councils. It is being carried out at the universities of Bergen, Gothenburg, Helsinki, Tromsø, and Uppsala, and at the Royal Danish School of Educational Studies at Copenhagen. 2. I use the abbreviation L1 for 'first language' and L2 for 'second language'. By L1 I mean the immigrant parent's first language, and by L2 the majority language in each of the Nordic countries, even though in some cases these languages are learned simultaneously, or in another order by the children. 3. Because the data from the Finnish group only partially represents the stratified sample at the moment, the picture emerging on the basis of it may be biased. 4. This official attitude demands even more support for Finnish in Sweden, often comparing the situation of Sweden-Finns to that of Finland-Swedes. 5. In this column, n is different from that in the other columns, excluding families with only one child, i. e. n = 29 for Turks, 15 for Finns, 29 for Vietnamese in Bergen, 24 for Vietnamese in Helsinki, and 24 for North Americans. 6. Data from Turks in Køge and Vietnamese in Bergen concern a specific child in the family as opposed to all the children in the family as in the other groups studied. 7. E.g. those few cases where the parents wanted to have more instruction in L2 for their children consisted of families that had recently arrived in the country. 8. As Huss (1990) points out, in order to maintain the minority language in a mixed-language family, the minority parent has to have a strong interest in language, be self-confident and committed to his or her goals. It is even more important that the majority parent is personally interested in the minority language and culture. In this data, it is only among the North American group where the majority parent shows such an interest. In nine families out of 21, the parents speak mostly English with each other. In four families of these, the Swedish-speaking parent speaks mostly English to his or her children. In the other groups, the minority language is never spoken by the majority parent within mixed-language families. References CHUNG, C. (1988), The language situation of Vietnamese Americans. In: S. McKay & S. Wong (eds.), Language diversity: problem or resource? New York: Newbury House.

< previous page

page_192

next page >

< previous page

page_193

next page > Page 193

EDWARDS, J. (1983), Language attitudes in multilingual settings: A general assessment. In: Journal of Multilingual and Multicultural Development 4, 225-236. EDWARDS, J. (1985), Language, society and identity. Oxford: Blackwell. FISHMAN, J. (1972), The sociology of language. Rowley, Mass.: Newbury House. HUSS, L. (1990), "Arkinen aherrus on rikkaampaa kahdella kielellä" Vanhempien ajatuksia lastensa kaksikielisyydestä. [Child bilingualism as parents see it.] In: I. Almqvist, P.-E. Cederholm & J. Lainio (eds.), Från Pohjolas pörten till Kognitiv Kontakt. Stockholm Studies in Finnish Language and Literature 6. Department of Finnish, University of Stockholm. MILROY, L. (1980), Language and social networks. Oxford: Blackwell. ROMAINE, S. (1989), Bilingualism. Oxford: Blackwell. SKUTNABB-KANGAS, T. & J. Cummins (eds.) (1988), Minority education: From shame to struggle. Clevedon, Avon: Multilingual Matters.

< previous page

page_193

next page >

< previous page

page_ii

next page >

< previous page

page_195

next page > Page 195

PART 3 CODE-SWITCHING

< previous page

page_195

next page >

< previous page

page_ii

next page >

< previous page

page_197

next page > Page 197

9 Code-copying in immigrant Turkish Lars Johanson Immigrant Turkish Through migration from Turkey and some Balkan countries, considerable Turkish-speaking diaspora groups have emerged in North Western Europe. These migrant speakers of Turkish have various backgrounds and do not constitute any uniform 'ethnic minorities'. Linguistically, they are, as a rule, living in unbalanced, asymmetrical contact situations, their first language fulfilling only community functions. Whereas several aspects of their acquisition and use of the dominant languages in the host societies have been studied, little attention has been given to Turkish as the dominated language. Only some preparatory work has been carried out in this field, e.g., in some excellent contributions by A. Backus, H. Boeschoten, C.W. Pfaff, L. Verhoeven, and others. The following remarks are based on observations made in the course of the short history of Turkish in Scandinavia and Germany. It is obvious that the present situation is a transitional one with processes of language loss and shift in the second and third generation. However, the chances for the survival of Turkish, the institutional responses to its presence, or the language maintenance efforts of the Turkish speech communities themselves will not be discussed here. The focus will be on the development of Turkish itself and the changes it is undergoing. Variational patterns Very little attention has been paid so far to the question of whether new local varieties of Turkish are developing. To answer such a question, thorough investigations at the phonological, lexical, morphosyntactic, and textual level would be necessary. What observable linguistic features - as to what is used and what is not used - distinguish Turkish diaspora speech? In this respect, several distinctions are necessary: we find rather diverse variational patterns in individuals, domains, sub-

< previous page

page_197

next page >

< previous page

page_198

next page > Page 198

groups, generations. The development will depend on the specific linguistic environments. In particular, studies of differences between the generations are important, especially with regard to the input. Most phenomena found in the speech of the first generation will certainly not be typical of future Turkish diaspora varieties. In the non-first generations, competence in the socially dominated language A and the dominant language B differs considerably from that of the first generation. We find signs of delay and stagnation in the acquisition of Turkish (Verhoeven & Boeschoten, 1986), of failure to acquire and use certain linguistic devices, of heavy restrictions in the use of Turkish, of breaks in the linguistic tradition, and thus of imminent language loss, namely individual attrition and erosion from generation to generation. B influence An interesting but highly controversial factor is the linguistic influence from the dominant co-areal majority language. In current research, there is a certain tendency to disparage such influence, obviously as a reaction to all too naive and premature contrastive analyses of the linguistic results of contact situations. In our view, the study of this influence should be vitalized in less naive attempts which take into account, synchronically, the complex variational patterns of the 'donor' and 'recipient' languages and, diachronically, their previous developmental stages. It is also important to note that linguistic influence cannot be studied in isolation from the sociolinguistic setting. Here, we will focus on the influence of German as the 'donor language' B on Turkish as the 'recipient language' A. Particularly, we will take up certain theoretical aspects and dwell upon some issues concerning how the study of this influence could contribute to our general insights into contact processes. Many analyses in this field have been based on the classical models of Weinreich (1953) and Haugen (e.g., 1972, 1973). Our own model of analysis deviates from these in some important points. It is meant as a framework for describing linguistic contact phenomena in a coherent way, also as a basis for typological comparison; and we intend to use a simple terminology which is easily applicable to various cases of language contact. The model is being elaborated in a comprehensive study which will also include linguistic analyses of other Turkic contact situations, past and present. Here, the purpose is only to give a very short

< previous page

page_198

next page >

< previous page

page_199

next page > Page 199

overview of some of the issues involved. Several important aspects will be left out altogether. The considerations include criticism of some basic concepts of traditional contact linguistics, for instance, 'borrowing', 'transfer', 'switching', 'interference', 'integration', and 'substitution'. The misleading metaphorics suggested by such terms often strongly influences the way linguists conceive of the processes characterizing interactions between languages in contact Code-copying Since we focus on the possible development of new varieties of Turkish, that is, on what is happening with one of the two codes involved, we restrict our attention to cases where the basic structural pattern of the clause is Turkish. At the dispense of other possible phenomena of code interaction ('language-mixing'), the focus will be on the linguistic analysis of intraclausal codecopying. We refer to cases where the minority language is the basic code and the majority language the code that is (in the usual terminology) 'switched to' or 'borrowed from'. Thus we disregard cases of 'code-switching' implying alternate choices of codes, in the sense of changing code over clauses, e.g., when German clauses, sentences or longer passages of speech are inserted into a discourse primarily conducted in Turkish. 'Code-switching' is also inadequate as a general cover term, since it generally refers to the juxtaposition of elements belonging to different grammatical systems or subsystems. Thus even the notion of 'intrasentential code-switching' implies an alternation between two codes. We prefer 'code interaction' as a neutral cover term for (intra- or extraclausal) 'code-alternation' and (intraclausal) 'code-copying'. The clausal level is chosen on empirical grounds, since it proves relatively easy to define the syntactical frame there, that is, to decide whether the basic ('matrix') code is A or B. It might well be that some products of certain other contact situations (e.g., Spanish-English contact situations in the USA) are more difficult to identify in this respect. Admittedly, for Turkish spoken in North Western europe, too, there may be marginal cases in which it is impossible to assign a particular language to a given clause. Anyhow, purely quantitative criteria are not decisive for the language assignment. A clause may obviously be A-coded in spite of a heavy amount of B elements. Normally, it seems appropriate to apply 'language indexing' to the clausal level and not to assign a deviant index to etymologically foreign

< previous page

page_199

next page >

< previous page

page_200

next page > Page 200

elements (of B origin) used in an A clause. Apart from quotations and some similar cases, the use of a B-element in A practically always represents more than just a switch in language. Our point is that B-elements are not just inserted and juxtaposed to A-elements. The clause has a strict structural economy, and there is always some adjustment involved, in some way and to some degree. Thus we will not say that German elements are 'switched' into Turkish speech, even if they are used only very occasionally. Kinds of code interaction Under the general heading of code interaction, we distinguish 'code-alternation' (CA) from 'code-copying' (CC), and 'extraclausal' code-alternation (ECA) from 'intraclausal' code-alternation (ICA). Finally, we distinguish 'momentary' code-copies (MCC) from more or less 'conventionalized' ones (CCC). As for the situation within the clause, we thus prefer a tripartite classification: ICA, MCC, CCC. ICA produces 'plain codeswitches', that is, B items that are not incorporated at all. Their status may be indicated by cues such as hesitation, asides, and translation or paraphrase (cf. Pfaff, 1979: 297). In the case of CC, a B item is copied into A, globally or not (v. infra). Calque phenomena are also taken to be cases of code-copying, not a different type of interaction. MCC refers to sporadic, ephemeral instances of copying (elements used 'only once'); it corresponds to Weinreich's 'nonce-borrowing' (1953: 11), which relies on the basic difference between langue and parole. Correspondingly, CCC refers to copies that are habitually used and have more lasting effects in the linguistic community. A further distinction might be based on the criterion whether a CCC necessarily presupposes bilingual ability (some degree of competence in both languages) or may occur in utterances of monolinguals. Note that examples of established 'loans' (e.g., from the history of Turkic languages) are also useful in general discussions of CC, since MCC must once also have been the point of departure for their conventionalization processes. Terminology Terms such as 'switching' and 'borrowing' have often been used to distinguish not only between different degrees of conventionalization but also, for instance, between 'morphologically' non-integrated and

< previous page

page_200

next page >

< previous page

page_201

next page > Page 201

integrated elements. Some researchers have even taken the distinction to be a matter of volume, discussing how large a chunk of B must be to represent a 'switch'. We doubt that it is possible to distinguish linguistically 'switched' elements from 'nonswitched' ones in a consistent way with such criteria, especially when they are mixed. Our own terminology reflects a classification which enables us to treat closely related phenomena, i.e., all types of copying, in a uniform theoretical framework, instead of postulating rather arbitrary absolute differences which cannot be consistently applied. The use of elements (units or qualities) emanating from one code within the context of another is often referred to as 'interference'. We also avoid this term because of its negative connotations implying deviations from monolingual norms, suggesting a phenomenon which causes impaired communication. The process of conventionalizing a 'foreign' element is a continuum with gliding transitions between degrees of acceptability. By using the term 'code-copying', we want to be able to refer to the insertion of a 'foreign' element without having to specify this degree and to determine at what point it ceases to be an 'interference'. So-called 'negative' and 'positive' effects of bilingualism are not our concern here; nor are the effects of the knowledge of L1 on the acquisition of L2. In recent literature on language acquisition, 'interference' often refers to negative effects of this kind, generally as opposed to [positive] 'transfer'. By avoiding such terms, we also want to avoid confusion with the concept linked to them. The above-mentioned term 'borrowing', which stands for one of the basic concepts of traditional contact linguistics, is based on a highly misleading metaphor. Nothing is borrowed in language contact: the 'donor language' is not deprived of anything; and more importantly - the 'recipient language' does not take over anything which is identical with an element in the 'donor language'. This is rather obvious and may seem trivial, but the use of the metaphor often has undesirable consequences. The same danger is connected with the term 'transfer'. It is too often ignored that there is no identity involved: the elements of the 'donor language' merely serve as models for imitation. Copying As already indicated, the concept of copying would account more adequately for the relevant phenomena described as 'borrowings' or 'transfers' in the literature. The basic concept behind the term is that linguis-

< previous page

page_201

next page >

< previous page

page_202

next page > Page 202

tic elements - both units and patterns - are copied from B to A. (Both segmental units as 'blocks' of qualities and individual qualities themselves are referred to as 'elements'; v. infra.) Therewith, the linguistic copies belong to other systems than their originals, which has important consequences. The copying can be global and selective. Global copying means that B units-'blocks' are copied into A as a whole. Selective copying implies that just one or more selected structural properties of B elements are copied onto A elements. Here we can only hint at some of the issues involved in copying processes and at some principles of classification. Alpha lects Code-copying does not mean any fusion of A and B, e.g., of Turkish and German. It is, however, typical of certain functional varieties of Turkish which - as part of the linguistic repertoire in the immigrant situation - are used for in-group communication and which we shall refer to as 'Alpha lects'. The same speaker of A will behave differently depending on the type of interlocutor involved. In communication with another bilingual who shares the same linguistic background he may use Alpha. If the interlocutor is not bilingual, Alpha is not likely to be chosen, since it would imply a breakdown in communication. The social consequences of the lect choice are predictable. Note that this definition does not specify the amount of code-copying. Since hardly any first-generation Turk is purely monolingual, the varieties of Turkish used in the diaspora communities at large are in principle not 'normal Turkish' any more, but already Alpha lects, even if they happen to be rather modest in code-copying. There are, however, also Alpha varieties which display extensive code-copying. In particular, young and more competent bilinguals of the second generation may use a relatively low-copying Alpha variety when speaking to their elders (or other firstgeneration Turks with a more limited B competence) and another relatively high-copying variety with each other. A highcopying variety reserved for in-group conversation between Holland-Turkish bilingual peers has been described by Backus (1989). Here we will not discuss the social reasons for the frequent code-copying sometimes encountered among adolescents growing up in an immigrant environment, for instance, possible 'neutrality strategies' to avoid choices expressing preference for A or B (as, e.g., in code-alternation).

< previous page

page_202

next page >

< previous page

page_203

next page > Page 203

Sources and differences The sources for copying are specific varieties of German which are usually also reflected in the bilingual speaker's own German. For the first generation, these are often, initially, makeshift communicative varieties which we refer to as 'Beta lects'. Later on, the speaker may copy elements from more native German varieties which are also likely to be reflected in more advanced stages of his German interlanguage. For the non-first generations there is a more complex input. The B elements serving as models may belong to the speaker's own German, which is, in general, relatively strong. They may, however, also belong to the Turkish Alpha lect of the first generation. (Examples of various types of copies in the language of the second generation of Turks living in Germany are to be found in Menz 1991.) There are various kinds and degrees of difference between original and copy with respect to certain properties. The differences may be considerable, and there are often structural reasons for them. Before making any statements on contrastive grounds, it is, however, necessary to identify the source of the copies. The result of copying cannot be analysed in terms of adaptation if it is not clear which variety of B the original belongs to. Anyhow, the properties of the copies do not always reflect the B proficiency of the speaker directly, that is, not all differences are (as is sometimes claimed) due to 'imperfect learning' of B. As just stated, B elements found in the speech of members of the second generation (with a native or native-like mastery of B) may also emanate from an Alpha 'home language' spoken by their parents. Global and selective copying in an unbalanced dominance situation is likely to lead to unidirectional convergence, making A more similar to B. Surface structures common to A and B provide equivalence positions and thus favour copying. (This might also mean that typological closeness of A and B lead to relatively rapid convergence.) On the other hand, all kinds of adaptation to be discussed further on, material reshaping, combinational restructuring etc., create more similarity with A and thus reduce structural conflicts. Global copying Global copying means insertion of copies of stretches of B speech, morphological-lexical-phraseological units of various kinds which we will refer to as 'blocks'. These are items which possess a material

< previous page

page_203

next page >

< previous page

page_204

next page > Page 204

shape, and may be morphemically simple or complex, bound or free, comprise one or more words, and belong to different morpheme categories, word classes and phrase types (including 'petrified' or 'idiomatic' expressions). They do not simply represent 'the phonemic shape of a word and its meaning' (in the sense of Haugen, 1953: 2, 90), but constitute blocks of different properties: in particular material, combinational, and semantic ones. Insertion The basic A-code provides the clausal frame, normally including combinational patterns and function elements. The global copy is inserted into one of the specific slots which its A 'equivalent' would fill; we refer to individual positions of this kind as 'insertion points'. Note that we speak of the clausal frame and the insertion of copies in a diachronically linguistic sense, and do not claim that, in any psycholinguistic sense, clauses are 'produced' or 'processed' in these two steps. The insertion is basically a question of the speaker's subjective assessment of surface-structure equivalence. Lack of real typological equivalence does not, however, prevent insertion. The global copy normally assumes grammatical A-morphology signalling clause hierarchy, e.g., banofa gittim 'I went to the station'. Also, copies larger than single words are 'integrated' into grammatical B structures in this way. This kind of 'adaptation to A morphology' not only means that bound morphemes are attached (banof-a 'to the station'); it is an insertion in the same sense if the function marker is an unbound one: (banof önünde 'in front of the station'). Consequently, it is just a natural result of the insertion if - in the awkward perspective of 'switching' theories - there seem to be 'asymmetrical patterns' which maintain the integrity of the A syntax, but one-sidedly 'suspend' B grammar. E.g., in the example just quoted, a Turkish suffix morpheme -(y)A and a Turkish morpheme order are used; German rules would require a yap- 'to write preposition (zum, with inherent definite article). According to our framework, in cases such as berist *-i the report', units which are intimately connected in B (here: object Bericht + verb schreiben) are not copied as a whole (e.g., den Bericht schreiben as an actional phrase), but as separate blocks, each one assuming function markers provided by the clausal A frame. thus, if B function markers (articles, plurals etc.) do not occur, it

< previous page

page_204

next page >

< previous page

page_205

next page > Page 205

is not because a whole B complex has been copied and undergone grammatical suspension. The very fact that German blocks are freely inserted into slots followed by bound Turkish clause-syntactic markers, falsifies the allegedly universal 'free morpheme constraint' (v. infra). Moreover, this morphological 'integration' into A cannot be referred to as a 'switch' from B to A, since the A clause is the very frame for the insertion. (Problems concerning what is actually already present in the clause when a copy is inserted cannot be discussed at length in the present chapter.) Nor can it be called a 'substitution', since there have never been any B function markers to replace. Thus this 'integration' is no criterion for distinguishing degrees of conventionalization. A morphemically complex A constituent (word, word-group, phrase) which is either entirely native or consists entirely of one or more B-copies is referred to as a solid complex. A mixed complex has some native and some copied part(s). Stages of development In the first generation of Turkish-speakers in Germany, there is, at an early stage, relatively large tolerance of globally copied German blocks, almost exclusively free forms, predominantly 'content units', mostly single nouns. Bound blocks and/or 'children', 'roll'. grammatical (derivational or relational) blocks are mostly copied as parts of complex ones, e.g., Although the clausal frame provided by the basic A-code normally includes the necessary clause-hierarchic function markers, these may also be replaced by an equivalent marker within a complex copy from B. Thus a complex such as sum banof 'to the station' may function like banofa. Globally copied German prepositional phrases show a high degree of structural equivalence with Turkish postpositional phrases with respect to external combinational properties. However, the originals of such complexes are often idiomatic, stereotyped phrases. At later stages, bound and/or grammatical blocks are copied isolatedly and used productively. This is valid for derivative 'content' units as well as for relational units (free or inflectional 'function elements'). Although the clausal A frame normally provides the clause-hierarchic function markers, these may also be replaced by productive global copies of equivalent B markers. The history of Turkic shows many cases where B function markers are - contrary to some other constraints

< previous page

page_205

next page >

< previous page

page_206

next page > Page 206

and principles proposed in the literature - used productively; cf. the copying of many Persian, Arabic and Slavic function words (such as conjunctions). Needless to say, it is often difficult to draw the borderline between content and relational units, e.g., in the case of B auxiliaries. The originals Between the original and the copy there are similarities and differences, which should be described At first, therefore, we should try to define which originals serve as the object of copying. Absence of elements such as gender/number inflection or articles does not represent syntactic 'simplification', if these elements have never been copied. No inflection or article has been 'lost' or 'deleted' if there is no reason to assume that it was present in the original copied. It is sometimes said that B grammar is 'suspended' within B stretches when they lack 'syntactic cohesion' (Muysken, 1987). Some copies which seem to lack syntactic cohesion have never had any 'fuller structures' to be syntactically 'simplified' or any function elements to be 'reduced'. Some 'telegraphic switches' (Boeschoten, 1991) certainly go back to global blocks which do not contain any 'fuller structures'. Others may be the results of material and/or combinational restructuring (v. infra). Substitution and reproduction Relatively close similarity to the original may be called 'reproduction'. Haugen's term 'importation' for successful copying again suggests that the corresponding units in B and A are identical. Replacement of B properties by others is known as 'substitution'. The differences are often due to adaptation, modification in the direction of A because of the grammatical and lexical incongruence between the two codes. The restructuring may, however, be basically A-independent and impossible to explain by simple contrastive analyses. Many copies which seem to lack 'syntactic cohesion' or 'integrity' - types reflecting 'a certain telegraphic sloppiness' (Boeschoten, 1991) may have been subjected to material and/or combinational restructuring. The analysis of such cases seems to suggest that typological distance does not favour 'suspension of grammar' in particular, but restructuring of code-copies in general.

< previous page

page_206

next page >

< previous page

page_207

next page > Page 207

As for the first generation of Turkish-speakers in North West Europe, there is much substitution at an early stage: little tolerance to foreign structures, imposition of A-properties upon the copied blocks. More advanced acquisition of B at later stages renders more similar copies possible. It is, however, obvious that differences between originals and copies do not always decrease steadily, as is often presumed. The modifications do not seem to reflect the speaker's level of B proficiency in a straight-forward way. Material reshaping With regard to material reshaping - restructuring of the shape of the block - there is a gliding scale, a continuum from reproduction to very extensive substitution. There is, inter alia, phonic adaptation in that phonic A-properties are substituted for B-properties. Thus, [x] in Dach 'roof may be rendered as [k], [ç] in Küche 'kitchen' as [h], etc. Differences in shape between originals and copies are often due to systemic differences of the elements in question. What is substituted for what, largely depends on how B segments are perceived and phonologically classified in A. For instance, since Turkish [k] in front syllables is generally predorsalpalatal, German [k] sounds '(crime) thriller' (copied from Krimi). Phonotactical B even in front environments tend to be classified as minus front, e.g., properties are replaced by A properties, e.g., in '(reported) sick' (from krank), 'night-duty' (from Spätdienst). Accent properties (stress, high tone) may be replaced by A-properties, e.g., [bá·:nho:f], [ú·nfal] by [báno·f], [únfa·l]. There is also more comprehensive reshaping of blocks: Weihnachten 'Christmas' is copied as vaynak, Überweisung 'transfer' as etc. It may go so far that a typical, salient single element of the block is selected to represent the whole of it, e.g., yap- or even an yap- from sich anmelden 'to register'; cf. Holland-Turkish spoel from gespoeld 'rinsed' (Backus, 1989). Copies of B adjectives are often inserted attributively in uniform, generalized shapes. There is absence of gender and number man 'the evil man'. An adjective may also be copied inflection on borrowed adjectives in reshaped complex blocks such as as such, not as part of a nominal phrase; e.g., in doyç adamlar 'German men', the copy of the variable adjective (deutsch) also exhibits a generalized shape. Generalized adjective forms may even be conventionalized. Thus, in some

< previous page

page_207

next page >

< previous page

page_208

next page > Page 208

Soviet Turkic languages, adjectives copied from Russian are used with generalized agreement markers (e.g., social'nïy naukalar 'social sciences'). Much of what has traditionally been classified as 'grammatical adaptation' belongs, in our view, to material reshaping. In a Holland-Turkish example such as op oog gözlük var 'on the eyes there are glasses' (Backus, 1989: 37) there is obviously no Dutch block op oog, in which Dutch grammar could have been 'suspended', just a reshaped material copy of a fuller Dutch 'employee' (from Angestellter). prepositional phrase. The same is valid for word-internal structures, e.g., As for the first generation of Turkish immigrants, there is at an early stage often strong and unpredictable reshaping, and, at later stages, more systematic substitution. Members of non-first generations may totally dispense with imposing Turkish phonic properties upon German blocks. However, they do not always. Apparently, there is not simply more reproduction, the more advanced stages of B acquisition are reached. The material shape of global copies is not permanently growing more similar to that of the originals. The whole problem is partly a question of the sources the copies are taken from; but it also concerns the diachrony (with respect to conventionalized B elements) of the specific Alpha lect serving as the 'matrix' code for insertion. We shall return to this question further on. Grammatical modification As for grammatical modification, the copies are realigned on the morphosyntactical A structure, classified - on the basis of some equivalence - into morphosyntactical categories to be able to function in the A clause, that is, prepared to be inserted into specific slots and to assume grammatical A-morphology signalling clause hierarchy. As already indicated, however, this very insertion does not mean any substitution, since the A clause is a natural frame already present to accept the global copies instead of A-blocks. If no element with clause-syntactical functions has been copied, none can be replaced. To be inserted as predicate cores, copies of German verbs are normally morphologically accommodated by a word formation device well-known from other Turkic contact situations, the set conversion formula infinitive + yap-, where the last element 'she wrote a letter', carries the predicator (predicative inflections). Thus, in bir mektup

< previous page

page_208

next page >

< previous page

page_209

next page > Page 209

the copy consists of , whereas the whole stretch yap- 'to write' has been inserted into the clause. In some cases, the accommodation does not take place; cf. the Norwegian-Turkish example kjøp-tüm 'I bought', where the copy of a Norwegian verbal stem is inserted directly (Brendemoen, 1987). The case ispiyonla- 'to spy', however, is essentially different, representing a Turkish denominal word formation from ispiyon 'spy' (cf. Spion). The accommodated element may be a more comprehensive yap- 'to lay tiles' (from Fliesen legen); cf. spoken Soviet Uzbek cases like ociri * zaynyat qil- 'to actional phrase, e.g., ). hold the place' (Russian One aspect of grammatical modification concerns change of combinability properties, often according to A patterns. Bproperties of combinability may be replaced by A-properties. A copy of the German word Gold 'gold' may assume the syntactic combinability of Turkish altin, so that uses like golt saat 'gold watch' are possible. Changes in the combinability of copies of Unfall 'accident', Umbau 'rebuilding' lead to such uses as unfal ol- 'to have an accident' and umbau ol- 'to be rebuilt, altered'. B auxiliaries such as muß may also shift word-class. In complex blocks with grammatical-relational constituents, e.g., in the plural 'children', the directive expression sum banof 'to the station', or the syntagma as 'as a baker', combinability properties 'the children' (used with a Turkish plural), sum banofa (gitti) (used with a Turkish case form may also be replaced: olarak (with olarak as a Turkish equivalent to al. indicating direction), as This means that the etymologically German function marker of the complex block may remain as a functionally redundant (possibly reinforcing) element. This does not imply any real duplication from a synchronical point of view, since the former B marker is not 'productive'. This case is encountered in other Turkic contact situations, where globally copied prepositional phrases involving etymologically non-Turkic prepositions are inserted in positions which entail Turkic case-marking. In Iranian Azeri, copies of Persian prepositional phrases such as hareg* äz sahr*'outside the town' are marked with Turkic case endings, which allow for adverbial use of the phrases, e.g.,harig* äz sähr dä (Kiral, 1991: 20). Analogous structures in Taijk result from a reverse process: complex results of copying of this kind in Uzbek dialects serve as models for further copying in the opposite direction. Thus Turkic case suffixes are used 'productively' as function markers in the Taijk clause frame, taking over (or reinforcing) the signalling of adverbial function.

< previous page

page_209

next page >

< previous page

page_210

next page > Page 210

From the synthetic point of view chosen so far, the blocks are taken to be parts of higher functional units. However, complex blocks can also be described analytic-ally, as functional units which are decomposed into constituents. As such, they can also be said to have internal combinational properties, which may be replaced by A properties. This kind of restructuring cannot be discussed in detail in the present chapter. Complex global copies are often the object of both material and combinational restructuring, as in the Holland-Turkish example collected by Backus (1989: 36): reet interesseren yap- 'not to interest [somebody] a damn' (without object and geen 'no'). Semantic modification As for the semantic modification, B-properties may be replaced, especially by A-properties, e.g., havuz (copied from Haus) is used for 'home' by identification with ev 'house, home', the significate of which is substituted for the German one. German influence may explain the underdifferentiating use of dogru * 'right' for 'really' (gerçekten) (cf. wirklich, which means both), or the confusion of synonyms such as baska*, diger*, öbür, öteki (Csató 1988); cf. anderer. We find many examples of seeming (cf. krank 'sick') 'unfit for work', (cf. semantic modification in Turkish spoken by the first generation. Copies such as Heim 'home') 'hostel' obviously differ in meaning from their Standard German equivalents. As long as it is not clear from which variety of German they are taken, they cannot, however, be analysed with any certainty in terms of adaptation. Syntactic constraints There are certainly syntactic constraints on global copying, that is, as to the syntactic contexts in which it is possible. Although it would be interesting to try to determine the possible orders of constituents in A sentences (clauses, morphemes, words, wordgroups, phrases) with regard to their B etymology, we will not discuss this issue here. Many investigators have tried to determine, e.g., at what points in a sentence a speaker may 'switch' to another language, and some allegedly universal constraints have been formulated. As for the 'free morpheme' and 'equivalence' constraints (Poplack, 1981; Poplack, Wheeler & Westwood, 1987), see Clyne, 1987, Boeschoten & Verhoeven, 1987, Backus, 1989, and Eliasson, 1989.

< previous page

page_210

next page >

< previous page

page_211

next page > Page 211

According to some constraints claimed in the literature, an A-affix would not, for instance, occur after a phonologically unassimilated B-stem. Thus banof-ta 'at the station' would be possible, whereas a more reproducing German pronunciation such as [ba:nho:fl, typical of the second generation, would render A-suffixing impossible. Counterexamples from many languages show that such alleged constraints just represent tendencies. In fact, word-internal global copies generally do occur more rarely than others. To take care of the exceptions to the 'free morpheme constraint', the notion of 'nonce-borrowing' has been adopted (Poplack & Sankoff, 1988), an auxiliary theoretical concept whose function is 'to protect the original theory from criticism' (Eliasson, 1989), and which leads to some circular reasoning (Boeschoten, 1991). Global copying in immigrant Turkish is obviously less restricted than these constraints allow. Generally, they do not even pay attention to very different functions of A and B elements in A-based clauses. In fact, some of the rules formulated by Hasselmo (1972) for morpheme orders within mixed complexes (e.g., lexemes with derivational and inflectional elements) seem to be more valid than most constraints claimed later on. In our view, however, the situation cannot be described properly without recognition of the concept of insertion already sketched and the concept of copied combinational patterns to be discussed below. Selective copying Global copying is primary to all other types. There is, however, also a good deal of selective copying in the sense that only certain aspects of the block serve as the model for imitation. This influence manifests itself as 'loan phonology', 'loan semantics', 'loan syntax', etc. There are, for instance, material, combinational, and semantic features in immigrant Turkish of Germany which owe their existence to German patterns although they do not occur in blocks of German origin. In the first generation, selective copies seem to appear at a late stage. In the second generation, they are rather common. In some varieties there is even excessive global and selective copying. Since selective copies appear at more advanced stages of acquisition, differences between originals and copies are often less marked, but here adaptation and other kinds of restructuring are possible also.

< previous page

page_211

next page >

< previous page

page_212

next page > Page 212

Haugen characterizes non-phonic copying as 'substitution', as if some phonic B properties had been replaced by A ones. This is misleading, if no global copying has taken place in the concrete case under investigation. Analyses of possible substitution in non-material copying cannot be based on comparisons with originals which also include material aspects. If no whole block with material properties - has been copied, there are no phonic properties to be replaced. It is sometimes even claimed that lexical 'loan translations' retain their source language shape, but show 'morphemic substitution'. If the term 'substitution' is used for, e.g., phonological adaptation, whereby A-properties are really substituted for B-properties, at least it cannot be used here in the same sense. Material copying One type of selective copying is an exclusively material one. Phonic aspects of B blocks may serve as models for imitation, so that segments and patterns typical of B are copied onto A blocks. Thus diaspora Turkish - especially in non-first generations, which are more familiar with B pronunciation - may acquire distinct phonological structures by copying sounds (such as, in Turkish spoken in Sweden and Norway, the pronunciation of rt as a kind of retroflex t, e.g., in dört 'four'), phonotactic patterns (e.g., clusters without any intervening vowel, as [sp] instead of [sïp] in spor 'sport'), or foreign accent patterns of stress and tone. Here, again, differences between original and copy can, in many cases, be explained as adaptation to the phonological system of the 'matrix' code rather than as the result of imperfect learning. On the other hand, we do not, of course, claim that every innovation of this kind is due to code-copying. In the history of Turkic, however, foreign phonic material has often been copied (e.g., form Iranian) and clearly adapted to the phonological systems of the 'recipient' languages (Johanson, 1990). Semantic copying Selective copying can also be restricted to non-material aspects. Exclusively semantic copying causes changes in content; cf. 'star', which has also assumed the meaning 'famous actor, singer, etc.' under English influence. There Standard Turkish may be differences at the denotative or connotative level. The semantics of a Turkish block

< previous page

page_212

next page >

< previous page

page_213

next page > Page 213

may be re-aligned under the influence of a German equivalent, e.g., altinda 'under' may be used for 'among', influenced by unter; or dur- for 'to be written', influenced by stehen. Combinational copying Copying of combinational properties typical of B blocks onto equivalent A-blocks is manifested by word order divergences, verbs used with unusual complement structures (e.g., bir kimseyi sor- 'to ask somebody'; cf. jemanden fragen), new rules for lexical subcategorisation, and other phenomena which coincide with German syntax (e.g., number marking: iki kardeslerim * 'my two brothers'), including word-internal syntax, e.g., redistribution of morphemic patterns within words. Copied combinability patterns may be more or less generalized and habitualized, for instance, leading to the emergence of new distributional classes (e.g., 'word classes') in A. Thus it is claimed that some Soviet Turkic languages have, on the model of Russian adjective endings, developed corresponding suffix classes. Analytically regarded, the internal combinational pattern typical of a complex B block can also be copied onto an equivalent A unit. Standard Turkish has, for instance, developed internal word-patterns such as bitkibilim 'botany' (instead of bitkibilimi or bitki bilimi). A phrase such as her ikinci hafta 'every second week' (instead of iki haftada bir) seems to copy properties from jede zweite Woche. Traditionally, non-phonic copies of at least combinational properties are mostly referred to as 'calques'. Semantic-combinational copies (e.g., çorba ye- 'to eat soup' instead of çorba iç-) are called 'loan translations', but generally only within the lexicon. Combinational copying may lead to under- or overmarking with respect to functional elements. There may be omission of the genitive suffix of the subject in lâzim-constructions under the influence of German müssen-constructions, or of the possessive 'he brushed the [= his] teeth' (cf. er putzte die Zähne) or in benim para var 'I have got money'; cf. suffix in disleri* German marking of the 'possessor', but not of the 'possessed' in ich habe Geld (Pfaff, 1988). On the other hand, the use of the indefinite article bir or of pronominals may be extended. In many cases, common Turkish does not spell out anaphoric pronouns for actants whose referents are recoverable from the immediate discourse (so-called 'pro drop' in generativist and cryptogenerativist terminology). Some diaspora Turkish varieties seem to be more explicit in this respect.

< previous page

page_213

next page >

< previous page

page_214

next page > Page 214

Also with respect to combinational copying there may be considerable differences between originals and copies, particularly due to adaptation to the A system. Thus some Turkic languages display clauses which seem to be right-branching subordinated complement and relative clauses, provided with finite verbs and introduced by conjunctions and relative elements. However, there are usually great differences between these imitations of Indo-European hypotaxis and their originals because of considerable adaptation to the Turkic systems. Combinational patterns can be copied onto solid as well as mixed complexes. Thus a mixed complex may assume a B morpheme order. Note that, although the A-coded clausal frame provides the combinational patterns, these can be changed by combinational copying. This leads to syntactic convergence between A and B. It is not less important to notice that, eo ipso, new equivalence positions for possible global copying come into being, an effect which might even be a reason for combinational copying. Obviously, internal combinational properties also determine the degree of complexity, which means that simplification of a complex A-block may be the result of a copied B-pattern. Even if some of the simplification met with in the second generation, grammatical reduction, reduction of the formal A-inventory (e.g., in the complicated Turkish verb morphology) reflects independent erosion tendencies, it may, at least partly, be due to copying of foreign combinational properties. Copying of frequency patterns Even the frequency of use peculiar to a B block may be copied onto an equivalent A-block so that the latter undergoes increase or decrease in frequency of occurrence. Of two stylistic A options, one may be favoured at the expense of the other. This also means that 'rejection patterns' may be copied from B to A. Thus Turkish spoken in Germany seems to exhibit an increased use of plural markers and deictic elements. In solid as well as mixed complexes, combinational patterns, e.g., constituent orders, that already exist in A but are more normal in B may gain ground in A and become less marked. Diaspora Turks may, under foreign influence, increase their use of SVO order, which is admitted but more marked in Turkish. Thus German influence may lead to underdifferentiation: reduced use of the rich word order devices of Turkish. Postpositive complement clauses with finite verbs may be used more frequently, at the expense of the genuine Turkish infinitized types

< previous page

page_214

next page >

< previous page

page_215

next page > Page 215

based on verbal nouns. Reichl supposes that, in Afghan Uzbek, 'the imitation of the Persian subordinate clause has led to a decrease of participial and gerundival constructions' (1983: 490). Even if the choice of such less synthetic constructions may, as has been claimed, be due to a general tendency towards simplification of grammar, its frequency of use may also be influenced by foreign, e.g., German, subordinative patterns. Similar shape As is well known from many language contacts, similarity in shape between an A block and its B equivalent may favour selective copying. The properties copied may be of a semantic nature, as in the case of the relatively homophonous sans 'luck' and Chance, direksiyon 'steering-wheel' and Direktion (Turkish müdürlük), or kontak 'ignition' and Kontakt. Onto blocks such as [táksi·], [puroble·m], German accent patterns may be copied: [tá·ksi], [puuróblé·m]. Similarly, a frequency pattern may be copied, so that Turkish words such as oto and even rare ones like optimal are activated under the influence of Auto and optimal. Mixed copies Apart from the 'pure' types of global and selective code-copying, there are also mixed types in the sense that a selective copy of (at least) combinational properties contains one or more global copies. yaz- 'to report sick', In Turkish spoken in Germany, we often find mixed copies of the type uban al- 'to take the tube' and where global copies are used as lexical cores in copied combinational patterns. Within the lexicon, this 'idiomatic' type is often called 'loanblend' (Haugen). We take it, however, to comprise all kinds of mixed global + combinational copying. With regard to diachronic development, a mixed copy often represents a transitory stage between a complex global copy and a purely combinational one, e.g., a B lexeme is copied globally together with some of its combinational rules. Holland-Turkish cases such as vrij al- 'to take off [work]' (cf. Dutch vrij nemen) quoted by Backus (1989: 32), belong to this type. We doubt that this is gel- 'to come to power' since here the combinational pattern is already present in Turkish iktidara gel-. valid for uitmaken yap- 'to finish one's relation with' However, as Boeschoten correctly states, in

< previous page

page_215

next page >

< previous page

page_216

next page > Page 216

(cf. het uitmaken met), the subcategorisation is 'copied into Turkish' (1991). In several Turkic languages, loan combinations such as in the postpositive complement clauses already mentioned have been used together with globally copied junctors, e.g., ki (anladim ki gelmez 'I realized she wouldn't come'). However, there are obvious differences between the original combinational patterns and the Turkish replicas, which are subject to several syntactic restrictions (Johanson, 1975). The structural equivalence between ki and Indo-European subjunctors is usually questionable. Some investigators have, however, expected the frequency of ki constructions to increase in immigrant Turkish, viz. under the influence of subordinated clauses introduced by subjunctors like German daß. In fact, such frequency copying has not been verified. One reason for the scarce use of ki constructions certainly is that daß and similar subjunctors offer no real equivalence positions for it. Remodeling of complex blocks If we consider the totality of copying devices, we conclude that complex blocks can be remodeled in diverse ways. We often find highly creative formations that presuppose the ability to analyze the B originals, to copy them or parts of them, and to rearrange the copies synthetically. The speaker can use mixed techniques, copying combinational and semantic properties onto i(Kindergeldstelle). The global copy may retain the A complexes which may also contain global copies, e.g., (Schule), in allomorphic form it has in the complex B model, even if this model is not copied as a whole, e.g., sul *, and not the compound sul binasi(Schulgebäude) 'school building'. Not every solid complex of B etymology must have been copied from B as a whole. The speaker may also use an analyticsynthetic technique, copying a complex block, but replacing parts of it by corresponding global copies known from other B contexts, replacive constituent copies. Thus a B complex originally serving as the model may in fact only be a 'quasi-original'. and ars kama, the constituents burot and ars are not parts of original copies of Brötchen and Ärztekammer, but are In themselves copies of Brot and Arzt, i.e., corresponding forms found in other B contexts. In other cases, larger parts of the complex - inflectional endings, articles, pronouns, etc. - may be absent or changed in the final A copy. At least in principle, such creative techniques should be distinguished from the purely material

< previous page

page_216

next page >

< previous page

page_217

next page > Page 217

reshaping discussed above. In practice, however, there are many ambiguous cases (e.g., Angestellte[r]).

'employee', copied from

Suspension of function marking One interesting phenomenon which we cannot discuss at length here is the possible suspension of function marking within the A-coded clausal frame. The latter may be simplified so that both A or B function markers are absent. In this case, if the B blocks involved have been copied without any grammatical 'garment', they cannot be considered to be grammatically 'stripped' either. Backus quotes interesting Holland-Turkish examples of bare forms such as the adverbially used nominal phrase handdoek '(with) a towel' and the direct object in kendi prijs söylüyor 'he names his price' (1989: 30), where the copied blocks are not marked with any function units (postpositions, case, possessive or accusative endings). Cases of obvious B influence (e.g., bin dokuzyüz yetmis * bes* buraya geldim 'I came here in 1975', cf. ich kam 1975 hier-her) are, of course, to be treated under the heading of combinational copying. New norms Code-copying is a normal developmental process and highly prominent in all Turkic languages. Its resultant distinctive speech characteristics may disappear or remain. Momentary copying is the point of departure for habitualization and conventionalization processes. Code-copying starting as performance phenomena may have diachronic effects on the language, i.e., lead to change. Copies may occur frequently, regularly, and even normally. The way from MCC to CCC is a continuum of changes in the sociolinguistic status, of degrees of recurrence in the individual and in the community. Deviations originally perceived as 'interference' phenomena may establish themselves as new sets of norms, and even replace their A equivalents. New systems are created in which conventionalized copies form integral parts. Thus, to account for conventionalization, the existence of separate Alpha grammars must be posited. Even a high-copying variety may become the norm within its group. It then constitutes a socially 'unmarked' Alpha variety and, as such, a new basic code ready to accept further copying. Its clauses serve as matrices for insertion of new copies. Thus, in principle, at every stage of development, the basic code must be defined anew; and

< previous page

page_217

next page >

< previous page

page_218

next page > Page 218

every given new Alpha norm may be deviated from by new 'marked' copying. This means that, for the analysis of current processes, knowledge of previous copying processes and its results is crucial. Influence of early Alpha As already mentioned, many phenomena observed in the speech of the first immigrant generations will certainly not be typical of possible German-Turkish varieties in the future. A certain number of global and selective copies will, however, remain. Some types of copying seem to give evidence for ongoing linguistic change. If Turkish survives in Germany and a more general Alpha lect stabilizes, it might rather bear features observed in the speech of the second generation. These features, however, partly emanate from Alpha varieties of the first generation, a kind of 'mother tongues' providing a specific input in addition to the speakers' own German. This 'double channel' input with two influence layers yields interesting results, for instance, with respect to the material shape of global copies. Some of the elements taken from the first generation may reflect previous stages of B development. The input situation contributes to the complex variational patterns and linguistic uncertainty of the second generation; but it also provides options which can be used for situational and stylistic variation. In any case, phenomena typical of the first generation clearly exert further influence even after the members of that generation have left the arena of linguistic contact. Integration In the discussion of 'integration' of code-copies, several variables have played prominent roles. In our view, the analyst should try to specify the degrees of development of a copy along these lines, register correlations between developmental stages reached along different parameters, but refrain from establishing absolute categories which conglomerate values of different variables in obscure and arbitrary ways. Parameters of structural 'integration' have proved problematic. In spite of all efforts, the boundaries between integrated and nonintegrated elements generally remain vague. The crucial difference has even been taken to lie in the speaker's intention rather than in the linguistic structure. Thus, according to Hasselmo, the stretches of speech themselves may be ambiguous, 'although the intention of the speaker may be a

< previous page

page_218

next page >

< previous page

page_219

next page > Page 219

matter of binary choice between code-switching and integration' (1970: 180). In our view, the basic problem arises from the vagueness of the general concept. E.g., 'morphological integration' mostly stands for phenomena essentially different from other kinds of integration. Thus, as already mentioned, inflectional A morphology is normally attached to global copies with their very insertion and cannot serve as a criterion of a more advanced stage of development. However, even with clearer definitional criteria for different kinds of 'integration', it seems dubious whether systematic correlations of a purely structural order can be found. High degrees of 'integration' with respect to social acceptance, frequency and nature of usage in the community have often been used as criteria for sorting out 'switches'. Other criteria are whether a given copy has replaced A elements and whether it is recognized as A (not perceived as 'foreign'). Not even such symptoms of conventionalization do always coincide with each other. And they are, of course, quite other parameters than the above-mentioned ones concerning linguistic adaptation. Neither a relatively low degree of adaptation nor a relatively low frequency of use excludes that a copy has been conventionalized and become part of a norm. The different kinds of 'integration' should be kept apart, even if some of them turn out to be empirically coincident features. The relationships between different structural stages and extra-structural factors should be studied attentively. To 'explain' copying and the development of copies it will certainly be necessary to pay attention to diverse synchronic, diachronic, socio- and psycholinguistic aspects. However, the categories defined until now on the basis of both social and structural criteria are far from discrete and possess little explanatory power. It is important not only to describe global and selective copying, but also to explore the reasons for, e.g., the actual occurrence of certain copies and the non-occurrence or low occurrence of others. The linguistic modelling briefly outlined in the present chapter only claims to offer an adequate descriptive framework. Descriptive adequacy, however, seems to be a necessary prerequisite for determining the structural as well as the non-structural conditions of code-copying. Note Thanks are due to H. Boeschoten and É.Á. Csató for highly valuable comments on an earlier version of this paper.

< previous page

page_219

next page >

< previous page

page_220

next page > Page 220

References BACKUS, A. (1989), Code-switching onder Turkse jongeren in Tilburg. Tilburg: Tilburg University (Unpublished M.A. Thesis). BOESCHOTEN, H. (1991), Asymmetrical code-switching in immigrant communities (in print). BOESCHOTEN, H. & L. Verhoeven (1987), Language-mixing in children's speech: Dutch language use in Turkish discourse. In: Language Learning 37, 191-215. BRENDEMOEN, B. (1987), Turkish in Europe: Some aspects of semantic transfer (unpublished paper). CLYNE, M. (1987), Constraints on code-switching. How universal are they? In: Linguistics 25, 739-764. CSATÓ, É. Á. (1988), Referential properties of some Turkish determiners. In: S. Koç (ed.), Studies on Turkish linguistics. Ankara: Middle East Technical 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. HASSELMO, N. (1970), Code-switching and modes of speaking. In: G. Gilbert (ed.), Texas studies in bilingualism (= Studia Linguistica Germanica 3), Berlin: de Gruyter. HASSELMO, N. (1972), Code-switching as ordered selection. In: E. Firchow et al. (eds.), Studies for Einar Haugen. The Hague: Mouton. HAUGEN, E. (1953), The Norwegian language in America: A study in bilingual behavior, 1-2. Philadelphia. HAUGEN, E. (1972), The ecology of language. Stanford. HAUGEN, E. (1973), Bilingualism, language contact, and immigrant languages in the United States: A research report. In: Th. Sebeok (ed.), Current trends in linguistics 10: 1. The Hague/ Paris: Mouton. JOHANSON, L. (1975), Some remarks on Turkic 'hypotaxis'. In: Ural-Altaische Jahrbücher 47, 104-118. JOHANSON, L. (1990), Zur Anpassung von Lehnelementen im Türkischen. Paper presented at the 2nd German Turcology Conference, Giessen 1990 (in print). KIRAL-SHAHIDI ASL, F. (1191). Persische Lehnsyntax im aserbaidschanischen Stadtdialekt van Täbriz. Mainz. MENZ, A. (1991). Studien zum Türkisch der zweiten deutschlandtürkischen Generation. Mainz. MUYSKEN, P. (1987), Neutrality in code-mixing. In: Eigen en Vreemd: Handelingen van het 39e Filologencongres Amsterdam. Amsterdam: Vrije Universiteit. PFAFF, C. (1979), Constraints on language mixing: Intrasentential code-switching and borrowing in Spanish/English. In: Language 55, 291-318. PFAFF, C. (1988), Turkish in contact with German: Language maintenance and loss among immigrant children in West Berlin. Paper presented at the Conference on Language Maintenance and Loss, Noordwijkerhout, August 28-30, 1988. POPLACK, S. (1981), Syntactic structure and social function in codeswitching. In: R. Durçn (ed.), Latino language and communicative behavior. Norwood, N. J.: Ablex Publishing Corporation.

< previous page

page_220

next page >

< previous page

page_221

next page > Page 221

POPLACK, S. & D. Sankoff (1988), A variationist approach to languages in contact. In: U. Ammon, N. Dittmar & K. Mattheier (eds.), Sociolinguistics. An international handbook of the science of language and society. Berlin: de Gruyter. POPLACK, S., S. Wheeler & A. Westwood (1987), Distinguishing language contact phenomena: Evidence from FinnishEnglish bilingualism. In: P. Lilius & M. Saari (eds.), The Nordic languages and modern linguistics 6. Proceedings of the Sixth International Conference of Nordic and General Linguistics, Helsinki 1986. Helsinki: Helsinki University Press. REICHL, K. (1983), Syntactic interference in Afghan Uzbek. In: Anthropos 78, 481-500. VERHOEVEN, L. & H. Boeschoten (1986), First language acquisition in a second language submersion environment. In: Applied Psycholinguistics 7, 241-256. WEINREICH, U. (1953), Languages in contact Findings and problems. In: Publications of the Linguistic Circle of New York 1, New York.

< previous page

page_221

next page >

< previous page

page_ii

next page >

< previous page

page_223

next page > Page 223

10 Turkish-Dutch code-switching and the Frame-Process Model Ad Backus The Netherlands is a multilingual country with several language contact situations. In virtually every contact situation that has been described in the linguistic literature code-switching occurs, so it is only natural that we would also encounter codeswitching in the bilingual communities in this country. In this chapter I will focus on the large Turkish community in the Netherlands, basing myself on data earlier presented in Backus (1989). Information on code-switching in a similar bilingual community, namely Moroccans in Holland, can be found in Nortier (1990). Being bilingual usually involves using two (or more) languages. The patterns of language alternation are very diverse, ranking on a continuum with diglossia and rapid intrasentential code-switching at its extremes. In diglossia-like situations bilinguals switch from one language to the other according to appropriate changes in the speech situation. If the speech situation does not change, they do not switch. At the other extreme of the continuum of bilingual speech we find rapid intrasentential switching. In this case switching takes place so often that we can't attribute it to changes in the context. Code-switching can apparently be the norm in the community. I will refer to this type of rapid switching henceforth as code-switching-mode. These distinctions are not new: the diglossia-type of code-switching is what Gumperz calls situational code-switching; the code-switching-mode is the prototype of his conversational code-switching (see, e.g., Penalosa, 1980). Some confusion can arise here. It is important to note that it is impossible to give a certain bilingual community a certain place on the continuum. Let us illustrate this with some data from the Turkish community in Holland. Adolescents have developed a code-switching-mode (as I will document below) which is reserved for in-group use. This would put them at the right end of the continuum. With older members of the Turkish community and with strangers, however, Turkish is used,

< previous page

page_223

next page >

< previous page

page_224

next page > Page 224

and if Dutch speakers enter the scene, a switch to Dutch is made. Now the same group would be ranked at the diglossia-end. What is of importance here is the existence of the code-switching-mode. As far as Turkish goes, the young bilinguals use two varieties: normal Turkish with their elders, and a kind of Dutch-Turkish with each other. Monolingual Turkish discourse is the norm for the community at large (i.e. when monolingual first-generation Turks are present) and the code-switching-mode is the norm for the peer-group. Below, we will focus on the code-switching-mode of bilingual Turkish adolescents. Interesting in this respect is that in an investigation of the linguistic characteristics of a very similar group, bilingual Turkish adolescents in Germany, no such mode exists (Bommes, 1989). This should make us cautious in making predictions about code-switchingpatterns on the basis of social variables alone. Research on code-switching Code-switching research is traditionally divided into three categories: sociolinguistic, psycholinguistic and theoretical linguistic research. We will argue below that integration of the three interests is badly needed. Functional and communicative aspects of code-switching have been extensively examined (see Redlinger, 1976; Scotton, 1979; for an overview: Giesbers, 1989). It was shown that code-switching is not a random phenomenon, but that it is rule-governed like any other linguistic expression. This finding led linguists to look for the syntactic contexts in which switching is possible. The research question was soon put in negative terms, leading to the formulations of constraints on code-switching. For some of these constraints, universal status was claimed, most notably for the two constraints put forward by Shana Poplack: Free Morpheme Constraint: ''Codes may be switched after any constituent in discourse provided that constituent is not a bound morpheme." (Poplack, 1980) Equivalence Constraint: "Code-switches will tend to occur at points in discourse where juxtaposition of L1 and L2 elements does not violate a syntactic rule of either language." (Poplack, 1980)

< previous page

page_224

next page >

< previous page

page_225

next page > Page 225

Both constraints have served as vantage points for a lot of subsequent studies. Sometimes they were confirmed and sometimes they were falsified (see Eliasson, 1989; Clyne, 1987 for overviews). Calls were made for modifying the constraints by incorporating typological differences between languages and situational differences between contact situations (see, e.g., Annamalai, 1989) We can illustrate this issue again with data from the Turkish/Dutch situation in Holland. The FM constraint, for instance, is typically violated in this situation, for two reasons: Turkish is clearly the main language of discourse, and Turkish is agglutinative. Thus, Turkish case markers, plural affixes and other derivative and inflectional endings are freely attached to Dutch nouns (Backus, 1989): (1) o blonde'dan alïyordum that blonde-ABL I got I took (lessons) from that blonde (girl). (2) ne yapïyor Jood'lar? what do-PRES-3 Jew-PL What are the Jews doing? (3) bir tane donkere jongen'nan a dark boy-WITH with a dark boy For evidence against the equivalence constraint in Turkish-Dutch code-switching, see Backus (1989) and Boeschoten and Verhoeven (1987). In recent papers Carol Myers Scotton has introduced a new model that attempts to account for the points in a sentence where a speaker may switch (Myers Scotton, 1990). This model, which she labels the "frame process model" is, in a further step, connected with the sociolinguistically oriented "markedness model" (Scotton, 1988), in order to present a unified account of code switching behaviour in multilingual communities. In the FP-model it is stressed that two languages in contact usually play different roles. It is a logical, but not a trivial remark. Domain separation has been documented for a lot of bilingual communities. Put differently, one language usually is dominant on a societal level, and the other one is a minority language. As a correlate of this, the roles are usually turned around in the actual speech situation: the minority language is the dominant language in conversation and the majority language is the code that is switched to (and borrowed from). This asym-

< previous page

page_225

next page >

< previous page

page_226

next page > Page 226

metry has largely been neglected in the search for universal constraints. The exceptions come mainly from studies that are based on data from languages with differing word order. Examples include Sridhar and Sridhar (1980), Nishimura (1986) and Bentahila and Davies (1983), as well as Hasselmo's early study on American Swedish (Hasselmo, 1972). The Frame-Process Model According to the Frame-Process Model three types of construction are possible: constructions that are either entirely Matrix Language or entirely Embedded Language, and mixed constituents. The ML is taken to be the main language of discourse and the EL is the intruding language. The constructions that are entirely in one language, are called islands and can be anything from a single constituent up to a big chunk of discourse. About mixed constituents further predictions are made. The model proposes that the matrix language sets the syntactic frame of a mixed utterance. This means that the ML provides the morpheme order and all productive system morphemes, or function elements. The FP-model says that mixed constituents follow two 'verifiable' principles: Morpheme Order Principle: "In mixed constituents, the morpheme order is that of the Matrix Language" System Morpheme Principle: "In mixed constituents, productive system morphemes are from the Matrix Language" The following forms are allowed by the MO principle: an order, either unmarked or marked, from the ML is followed, or bare EL forms are used. Let us look more closely at each of these in turn. The first strategy, that states that the morpheme order of the ML is followed, is illustrated by:

< previous page

page_226

next page >

< previous page

page_227

next page > Page 227

(4) Bunlarïn verjaardagen'a gitmesi güzel oluyor. them-GEN birthdays-DAT going nice is It's nice that they go to birthdays. The Dutch noun is inserted into the slot that its Turkish equivalent would have occupied in a monolingual utterance. The Turkish word order is preserved, just like the morpheme order in the mixed constituent. Dutch rules would require a preposition, but Dutch rules are irrelevant, since Turkish morpheme order is followed. The second strategy says that an order is followed that is marked, but permissible in the ML, and that happens to be the basic order in the EL. We have not encountered this strategy in our corpus, but it seems to be quite frequent in Scotton's data from East-Africa. We will come back to this strategy later. Bare forms, Scotton's third strategy, do crop up in the Turkish data: (5) Nee, bunlar herkes kendi prijs söylüyor. no this-PL every REFL price name-PRES-3 No, here everybody names his own price. If strategy I was used, the Dutch noun would have been marked with the possessive and accusative endings, but instead a form is used that is not marked for anything. The point is that Turkish morpheme order is not violated. The MO Principle is clearly reminiscent of the Equivalence Constraint. The big difference is the recognition of the different roles languages play in bilingual discourse. It is straightforward that one language exerts more influence over the other than the other way round. The influencing language will usually be the EL, since it would not crop up in ML discourse at all if it did not have any influence. For instance, Welsh borrowed a massive amount of words from English, but English did not borrow any from Welsh (see Thomas, 1984, for a few exceptions). We can assume that the starting point of the borrowing process is the emergence of the borrowed form (English) in ML (Welsh) discourse. This also explains why vocabulary from disappearing languages seldom lives on in the overtaking language. That language simply does not borrow from the dominated language, neither does it codeswitch into its direction. Grammatical properties, transmitted through convergence, seem to have a better chance for survival, in the form of an ML substratum. If we assume that code-switching plays a

< previous page

page_227

next page >

< previous page

page_228

next page > Page 228

major role in the transfer process of forms from one language to another, it is necessary to keep the results from contact in mind, as we can see them in synchronic descriptions of the syntax and lexicon of almost every language. The SM Principle also allows three strategies, the most common one of which is using ML system morphemes. Turkish-Dutch code-switching provides a huge amount of illustrations of this strategy. Consider the following example: (6) Cassetterecorder'nan friettent'e gidelim lan. taperecorder-WITH chip shop-DAT let's go hey Hey, let's go to the chip shop with the taperecorder. The content words are Dutch and the function elements (the postposition -nan and the dative case-marker) are Turkish. Object + Verb switches offer strong evidence for the existence of this strategy: (7) Bu bisürü taal'larï beheersen yapïyorken . . . he a lot language-PL-ACC know do-PROG-WHILE While he knows a lot of languages, ... (8) Türkler alsa, klant'lar wegjagen yapïyor. Turks take-IF customer-PL chase away do-PRES-3 If he would take Turks, he would chase away his customers. All function elements in the mixed constituents are Turkish, even if they have to break up the highly idiomatic Dutch object + verb combinations. The second strategy allowed by the SM Principle consists in allowing EL system morphemes as long as they are not productive. Consider, for example, the next utterance: (9) Polen'lara hollandaca ders verdi. Poles-PL-DAT Dutch lesson give-PAST-3 He taught Dutch to Poles. The plural is marked twice here, by the Dutch -en morpheme, and the Turkish -lar. Myers Scotton would contend here that -lar is productive and -en is not. This is acceptable, since all other morphemes are Turkish. The model again allows for the use of bare forms under this principle. As an example of a bare form we can cite prijs again (example

< previous page

page_228

next page >

< previous page

page_229

next page > Page 229

5). The possessive and accusative markers are not present, but there is also no Dutch possessive used. Instead of using a Dutch function element, no function element at all is used. To close this section on the SM Principle it should be remarked that it is very similar to the Free Morpheme Constraint, but that it allows the major type of counterexample to that constraint, namely ML bound morphemes attached to EL roots. Problems for the model Before the model can be seriously tested, islandhood, Matrix Language, productiveness and system morphemes need to be unambiguously defined. Before the notion of islandhood is specifically defined for a particular language pair the model runs the risk of explaining away every counterexample to the two principles as being an EL island. The only criterium required so far for islands is internal structural dependency, so single EL morphemes can never be islands. Prototypes of an island are frozen chunks, or conventionalized idiomatic expressions. This issue needs clarification before the model can be tested. Another problem that has to be solved before the model can be rigorously tested is the problem of determining the ML. It is intuitively obvious to say that the dominant language in the speech community is the ML (Turkish is a minority language in the Netherlands, but it is the dominant language in the Turkish community). It is not always easy to assign a base language to an utterance. The language of the finite verb has been suggested (Treffers, 1990) or the language of the first major constituent of the utterance (Joshi, 1985), but both methods have their drawbacks (Romaine, 1989, concludes in her review of the matter that language assignment is "essentially ad-hoc". p.137). Nortier (1990) however, provides a workable solution, based mainly on various criteria proposed in the literature, but leaving the option open that the ML sometimes simply cannot be assigned. Agreement on this matter has to be reached. It is also worthwhile to ask ourselves whether a matrix language can be assigned in a situation where there is a code-switchingmode. The unmarked language variety in in-group speech is this code-switching-mode. Taken to extremes, this means that the frame is set according to the rules of the code-switching-mode, not those of Turkish. This leads to another question: what are the characteristics of a code-switching-mode? Part of the definition is of a quantitative nature. Part

< previous page

page_229

next page >

< previous page

page_230

next page > Page 230

of the data in Backus (1989) were classified according to their language: Turkish 116 Dutch 33 Mixed 87 Quantification alone, however, does not necessarily lead to an unquestionable decision as to what the main language of discourse is. The model presupposes that we can distinguish content morphemes from system morphemes, and productive system morphemes from nonproductive ones. Surely there are prototypes of both system morphemes and content morphemes, but where do we draw the line in the grey area between them? And, if it is true that EL function words are never used, how does the model account for the borrowing of function words (like 'they' in English)? Myers Scotton acknowledges that system morphemes from the EL can occur, but that in that case they are non-productive. Crucial for this matter is the establishing of criteria for productiveness. Consider again example (9), repeated below: Polen'lara hollandaca ders verdi. We can accept that the Dutch plural marker is not meant to be productive since plurality is also marked in Turkish. But what about the next examples? (10) O normale dag boyle. that normal day such-a It was a normal day. (11) ...yani cok bakkallarï vardi, economische yükseldi. i.e. many grocers there-were economically rise-PAST-3 ... that is, they had many grocers, they had advanced economically. (12) Je moet naar een Türk evine bakacagïm. you must to a turkish house-DAT look-FUT-2sg You have to look at a Turkish house.

< previous page

page_230

next page >

< previous page

page_231

next page > Page 231

In (11) we encounter the Dutch adjective-agreement marker -e in the adjective normale. The agreement marker is a function element According to the FP-model it must be unproductive. Evidence that that is the case seems to come from the same marker in economische. Dutch rules would not allow for the -e in this position. Probably EL words are inserted in their conventionalized form. For adjectives this could mean that they are used with their agreement marker, that being their conventionalized, unmarked form. In the last example, however, it is hard to maintain that either the Dutch preposition, or the Turkish dative is an unproductive, fossilized morpheme. A strategy the MO Principle allows is using a morpheme order that is permissible but marked in the ML and unmarked in the EL. This is a sticky issue, since a lot of languages have relatively free word order. A sentence with SVO order is marked, but permissible in Turkish; so, if bilinguals would use SVO order in all their mixed utterances one could still say that the MO principle is followed, since SVO is possible in the ML. But wouldn't we rather term this a case of syntactic convergence to Dutch? In fact, Myers Scotton stresses that in Swahili-English code-switching "using the order which is marked for Swahili, but unmarked for English, is a favorite strategy" (p. 12). This is what you would expect: English material is marked, and, to foreground that, a marked sentence structure is used. The danger is apparent: the ML sets the morpheme order, but a favorite strategy is using an order that does not occur too often in the ML, but instead is highly frequent in the EL. Why not call this EL order? A last remark on the formulation of the principles is that using bare forms means that no morpheme order and no system morphemes are really used. More than towards forcing ML structure on an utterance, the principles seem to be directed towards exclusion of EL morpheme order and EL system morphemes and should, I maintain, be formulated in that vein. Theoretical issues Turning away from the actual formulation of the model we can now focus on its underlying assumptions. In proposing this model an assumption is implicit which is a little problematic: sentences are supposed to be produced in two steps: first the frame is set, then the content words are inserted. Psycholinguistic

< previous page

page_231

next page >

< previous page

page_232

next page > Page 232

evidence for this claim is virtually impossible to get, given the present state of our knowledge. Convergence poses a problem for the model. In a lot of contact situations languages tend to create equivalences in their grammar and vocabulary (usually this means that the dominated language adopts words and structures from the dominant language). In an extremely stable situation, like the one in Northern India described in Gumperz and Wilson (1971), this can lead to different languages with almost identical grammars. How can one decide whether the ML or the EL is being spoken? This reminds us of the related phenomenon of triggering. Cognates are notorious triggers for switches. It is often impossible to assign a language index to a word like that. It is useful here to stress the stand the model takes in the borrowing debate: if a form is used only once (prototypical for constituents; less so for single words) it is a code-switch. If it becomes conventionalized (prototypical for single words, especially nouns, and idiomatic phrases) it becomes a loan. Some scholars wish to classify nonce-borrowings (borrowings that are used only once) under the same heading as borrowings (e.g., Sankoff, Poplack and Vanniarajan, 1990). The underlying criterium is that borrowings are morphologically integrated into the ML, whereas a code-switch is not. For the FP-model this distinction has no importance, since the ML is supposed to build the frame for every mixed utterance (every utterance that is not an EL island), so all other EL material is naturally morphologically integrated. Conclusions and discussion One basic assumption in Scotton's work is that code-switching serves a function. Somebody who is switching codes is negotiating interpersonal relationships, just like somebody could do in a monolingual setting by switching registers. It is assumed that speakers know what they are doing: they know what the social consequences are of the way they speak. Knowledge of the social meaning of language choice motivates that very choice. Code-switching, in Scotton's and in my view, can very well be the unmarked variety of a language. This is typically the case in in-group conversation between bilingual peers who use this code-switching-mode with each other, but not in other situations (e.g., with strangers, parents

< previous page

page_232

next page >

< previous page

page_233

next page > Page 233

etc.). Within the group, though, the mode is socially accepted and conventionalized, which makes it the unmarked form of conversation (see Hasselmo, 1972, for American Swedish; Agnihotri, 1979, for British Punjabi). In our own study of DutchTurkish code-switching we found strong evidence for this mode. Myers Scotton also recognizes another form of code-switching: code-switching as a marked choice. This is in accordance with the findings of much, usually sociolinguistically oriented work on code-switching, especially the work by Gumperz. Speakers switch to attract attention, to give emphasis etc. It stands to reason that this is only possible in situations where code-switching cannot be unmarked, i.e., where there is no code-switching-mode. Thus, the bilingual peers that we discussed in the previous paragraph would have a hard time using code-switching as a marked choice when speaking to one another, but could easily do just that when speaking to their parents. Or are there ways of marking switches, even in the code-switching-mode? The F-P Model seems to suggest so. A crucial conclusion that follows from combining the sociolinguistic markedness model and the linguistic FP-model is that the principles of the FP-model can be violated in order to use code-switching as a marked choice. If a code-switching-mode that follows the two principles is the unmarked code, then deviating from the principles is used to make the utterance marked. This demands from the counterexamples that semantically they have a marked character. This means firstly that they should have a very low frequency of occurrence. A second condition, which is rather more difficult to demonstrate, is that it should be clear from the context that the switch was actually intended by the speaker to be marked. An important role is assumed here for discourse analysis. A prototypical form of code-switching as an unmarked choice is the smooth switching, that is documented for a number of bilingual situations. The model predicts that normally code-switching follows the two principles, but that it does not have to when a particular switch is intended as marked. That switch might, or might not, also be flagged, depending on the conventions that have been established in the particular situation. In a lot of bilingual communities though, code-switching is marked. The normal way to speak is monolingual, where the used code can differ according to variables of the speech situation. Intersections of the markedness model and the FP-model would predict here that a code-

< previous page

page_233

next page >

< previous page

page_234

next page > Page 234

switch does not follow the two principles, or in any case that it does not have to. Clearly this is not a very desirable situation. Firstly, the FP-model would have to be restricted to those situations where there is a code-switching-mode. Secondly, judging from the support the two principles get from data all over the world they can explain much more than that. References AGNIHOTRI, R. (1979), Processes of assimilation: A socioinguistic study of Sikh children in Leeds. York: York University (Unpublished PhD. Thesis). ANNAMALAI, E. (1989), The language factor in code-mixing, In: International Journal of the Sociology of Language 75, 4754. BACKUS, A. (1989), Codeswitching onder Turkse jongeren in Tilburg. Tilburg: Tilburg University (Unpublished MA Thesis). BENTAHILA, A. & E. Davies (1983), The syntax of Arabic-French code-switching. In: Lingua 59, 301-330. BOESCHOTEN, H. & L. Verhoeven (1987), Language mixing in children's speech: Dutch language use in Turkish discourse. In: Language Learning 37, 191-215. BOMMES, M. (1989), Sprachliche Verarbeitung der Migrationssituation bei Migrantenjugendliche ttrkischer Herkunft Osnabrück: Osnabrück University (Unpublished PhD. Thesis). CLYNE, M. (1987), Constraints on code-switching: How universal are they? In: Linguistics 25, 739-764. ELIASSON, S. (1989), English-Maori language contact: Code switching and the Free Morpheme Constraint. Reports from Uppsala University Department of Linguistics 18, 1-28. GIESBERS, H. (1989), Code-switching tussen dialect en standaardtaal. Amsterdam: P.J. Meertensinstituut. GUMPERZ, J. & R. Wilson (1971). Convergence and creolization: A case from the IndoAryan/Dravidian border. In: D. Hymes (ed.), Pidginization and creolization of languages, Cambridge: Cambridge University Press, 151-167. HASSELMO, N. (1972), Code-switching as ordered selection. In: D. Pinchow et al. (eds.), Studies for Einar Haugen. Den Haag: Mouton, 261-280. JOSHI, A. (1985), Processes of sentences with intrasentential code switching. In: L. Karttunen, D. Dowty & A. Zwicky (eds.), Natural language parsing: Psychological, computational and theoretical perspectives, Cambridge: Cambridge University Press, 190-205. MYERS SCOTTON, C. (1990), Intersections between social motivations and structural processing in codeswitching. Paper read at the Workshop on Constraints, conditions and models, ESF Scientific Network, London. NISHIMURA, J. (1986), Intrasentential code-switching: The case of language assignment. In: J. Vaid, (ed.), Language processing in bilinguals: Psycholinguistic and neuropsychological perspectives. New Jersey: Hillsdale, 123-143.

< previous page

page_234

next page >

< previous page

page_235

next page > Page 235

NORTIER, J. (1990), Dutch and Moroccan Arabic in contact: Code switching among Moroccans in the Netherlands. Dordrecht: Foris. PENALOSA, F. (1980), Chicano Sociolinguistics. A brief introduction. Rowley, Mass.: Newbury House. POPLACK, S. (1980), Sometimes I'll start a sentence in Spanish y termino en Español. In: Linguistics 18, 581-618. REDLINGER, W. (1976), A description of transference and code-switching in Mexican-American English and Spanish. In: G. Keller et al. (eds.), Bilingualism in the bicentennial and beyond, New York: Editorial Bilingual, 41-52. ROMAINE, S. (1989), Bilingualism. Oxford: Blackwell. SANKOFF, D., S. Poplack & S. Vanniarajan (1990), The case of the nonce-loan in Tamil. In: Language variation and change 2 (1), 71-101. SCOTTON, C. (1979), Strategies of neutrality: Language choice in uncertain situations. In: W. McCormack & S. Wurm, Language and society: Anthropological issues, Den Haag: Mouton, 71-87. SCOTTON, C. (1988), Codeswitching as indexical of social relationships. In: M. Heller (ed.), Codeswitching: Anthropological and sociolinguistic perspectives, Berlin: Mouton de Gruyter, 151-186. SRIDHAR, S. & K. Sridhar (1980), The syntax and psycholinguistics of bilingual code-mixing. In: Canadian Journal of Psychology 34 (4), 407-416. THOMAS, A. (1984), Welsh English. In: P. Trudgill (ed.), Language in the British Isles, Cambridge: Cambridge University Press, 178-194. TREFFERS, J. (1990), Towards a uniform approach to code-switching and borrowing. Paper read at the Workshop on Constraints, conditions and models. London: ESF Scientific Network.

< previous page

page_235

next page >

< previous page

page_ii

next page >

< previous page

page_237

next page > Page 237

11 Code-switching and borrowing in an Arabic-Dutch context Jacomine Nortier In the literature over the last decades, several definitions and characteristics of borrowing as opposed to code-switching have been developed. Muysken (1990) gave a summary of definitions and characteristics that are currently used. According to Muysken a borrowing can be considered if only one word is involved which is adapted grammatically, used frequently, accepted as an own word that has replaced another word, leading to a semantic change. The first characteristic ''no more than one word" should perhaps be expanded to "no more than one word or short idiomatic expression". An example of the latter is faux pas or savoir faire (examples from Bentahila and Davies, 1983). Adaptation on three grammatical levels is illustrated in the possible Dutch sentence (1): (1) Het jeugdelftal had een uitzonderlijk goed keepertje. The youth team had an extraordinary good keeper (small). In (1), keepertje is pronounced according to Dutch rules and adapted to Dutch morphology (the diminutive -tje). As a consequence, the adjective goed is used. Neuter and diminutive nouns in Dutch are accompanied by adjectives with a zeroending. Keeper is frequently used in Dutch, at least several times in every soccer report, article or conversation. In the interpretation of Heath (1990), frequency of usage is not easy to determine since it depends on different individuals. According to him, an element can be code-switching on formal grounds, and at the same time an instance of borrowing on functional grounds. Heath proposes, therefore, to see the difference between code-switching and borrowing as a gradual one. However, the noun keeper does not raise any problems. I am not sure whether it replaces a Dutch word (perhaps doelman), but it is certainly recognized as an own word by Dutch

< previous page

page_237

next page >

< previous page

page_238

next page > Page 238

people who speak English as well as people who don't In English the word keeper may have other meanings beside a person who defends a goal in soccer, but in the Netherlands it is the only possible interpretation. "Semantic change" in this case means "semantic narrowing". Although the characteristic ("replaces own word'') is questionable, it can be said that keeper is a clear example of borrowing from English into Dutch. In this chapter the status of Dutch single word switches in the Moroccan-Arabic language of Moroccan adolescents living in the Netherlands will be discussed. One of the questions is whether it is appropriate to consider single word switches as a separate category rather than a particular kind of intrasentential switching. Another topic directly linked to this matter is the case of borrowing: Can the single word switches be interpreted as borrowings? It will appear that this is not the case. I will also point out that solutions like the notions of nonce borrowing and constituent insertion cannot be used in the Dutch/Moroccan Arabic context. Some seemingly logical explanations will be discussed finally. Code-switching among Moroccans in the Netherlands From 1985 to 1989 the project 'Code-switching among Moroccans in the Netherlands' was carried out at the Linguistics Department of the University of Amsterdam. The report, originally a doctoral dissertation, is published in 1990 (Nortier, 1989, 1990). In this study nine informal conversations between Dutch/Moroccan Arabic bilingual Moroccan men and women (average age about 23) were analyzed. Fifteen informants were involved. Berber speaking Moroccans were excluded. The conversations were taped by a bilingual Moroccan research assistant. The informants were selected among the assistant's friends and family and further along (their) personal networks and a sports school in Utrecht. Places where the conversations were held were chosen in order to guarantee informality, such as cafés, a shopping centre and the assistant's living room. It was expected that these settings would constitute optimal conditions for codeswitching. Initially bilingual proficiency was assessed intuitively. At a later stage a post hoc ordering from Dutch dominant via balanced bilingual to Moroccan Arabic dominant was made on the basis of four measures.

< previous page

page_238

next page >

< previous page

page_239

next page > Page 239

In sum, about one thousand code-switches have been found and analyzed. These switches have been classified as intrasentential (within sentence boundaries), intersentential (between sentences) and extrasentential (term used by Boucherit, 1987, and Scotton, 1987, referring to words which are part of the discourse structure, but outside the sentence's syntactic structure). One third consisted of inter- and extrasentential switches (examples (2) and (3), respectively), while two thirds of all switches were intrasentential, as in (4). (2) Iedereen zit te kijken / ta-ygulu l-ek: bada ma-l-u? Everybody is watching. / They-say to-you: What's wrong with him? (moha096) (3) Maar de tijd die gaat toch voorbij, / fbemti? But the time it goes still by, / you-understand? (dria379) (4) hna ka-yseknu Here they-are-living (kalb056)

/ zowel Marokkanen als Nederlanders. / both Moroccans and Dutch.

Among 677 intrasentential switches 402 were single words, as in (5): (5) u hada huwa l/ voordeel / lli . . . and that it the- / advantage / which . . . 'and that is the advantage which ... (nadb422) Single word switches Of 402 switched single words 371 were Dutch in Moroccan Arabic sentences and only 31 were Moroccan Arabic in Dutch sentences. Table 1 gives more details (see Nortier, 1990). Most single Dutch words were nouns (285). Without denying the importance of analyzing all single word switches, I will concentrate here on the Dutch nouns in Moroccan Arabic sentences because of their numerical preponderance.

< previous page

page_239

next page >

< previous page

page_240

next page > Page 240

Table 1 Word classes to which single words belong

Nouns Adverbs Adjectives Conjunctions Verbs Prepositions Pronouns Numerals Total

Total abs in % 286 71.1 41 10.2 26 6.5 20 5.0 11 2.7 8 2.0 6 1.5 4 1.0 402

100

Du in MA MA in Du sentence sentence abs in % abs in % 285 99.7 1 0.3 29 70.7 12 29.3 26 100 0 0 6 30.0 14 70.0 11 100 0 0 5 62.5 3 37.5 5 83.3 1 16.7 4 100 0 0 371

92.3

31

7.7

In Nortier and Schatz (1988) it was shown that in borrowing mostly nouns are involved. Monolingual use of the native language, in which frequent borrowing occurs, shows an average of almost 19% nouns. Of all borrowings, 57% are nouns. In the Dutch/Moroccan Arabic data (the same as in the present study) the percentage of nouns in monolingual use was 13, whereas 77.4% of the borrowings consisted of nouns. The Dutch/Moroccan Arabic single word switches were treated as borrowings in Nortier and Schatz (1988). The other language pairs under investigation were French/English in Canada; Flemish/ French in Brussels; Quechua/Spanish in the Andes mountains, and English/Dutch in the United States. In the distribution among word classes and semantic fields the single words in the Dutch/Moroccan Arabic material did not differ from the 'true' borrowings in the other language pairs. Also in accordance with the other language pairs is the fact that Dutch elements occur in Moroccan Arabic sentences, rather than the reverse. The language that has the most impact on the other, i.e., the majority language, the language of the colonists or of the host society, is the one from which elements are borrowed. Incorporation of Dutch nouns Now I will return to the Dutch single nouns in Moroccan Arabic sentences and illustrate why those nouns cannot be considered borrowings. It would be easy if none of the criteria mentioned by Muysken would

< previous page

page_240

next page >

< previous page

page_241

next page > Page 241

apply. Then I could state, without further discussion, that all single Dutch nouns in Moroccan Arabic sentences are in fact codeswitches and borrowing is out of the question. But, as it is, some criteria are met while others are not. The first criterion, of course, is met because I only discuss single nouns. Serious problems arise when adaptation is concerned. Phonological adaptation is dubious since - in my data - none of the Dutch single nouns is pronounced identical to Moroccan Arabic rules. In spreekuur (example (328) in Nortier, 1990) a cluster of three consonants, among which -p-, is heard, although Moroccan Arabic doesn't have clusters longer than two (different) consonants and -p- is nonexistent. The absence of morphological and syntactic adaptation is more difficult to demonstrate. When a plural ending is required (45 times in the Dutch/Moroccan Arabic material), it is either not realized (eight times) as in (6), or it is expressed in Dutch (37 times) as in (7): (6) ka-ncemlu Gir / gebouw / kbaR we-are-doing only / building (sg.) / large (pl.) 'We only do large buildings' (the speaker is a heating engineer) (ahma125) (7) ka-tetkellemcla / buitenlanders You-are-talking about / foreigners (kala 168) Dutch nouns with Moroccan Arabic plural endings did not occur. Whether or not an item is morphologically and syntactically integrated is not always visible. For example, a singular noun is neither marked for number in Dutch nor in Moroccan Arabic. The use of a Moroccan Arabic plural ending or a Moroccan Arabic plural ending or a Moroccan Arabic diminutive with an originally Dutch noun would be an indication of morphological integration in Moroccan Arabic. However, in the four cases where a diminutive was used it was always in Dutch, and in two out of the four cases it was combined with a Dutch plural marker, as in (8): (8) Hetta / liedjes also / songs (nadb289) A correct Dutch plural marker was used 37 times, as in (7) above. This group makes up the majority of single Dutch words which apparently

< previous page

page_241

next page >

< previous page

page_242

next page > Page 242

are not integrated in Moroccan Arabic morphologically. A plural was required in 45 cases. Eight times (i.e., 17.8%) a singular form was used where a plural would have been expected on the basis of the context, as in (6) above. Nait M'Barek and Sankoff (1988) have stated that when a noun takes the article of the host language, it is seen as an argument of syntactic integration. According to the theory of borrowing (in the vision of Nait M'Barek and Sankoff), if those nouns were borrowings, they would behave as simple nouns in Arabic and they would not demonstrate French flection or syntax. (They investigated French/Arabic codeswitching among Moroccan students in Canada.) In my material, the Dutch nouns often seem to lack definite articles, either in Dutch or in Moroccan Arabic. One of the striking findings is that only once the inserted Dutch noun has a Dutch definite article. Out of 285 contexts where a Dutch noun is inserted, 88 times an article is not required, 53 times Moroccan Arabic definite articles are used, and 82 times an obligatory definite article is not realized at all. In the remaining cases, an indefinite article is required (and often not used), which will not be discussed here. Details are given elsewhere (Nortier, 1990). In sum, I have showed here that Dutch nouns in Moroccan Arabic sentences are not fully integrated grammatically. On the one hand, morphological integration seems impossible because of the apparent lack of Moroccan Arabic plural endings and diminutives, while on the other hand morpho-syntactic integration could be concluded from the (almost) exclusive use of Moroccan Arabic definite articles. The picture is disturbed by the fact that rather categorically compulsory articles are not used at all. Syntactic simplification should be rejected as an explanation for the frequent deletion of definite articles. Other characteristics of simplification, such as the omission of plural markers or prepositions, only occur sporadically in my data. The criterion "frequency of use", could not be investigated in a relatively small sample. The criterion "replaces own word", however, was investigated. It appeared that for one fifth of the nouns involved a Moroccan Arabic equivalent was used elsewhere in the material. An analysis with respect to semantic fields to which single switched nouns and nouns in longer switched stretches belong yielded a slight preponderance for the single nouns to belong to typical Dutch semantic fields. An example is the field of social and professional position.

< previous page

page_242

next page >

< previous page

page_243

next page > Page 243

With respect to the latter criterion it should be concluded that single nouns functionally show a greater tendency towards borrowing than nouns in multiword switching. A stronger statement cannot be made. After the data collection, informants were not interviewed again, which is the reason why nothing is known about the acceptance as own (Moroccan Arabic) words of the nouns under consideration. Matters concerning semantic change were not investigated. Although some criteria concerning borrowings have not been studied, it is obvious from the discussion above that Dutch single nouns in Moroccan Arabic sentences cannot simply be interpreted as established borrowings or as intrasentential code-switches. Formal integration was incomplete and functionally some nouns behaved as borrowings, while others did not. Nonce borrowing Sankoff and Poplack, in several publications (1985, among others), have developed the notion of nonce borrowing in order to classify in-between-cases that were neither established borrowings nor instances of intrasentential code-switching. In 1953, Weinreich had already introduced the term nonce borrowing (Weinreich, 1953: 11). The current definition is as follows: " ... it differs from borrowing in the traditional usage mainly in the social sense - nonce borrowings are not necessarily recurrent or widely recognized in the community as loanwords. They are, however, morphologically and syntactically incorporated into the host language. The necessity of phonological adoption is not clear at this point and may differ from community to community and individual to individual Thus nonce and established borrowing are two extremes of a continuum in which the distinguishing parameter is not linguistic but social." (Sankoff and Poplack, 1985: 12) However, as was demonstrated in the previous section, the Dutch single nouns in Moroccan Arabic sentences were not only difficult to classify with respect to their use or social function. Problems arose in the formal sense in particular. Nonce borrowing is not a solution because it presumes at least some grammatical integration which is absent in the Dutch/Moroccan Arabic data.

< previous page

page_243

next page >

< previous page

page_244

next page > Page 244

Constituent insertion Nait M'Barek and Sankoff (1988) found many instances of French nouns with a French definite article in Arabic sentences. The occurrence of an article in the language of the noun (French) made it undesirable to classify that NP as a (nonce) borrowing, since the latter is defined as a morphologically integrated element Therefore, they used a new category, constituent insertion which is defined as: " . . . insertion of a French NP, including at least article and noun (both inflected for person, number and gender) and optionally other elements in a syntactic slot for an Arabic NP." (Sankoff and Poplack, 1988: 5) There are several similarities between the French constituent insertions in Nait M'Barek and Sankoffs material and the Dutch/Moroccan Arabic corpus. In the first place, Moroccan Arabic is the language in which the greatest number of single words from the other language (French or Dutch) are inserted. Furthermore, most single switched words are nouns. Finally, the inserted nouns function chiefly as direct object or prepositional complement, in both corpora. (9) is an example where the inserted Dutch noun is a direct object: (9) daba cTaw-ni / ontslag Now they-gave-me / notice 'Now they fired me.' (nadb277) In (7) above, the inserted Dutch noun is a prepositional complement. Unfortunately, the differences between the constituent insertions and my single switched nouns cannot be overlooked: In the French/Arabic data, the French nouns were usually (91%) accompanied by a French article. The remaining 9% were interpreted as (nonce) borrowings. I have showed earlier that it was exactly the group of nouns without a definite article (in either language) that caused classification problems. Adopting the notion of constituent insertion would solve the difficulties for the Dutch nouns in Moroccan Arabic sentences in only one out of 285 cases.

< previous page

page_244

next page >

< previous page

page_245

next page > Page 245

Asymmetry and the suspension of grammar Boeschoten (1990) compared the results of my investigation to those by Backus (1989) on Turkish/Dutch code-switching. At the end of his article, Boeschoten states, as a conclusion, that "... typological closeness between the minority language and the dominant language leads to rapid one-sided convergence (Clyne, 1987), and that typological distance leads to a type of intra-sentential code-switching in which L2-grammar is suspended mostly in L1-structure." (Boeschoten, 1990: 15) Particularly in the Turkish/Dutch material, Boeschoten found a certain "telegraphic (in the sense of 'reductionist as to function elements') sloppiness" (1990: 9). With respect to switches where syntactic cohesion is absent, Muysken (1990) applies the term "suspension of grammar". For these phenomena, which are interrelated (if not the same), Boeschoten gives three reasons: (1) The switches reflect previous stage of L2 development of the informants and of others. (2) Switches serve to avoid problems with surface structure equivalence. (3) The code-switching mode forms part of the Turkish repertoire. In his book on borrowing in Moroccan Arabic, Heath (1990) discusses in short the Algerian/French language contact situation. Perhaps his remarks can be seen as a justification for (1) above: borrowings from French into Moroccan Arabic usually have a Moroccan Arabic article and pronunciation. In Algerian Arabic, borrowings from French are not integrated. The nouns keep their original (French) article and pronunciation. Heath explains this by stating that Algerians usually have a better command of French than Moroccans. Although Boeschoten, in his 'three reasons' referred to Turkish/Dutch code-switching, I think it is possible to translate Heath's remarks to the Dutch/Moroccan Arabic situation and link telegraphic switching to article deletion in my data. Applying Heath's remarks would mean, e.g., that Moroccans with a good command of Dutch use Dutch nouns with Dutch articles and plural endings in otherwise Moroccan Arabic sentences, because the borrowings would not be integrated. In fact, all speakers in my sample had a good command of Dutch and Dutch nouns never had Moroccan

< previous page

page_245

next page >

< previous page

page_246

next page > Page 246

Arabic plural endings. So far, what Heath stated seems plausible. However, the definite article is only once realized in Dutch. Preferably, they are either not expressed or realized in Moroccan Arabic. With respect to reason (2) above, Boeschoten makes the following remark: In Dutch/Moroccan Arabic code-switching the syntactic integrity is upheld to a much greater extent, which is caused by the fact that Dutch and Moroccan Arabic are more equivalent than Turkish and Dutch. I would like to suggest, therefore, that in future research more attention should be paid to non-equivalence sites between Dutch and Moroccan Arabic. If necessary, elicitation and judgement procedures should be used beside spontaneous speech. I am sure that in the analysis asymmetry and 'telegraphic sloppiness' should be involved to gain insight in what lies behind the facts. This does not mean that a grammatical analysis is useless. On the contrary: it is at least necessary to understand where the problems begin. Diachrony Linguists seem to be convinced that the factor time plays an important role in explaining and interpreting language contact phenomena. Boeschoten suggests that we consider ". . . at least certain types of ragged [i.e. non-constituent, JMN] switching as evidence for ongoing linguistic change." (1990: 2) Heath (1990) considers frequent code-switching as a necessary stage for conversion patterns to develop. Once the patterns are established, frequent borrowing can occur. In Nortier (1990) it was argued that the nouns without obligatory articles possibly reflect an intermediate stage between code-switching (with Dutch characteristics) and lexical borrowing (with Moroccan Arabic morphology) Diverse as they are, the three visions above have in common that they consider time a crucial factor. The most logical way to investigate the justification of this statement is to study a new corpus of code-switching material which will be collected in the near future. When in future research speakers, their backgrounds and settings are comparable to those in the material analyzed now, differences between the corpora can be ascribed to diachronical influence (among others, of course).

< previous page

page_246

next page >

< previous page

page_247

next page > Page 247

Conclusions and discussion The above explanations cover only part of the problematic Dutch single noun switches in Moroccan Arabic sentences. Although an absolute solution seems far away, I think that future research will shed more light on the phenomenon. I assume that time plays a crucial role in language contact. Therefore, a code-switching corpus which will be collected five to eight years after the one this article is concerned with, should at least demonstrate whether or not a development is going on and if so, in what direction. I hope to find indications like a changing pattern of article use and morphological integration of Dutch elements on other levels, such as the use of Moroccan Arabic plural endings. Finally, I would like to stress the importance of considering code-switching not as an independent entity. In the interuniversitary project that recently started, Turkish/Dutch and Moroccan Arabic/Dutch code-switching will be studied in close connection with bilingualism in children and language loss and thus receive the treatment it deserves. Acknowledgements This research was supported by the Foundation for Linguistic Research, which is funded by the Netherlands Organization for Research (NWO). I would like to thank Kees Versteegh for comments on an earlier version (and Pim van Straaten for the work she did on typing this article). References BACKUS, A. (1989), Code-switching onder Turkse jongeren in Tilburg. Tilburg: Tilburg University (MA Thesis). BENTAHILA, A. & E. Davies (1983), The syntax of Arabic-French code-switching. In: Lingua 59, 301-330. BOESCHOTEN, H. (1990), Asymmetrical code-switching in immigrant communities, Paper held at the ESF workshop on codeswitching in London. CLYNE, M. (1987), Constraints on code-switching: How universal are they? In: Linguistics 25, 739-764. HEATH, J. (1990), From code-switching to borrowing: A case study of Moroccan Arabic. London: Kegan Paul. MUYSKEN, P. (1990), Concepts, methodology and data in language contact research. Papers for the Workshop on Concepts, methodology and data. ESF, 15-30.

< previous page

page_247

next page >

< previous page

page_248

next page > Page 248

NAIT M'BAREK, M. & D. Sankoff (1988), Le discours mixte Arabe/Français: Emprunts ou alternances de langue? In: Canadian Journal of Linguistics/Revue Canadienne de Linguistique 33(2), 143-154. NORTIER, J. (1989), Dutch and Moroccan Arabic in contact: Code-switching among Moroccans in the Netherlands. Amsterdam: University of Amsterdam (PhD. Thesis). NORTIER J. (1990), Dutch-Moroccan Arabic code switching among Moroccans in the Netherlands. Dordrecht: Foris Publications. NORTIER, J. & H. Schatz (1988), Van éénwoordwisseling naar ontlening, een vergelijkend onderzoek. In: Forum der Letteren 29(3), 161-178. SANKOFF, D. & S. Poplack (1985), Adposition and article in Finnish-English code-switching. Montréal (ms). SANKOFF, D. & S. Poplack (1988), A variationist approach to languages in contact. Updated version of 'Code-switching' (S. Poplack & D. Sankoff). In: U. Ammon, N. Dittmar & K. Mattheier (eds.), Sociolinguistics. An international handbook of the science of language and society. Berlin: Walter de Gruyter. WEINREICH, U. (1953), Languages in contact. Den Haag: Mouton.

< previous page

page_248

next page >

< previous page

page_249

next page > Page 249

12 Finns and Americans in Sweden: Patterns of linguistic incorporation from Swedish Paula Andersson This chapter examines language contact phenomena, often referred to by a dichotomy of code switching and borrowing, in two language pairs. Finnish and American English are currently being spoken as immigrant minority languages in Sweden. In this study, the patterns of incorporation of Swedish lexical items into Finnish and American English are compared. Instead of treating code switching and borrowing as clearly different phenomena, they are seen as end-points on an integration continuum. Differences in the patterns of incorporation of Swedish items have been found between Finns and Americans. Finns tend to integrate the Swedish items to different degrees, like borrowings, while Americans tend to use the Swedish items in a nonintegrated shape, like single-word code-switches. Factors like the structure of the languages, the history of language contact, the social context of immigration and consequent degree of linguistic neutrality on different language levels, are suggested to explain the different patterns of incorporation found between Finns and Americans. Code-switches and borrowings Many of the researchers working on language contact phenomena agree upon a theoretical differentiation between codeswitching and borrowing. Code-switching would be the alternative use of two languages and two grammatical systems while borrowing would involve the grammatical system of one language only and non-native items adapted to that system. In practice, however, there is no agreement upon reliable criteria for differentiating code-switches from borrowings. This disagreement can be illustrated by the different views presented in Clyne (1987),

< previous page

page_249

next page >

< previous page

page_250

next page > Page 250

Sankoff, Poplack & Vanniarajan (1986), Poplack, Wheeler & Westwood (1987) and Scotton (1987). Clyne contrasts code-switching with lexical transference. A lexical transference is a single item, whether integrated into the grammatical and/or phonological system of the recipient language or not. Poplack and her co-authors acknowledge that particularly single-word incorporations are often ambiguous as to their status as code-switches or borrowings. Denying the usefulness of phonological criteria, they develop specific morphological and syntactic criteria in order to be able to separate single-word code-switches from borrowings. Established borrowings and so- called nonce borrowings are to be seen basically as the same process, while code-switching as a process of incorporation is totally different. The difference between these two views concerns the status of single-word incorporations: Clyne excludes all of them from code-switches, while Poplack et al. categorize, on morphological and syntactic grounds, some of them as code-switches and some of them as borrowings. Scotton, representing a third standpoint, claims that criteria relying on phonological, morphological or syntactic integration are not reliable to distinguish borrowings. Instead, borrowed forms should be distinguishable by their frequencies. Borrowings belong to a specifiable set from the embedded language and they are available to many speakers, while embedded forms in switching are not. Appel and Muysken (1987), among others, conclude that the integration of borrowed elements is a very gradual process and that the degree of integration is generally indicative of how long the borrowed item has been used within the speech community. On the basis of simple diagnostic criteria, it is not possible to distinguish cases of code-switching from not-yet-integrated borrowings. Also in relation to the current data from Finnish-Swedish and English-Swedish bilingualism, it can be claimed that codeswitching and borrowing are not two clearly different phenomena. The incorporation of specific items is a variable phenomenon, both in the speech of bilinguals and from the point of view of the speech community. Instead of as a dichotomy, typical codeswitches, i.e., phonologically and morphosyntactically non-integrated multi-word incorporations, are treated as one end-point, and typical borrowings, i.e., phonologically and morphosyntactically integrated single-word incorporations, as the other endpoint on the same continuum (cf. Boyd, 1990). Incorporations showing

< previous page

page_250

next page >

< previous page

page_251

next page > Page 251

different degrees of integration are to be found between these two ideal types. Linguistic neutrality and structural comparison of languages Studies on code-switching have mainly been carried out using data from typologically similar Indo-European languages. Some of these studies have suggested universal principles of code-switching, e.g., the Equivalence constraint and Free morpheme constraint formulated by Poplack (1980), and the Government constraint by Di Sciullo, Muysken & Singh (1986). Although many counter-examples to the universal principles have been reported from different language pairs and multilingual communities (some of them discussed in, e.g., Scotton, 1987), the suggested principles still seem to be relevant, if not as rules, at least as strong tendencies constraining code-switching. Therefore, as Appel and Muysken (1987) point out, there is a need to search for relativized constraints. These constraints are intended to be able to take into account how these universal principles of code-switching interact with constraints particular to different types of language-pairs involved and different types of bilingual communities. These relativized constraints underlie the concept of linguistic neutrality. The term linguistic neutrality, as it is presented by Appel and Muysken, refers to the points of similarity between the structures of the two languages involved in code-switching. The universal principles of code-switching mentioned above are special cases of neutrality, claimed to allow for intra-sentential switching. The Equivalence constraint involves linear neutrality, i.e., the word order is parallel in the two languages around the switch point. The Government constraint involves grammatical independence, i.e., ungoverned elements such as tags, exclamations, interjections and adverbs can easily be switched. Other types of linguistic neutrality mentioned by Appel and Muysken include neutrality achieved by words which are phonetically identical or very similar in both languages, and neutrality achieved by morphological means. Case suffixes for nouns, and verbs such as 'make' or 'do', which neutralize a non-native verb by forming a complex with it, are examples of morphological neutralizers. It seems to me that the concept of linguistic neutrality, as it is presented by Appel and Muysken, can be understood from two perspec-

< previous page

page_251

next page >

< previous page

page_252

next page > Page 252

tives, which are not clearly distinguished. On one hand, neutrality refers to similarities in the structures of the two languages. In that case, it would facilitate possible code-switching. On the other hand, neutrality seems to refer to the result of using auxiliary strategies when incorporating foreign items, when there is no overlap (or neutrality in the first sense) between the structures of languages, as in the case of morphological neutralizers mentioned above. Neutrality in the first sense is cross-linguistic, and neutrality in the latter sense is intra-linguistic, i.e., neutrality achieved in relation to the norms of the recipient language. For the purpose of the present study, I will adopt the concept of neutrality in the sense that it refers to similarities in two languages on various linguistic levels, e.g., morphology, phonology or lexicon. The neutral areas between languages would make the language boundaries more fuzzy and therefore easier for the bilingual speaker to cross. From the point of view of a bilingual speaker, these similarities would enable incorporations from one language into the other, including possibly both codeswitching type and borrowing type incorporations and others like loan translations. A narrow area of neutrality on one language level could be overcome by using the possibilities provided on another level where the neutral area is larger. The dilemma for the bilingual is that when strong negative attitudes are associated with language mixing, which is considered to be a sign of decay, at the same time inexhaustible linguistic resources which would fit the bilingual's needs are provided on the forbidden side of the language boundary. One way out is to utilize the neutral areas between the languages. In this study, three different languages and two language pairs are involved. Finnish is a suffixal language and typologically distant from Swedish, while English and Swedish are closely related, relatively analytic languages. Typological comparison would predict narrow areas of neutrality for the Finnish-Swedish language pair, and large areas of neutrality for the EnglishSwedish language pair. This would imply that incorporations from Swedish into English would be easy and frequent, while incorporations from Swedish into Finnish would cause problems and would not be so frequent. We shall see below that the picture is far more complex than a general typological comparison might suggest.

< previous page

page_252

next page >

< previous page

page_253

next page > Page 253

Finns and Americans in Göteborg The Finns and the Americans are, as minority groups in Sweden and in Göteborg, very different in many respects. Göteborg as a whole has about half a million inhabitants. The Finns are the largest immigrant group in Göteborg (approximately 10 000 Finnish citizens and at least as many Swedish citizens with a Finnish background). Most of the first generation immigrants came to Sweden in the late 1960s or early 1970s. Volvo has steadily offered, up to now, jobs for the Finns in search of work in industry. Many of the Finns in Sweden tend to have their social networks largely within the Finnish speaking community. Even many of those who are skilled in Swedish primarily spend their leisure time in activities related to the community of Finns living in Sweden. Alongside the informal social networks, a network of formal organizations and institutions has emerged. During the late 1970s and early 1980s, the community of Finnish immigrants in Sweden has gradually acquired the characteristics of a permanent ethnic minority (Jaakkola, 1984). The American group in Göteborg is small (approximately 600 USA-citizens and as many Swedish citizens with American background) and lives scattered in different parts of the city. The occupation of the Americans is often middle-class, and many of them have language-related jobs, e.g., as teachers. A large majority of those living in Sweden for a longer period of time are married to Swedes, and they are usually well-integrated into Swedish life but have varying degrees of contact with other Americans (Boyd, 1990). The differences between these two groups as regards the social determinants of immigration, such as the number of group members, the length of stay in the host country, the social network structure or ethnic mobilization, can be expected to be decisive, when comparing the patterns of incorporations. The possible links between these factors and the idea of linguistic neutrality will be discussed below. Data and principles of analysis The data of incorporations of Swedish-origin items into Finnish and American English comes from audio-recorded interviews with 18 first generation immigrants in the Finnish group and 12 in the American group. The incorporations from these two groups will be compared in

< previous page

page_253

next page >

< previous page

page_254

next page > Page 254

relation to 1) word class and phrase type categorization and 2) morphosyntactic integration. Typical code switches, multi-word incorporations showing Swedish phonology, morphology and syntax, are considered as one instance of incorporation, as well as single-word incorporations, integrated into Finnish and English phonology and morphology. These two types of incorporations are thought to be the end-points on a continuum from non-integrated items to fully nativized ones. Examples (1a) and (1b) show what has been categorized as non-integrated items, like code-switches, and example (2a) shows an example which is morpho-syntactically integrated: (la) I started going to SVENSKA FÖR UTLÄNNINGAR and 'Swedish for foreigners' (1b) This sort of VISADE SEJ / showed // out to be very easy 'turned out' (2a) They were simply foreign lecturer TJÄNSTs 'positions' The word class and phrase type categorization provides an overall characterization of the incorporations used by Finns and Americans. When searching for possible explanations for the differences found between the two groups, the concept of linguistic neutrality will be discussed. The description of patterns of morpho-syntactic integration of the incorporations is the other main focus of the present study. That description ends up with a suggestion for a cross-linguistic comparison of the incorporation patterns, the concept of linguistic neutrality being a central issue again. Phonological integration of the incorporations has not been taken into account within this study. Phonology is not accepted by all the researchers as a reliable criterion for the degree of integration of nonnative items. Poplack (1987), for instance, rejects phonological criteria by pointing out that there is individual variability and/or similarity in the phonological systems of a bilingual's two speech varieties. In relation to the present data, though, phonological integration seems to be a central factor, interacting with morphological integration. The concept of neutrality could very well be applied to phonology, when the interaction between phonological and morphological integration of the incorporations is described. Nevertheless, as a first step, morpho-syntactic integration of the incorporations has been studied independently. I focus the discussion on the Finnish-Swedish language pair. The figures concerning the English data are drawn from Boyd (1990), and

< previous page

page_254

next page >

< previous page

page_255

next page > Page 255

more details concerning the incorporations in the English data can be found in that report. Swedish incorporations into Finnish and English 614 instances of incorporation in the Finnish data and 207 in the English data were found in the corpus. On average, the Finnish informants use more Swedish incorporations than the Americans do: 34 and 17 per informant in interviews of roughly the same length of time respectively. In Table 1, the incorporations (tokens) into Finnish have been divided into different word and phrase types. Table 1 Swedish incorporations into Finnish by word class and phrase type Word classes and phrase types N % single nouns 397 64.7 single verbs 92 15.0 single adjectives 49 8.0 interjections 29 4.7 noun phrases 19 3.1 adverbs 11 1.8 verb phrases 6 1.0 conjunctions 5 0.7 preposition phrases 3 0.5 adjective phases 2 0.3 pronouns 1 0.2 Total 614 100 Single nouns clearly dominate this set of incorporations (64.7%). There is a fairly high proportion of verbs (15%) but even adjectives (8.0%), interjections (4.7%) and adverbs (1.8%) occur to a certain extent. As to phrases, noun phrases are most frequent while there are only a few occurrences of verb phrases, prepositional phrases and adjective phrases. The phrases are all instances of unambiguous multi-word code-switches, while all other categories represent single-word items. The proportion of phrases is small, i.e., only 30 instances out of 614, which makes up 4.9% of all incorporations. In the English data, as reported by Boyd (1990), the domination of nouns is even more striking: 86% compared to 65% in the Finnish data.

< previous page

page_255

next page >

< previous page

page_256

next page > Page 256

In the English data, besides the nouns, only single adjectives (6.8%) occur in any appreciable number. Single verbs, which are fairly frequent in the Finnish data, occur only a few times in the English data. When it comes to incorporations, the domination of nouns has been reported from many studies. The growth of the proportion of other word classes, especially verbs in relation to nouns, has been claimed to be indicative of a close contact between the languages (cf. Ruke-Dravina, 1969). For Finnish and Swedish, the distribution of the different types of incorporations would indicate long-term co-existence between the languages, and the strong dominance of nouns in the set of incorporations from Swedish into English would indicate a lack of this kind of contact between the two languages. In the present case of FinnishSwedish and English-Swedish bilingualism, this explanation seems reasonable. The patterns of incorporation can be discussed in two respects: in terms of the history of the co-existence of the languages and in terms of the social in the context of immigration. Incorporation and history of co-existence In spite of the typological distance between Finnish and Swedish, there is strong, mainly lexical influence from Swedish both in standard Finnish and in all the Finnish dialects and other spoken varieties of Finnish in Finland. This influence is a result of close language contact between Swedish and Finnish for hundreds of years, on both sides of the boundary between present-day Sweden and Finland, and the previous dominance of Swedish in this contact. Besides, much of the content of western culture has been passed on into Finnish with the Swedish language acting as an intermediary. Consequently, more than half of all the loan words in Finnish come from Swedish or via Swedish (cf. Hakulinen, 1979; Ravila, 1967). This estimate is based on standard Finnish. The Swedish influence is probably even more extensive in the spoken varieties of Finnish in Finland. In a recent report, Swedish loan words in a Finnish dialect, spoken in a bilingual area in southwestern Finland, were investigated (Grönholm, 1988). Only 5% of these loans occur exclusively in this dialect, 80% occur also in some other dialects, and 36% of the loans occur in all Finnish dialects, not only in those closest to the regions where Swedish is spoken. Not only the (almost exclusively) one-way flow of the loan words from and via Swedish but also a considerable number of loan translations in Finnish are a result of the thorough Europeanizing, via

< previous page

page_256

next page >

< previous page

page_257

next page > Page 257

Swedish, of the cultural content of the language. Many of the compounds and phrasal expressions in Finnish are loan translations which, also as to their form, reflect their Swedish sources (cf. Hakulinen, 1979). The rules for integrating loan words into Finnish are productive. The most important ones are to add a vowel after a final consonant cluster and to delete initial consonant clusters, or to adjust verbs to a certain conjugation paradigm, e.g., Swedish stopp 'stop' > toppi or Swedish stoppa 'to stop' > topata. Lexical parallels between Swedish and different varieties of Finnish in Finland provide a large area of neutrality between the languages, which can be expected to provide an avenue of easy incorporations from Swedish into the immigrant varieties of Finnish. Many Finnish immigrants claim that they did not know a single word of Swedish when they came to Sweden, but the knowledge of the lexical parallels which maybe dawned upon the immigrants only when they begin to learn some Swedish - and the knowledge of the integration rules ought to help them to recognize some loan words in their Finnish and to pick up new elements from Swedish and to integrate them. The role of this kind of a ''potential vocabulary" in foreign language learning has been emphasized by Ringbom (1987). Potential vocabulary has both cross-linguistic and intra-linguistic aspects: it is based on the learner's ability to analyze his vocabulary and to make creative use of its different elements, as well as on his ability to recognize relationships between languages. The learner can usually draw upon a considerable potential vocabulary in a target language which is closely related to his first language, but learners differ greatly in how easily they can make the relevant associations. Between English and Swedish, the lexical parallels are probably even more striking than between Finnish and Swedish, due to the typological similarities. As to the basic vocabulary, for most high-frequency core vocabulary, Finnish differs radically from Swedish. In closely related languages, like English and Swedish, even the core vocabulary shows relatively great similarities (cf. Karlsson, 1976). There is a considerable borrowed lexicon, parallel in English and Swedish, but, in contrast to the FinnishSwedish situation, there is currently not any major flow of lexical influence from Swedish to English except in small bilingual groups such as that in Göteborg. Maybe the strong domination of nouns and the lack of verbs in the set of Swedish incorporations into English results at least partly from the lexical similarity due to these

< previous page

page_257

next page >

< previous page

page_258

next page > Page 258

facts. English speakers do not need to incorporate, for instance, the Swedish verbs kommunicera 'to communicate', reglera 'to regulate', or serva 'to serve' because they use the English equivalents of the verbs. On the other hand, many Finnish speaking immigrants extend their vocabulary with kommuniseerata, regyleerata and sörvata. The examples above lead to one more comment concerning standard Finnish and language policy in Finland. An important issue for Finnish language purists has been, from 1600 to the present, to rid at least the Finnish written language of all foreign, especially Swedish influence. Loan translations, contributions from the dialects, and the use of possibilities offered by Finnish derivational morphology, have been the means relied upon to attain this goal. Hundreds of words in different domains, where other languages use international loan words, have been replaced as a result of active language planning, e.g., puhelin 'telephone', kirjallisuus 'literature', tasavalta 'republic', kielioppi 'grammar', yliopisto 'university' (Hakulinen, 1979; Lainio, 1989). This kind of vocabulary is very easily incorporated into Sweden Finnish, like Swedish universitet > universiteetti, Swedish trafik > trafiikki (StFi liikenne), Swedish fixa > viksata. Over two thirds of (all) cases of incorporation of Swedish items into English involve Swedish or partly Swedish compounds. This suggests that Swedish words are resorted to primarily for semantically "heavy" concepts which in many cases do not have a direct English equivalent (Boyd, 1990). Let us illustrate this by Swedish dagmamma, literally 'day mother', a person who takes care of children in her home when the parents are working outside their homes. The English speakers incorporate the compound in a non-integrated shape, like a code-switch, while most of the Finnish speakers convert the compound into the loan translation päivä-äiti. Day care of children has been institutionalized in a similar way in Sweden and Finland, and the linguistic expressions for the phenomena are organized in a similar way as well. This allows for a loan translation in Finnish. Incorporation and immigrant networks Another factor that supports incorporation into Finnish is related to the social context of immigration. Finns tend to have dense social networks within their own speech community. For example, the Volvo factory is an environment where work-related terminology can be acquired, not from Swedes in the first place but from Finnish work mates who intro-

< previous page

page_258

next page >

< previous page

page_259

next page > Page 259

duce the terms in an already integrated shape to the newcomers. It might be the case that a newcomer acquires the Finnicized verb templata far before its Swedish equivalence stämpla 'punch in/out'. New norms for language use can be established among Finnish immigrants, including a certain amount of established borrowings. Some of these established borrowings are probably accepted as Finnicized by many members of the speech community. The Finnicized elements become a part of the neutral lexicon, which can gradually be extended. The distribution of word classes in Grönholm's (1988) corpus of 1400 borrowed lexemes in the southwestern Finnish dialect is 60.4% nouns, 28.4% verbs, 6.3% adjectives, 1.5% adverbs and 0.7% interjections. The Göteborg corpus reflects these proportions fairly well with the exception of verbs, in spite of the fact that Grönholm's calculations are based on What she calls lexemes, i.e., more or less established loans, while in the Göteborg data calculations are based on tokens and so-called nonce loans as well as code-switches are included. If the word class distribution in Grönholm's study reflects a long-term language contact, it could be argued that according to the Göteborg data, the situation of Sweden Finnish is approaching the nature of the long-term language contact on the Finnish side of the boundary. This type of a speech community does not exist for Americans in Göteborg. Because of their family and work situations, their social networks are mainly among Swedish speakers. The Americans do not receive the same kind of support from the other English speakers when incorporating Swedish elements. They must rely on individual solutions (see below). Lexical neutrality and language pairs During the coding work in this project, one of the first clues to differences between the two groups was actually that decisions about which items to consider to be of Swedish origin were relatively clear-cut in relation to the English data but not in the relation to the Finnish data. The decision as to whether a specific item should be counted as being of Swedish origin or not, could have been done from different points of view. A liberal criterion was applied: if there was reason to believe that the speaker had acquired the specific lexical item in the immigrant context, that item was considered to be an incorporation, whether it occurs in some of the varieties of Finnish in Finland or not. Verbs like kommuniseerata and analyseerata are used by many Sweden Finns. The

< previous page

page_259

next page >

< previous page

page_260

next page > Page 260

Swedish derivational suffix -(is)era has been maintained when the verbs have been incorporated into the immigrant varieties of Finnish. In standard Finnish, the Swedish derivational suffix has been substituted by a native one -oi, StFi kommunikoida, analysoida. The lexical neutrality between Finnish and Swedish is probably greater than one might expect for such typologically different languages. The boundaries between the areas of overlap and non-overlap are fuzzy. It seems that this kind of similarity between the languages provides a lexically neutral area which the bilingual speaker can utilize without diverging too much from what he or she considers native norms. It also provides an easy avenue for further incorporations and, on the other hand, possibilities for avoiding borrowing when loan translations can easily be used. The dense social networks within the Finnish speech community allow for nativization of a certain amount of incorporations, which extends the area of neutral vocabulary for the individuals. For the English speakers, the lexically neutral area might even be larger than for the Finnish speakers, but it seems that the role of lexical neutrality is different for incorporations from Swedish into English. The large area of common vocabulary decreases the need of incorporations and restricts the incorporations mainly to nouns and especially compounds which refer to different aspects of Swedish society. These nouns are definitely outside the potential neutral lexicon. In the English data, the Swedish incorporations are easily identified. Most of the incorporations are heavy compounds, like bostadsförmedlingen 'the housing exchange' or fortbildningsavdelningen 'the department for further education'. They are markedly distant from English nouns and they can rather be paralleled with noun phrases. Due to the nature of Swedish-dominated networks, for most of the Americans the incorporations do not easily achieve a status of established loans within the speech community. Furthermore, because of their language-related jobs, many Americans probably try to avoid Swedish incorporations (Boyd, 1990). Morpho-syntactic integration of Swedish incorporations For the morpho-syntactic integration of Swedish incorporations, four possibilities are taken into account in principle.

< previous page

page_260

next page >

< previous page

page_261

next page > Page 261

1) Native morphological markers, attached to the incorporated items where morphological markers are expected, indicate that the incorporations are morpho-syntactically integrated into the discourse. In Table 5 below, the category 'Finnicized/Anglicized' refers to these cases. 2) When the incorporations are morphologically unmarked, they can be accurate in their context. The category 'Neutralized' in Table 5 refers to this utilization of morphological unmarkedness. 3) Cases of unmarkedness may also indicate that the speaker has not been able to solve an integration problem which has arisen when a certain function is marked in different ways in the two languages. Another indication of integration problems is double marking of a certain function. In Table 5, these cases have been put into the category 'Conflict'. 4) Finally, the speaker can choose to maintain Swedish morphological markers and/or syntax. These four possibilities will be used for a cross-linguistic comparison of morpho-syntactic integration. The comparison will be based on the data analyzed from somewhat different points of view and by applying somewhat different categories for Finnish and English morphology. Note, therefore, that the categories used in Tables 2-4 for the presentation of Finnish and English data respectively are not directly comparable with these four possibilities of integration. Morpho-syntactic integration into Finnish Finnish is a relatively synthetic language, suffixation being its most characteristic morphological process. Finnish morphology is two-fold in that the processes of inflection, derivation and compounding are different for nominals (nouns, adjectives, pronouns and numerals) and verbs. A third word class category, particles, is a rather heterogeneous category of uninflected words or words which only have incomplete inflectional paradigms, i.e., adverbs, conjunctions, interjections, post (and pre-)positions. A nominal word stem can take suffixes for derivation(s), number, case, possessive and clitic(s). Number and case marking are the most characteristic inflectional processes, because they are applied to all the constituents in a noun phrase. The nominals can also occur morphologically unmarked as free morphemes. The morpho-syntactic feature the

< previous page

page_261

next page >

< previous page

page_262

next page > Page 262

free allomorph stands for is nominative singular, which is considered to be the base form of the nominals. Finite verbs take suffixes for derivation, passive, tense/mood, person and clitics, and infinitive verbs for derivation, passive, infinitive, number, case, possessive and clitics. Verbs are not as clearly free morphemes as the nominals are, because in most of the verb forms the stem morpheme takes one or more bound morphemes. In Table 2 below, the degree of morpho-syntactic integration of Swedish incorporations is shown by using four levels. Table 2 Morpho-syntactic integration of Swedish incorporations into Finnish Possibilities of integration N % 1) Finnish inflection 320 52.1% 2) Other Finnish morphology 78 12.7% 3) Unmarked 145 23.6% 4) Swedish inflection or syntax 71 11.6% Total 614 100% Items belonging to group I are marked with one or more Finnish inflectional suffixes; they can be considered unambiguously to be morphosyntactically integrated into Finnish discourse. The Swedish compound arbetsskada, 'work-related injury', in examples (3a) and (3b) receives Finnish case suffixes for elative and accusative respectively: (3a) Jouduin puhutteluun täst ARBETSSKADAst this-EL -EL I was called up for an enquiry about the work-related injury (3b) Oisitko sä voinu tämän ARBETSSKADAn estää? this-ACC -ACC Could you have prevented this work-related injury? Groups 2 and 3 contain items which are unmarked from the point of view of inflection. In group 2, though, there are other markers of integration, i.e., a derivational suffix, like in Swedish slipare > liippaaja 'grinder', a partly Finnicized compound, like in invandrarjuttu 'immigrant stuff, or a vocalic "suffix", used for phonotactical integration, like in Swedish rast > rasti 'break'. On these grounds, the morpho-

< previous page

page_262

next page >

< previous page

page_263

next page > Page 263

syntactic status of these incorporations is clear: they are to be considered nominative singular. We can add group 2 to group 1, and we get a total of 64.8% of the incorporations that can be considered morpho-syntactically integrated. Items in group 3 are not marked with any Finnish suffixes at all. Most of these morphologically unmarked incorporations (114 instances) are nouns or adjectives which can be interpreted as being marked for nominative singular. In these instances, one has to rely on syntactic, contextual criteria when judging whether nominative singular could be an accurate case form. In fact, mostly it seems to be accurate, like in the following examples: (4a) se on mun sisäinen KÄNSLA it is my inner feeling (4b) suomi on jotenkin ni OPRECIS kieli Finnish is somehow such an unprecise language In about 25 instances, some other case form than nominative singular could be more accurate. In those instances, it seems that there is a complication in the incorporation process, which the speaker has not been able to solve in a satisfactory way. The rest of the items belonging to this group of morphologically unmarked items are interjections, like herregud 'oh my God', precis 'precisely' and a few conjunctions. Those instances of morphologically unmarked incorporations which do not cause an apparent syntactic confusion, can be added to group 1 and 2, which are unambiguously integrated. If we exclude the 25 questionable instances, we end up with a total of 84% of all Swedish incorporations, and we consider these to be morpho-syntactically integrated into Finnish discourse. From this it can be concluded that morphosyntactic integration into Finnish is the norm for Finnish speakers. The incorporations discussed above are single-word items. In group 4, both single-word incorporations maintaining Swedish inflection, and multi-word incorporations showing Swedish syntactic features, are included. These items comprise only 11.6% of all instances. These are the incorporations that can be considered "real" code-switches. What about linguistic neutrality? A narrow area of morphological neutrality was predicted for Finnish and Swedish, on the basis of the typological distance between the languages. When Finnish is in contact with Swedish, the areas of morpho-syntactic neutrality ought to be

< previous page

page_263

next page >

< previous page

page_264

next page > Page 264

found in the occurrences of free morphemes, i.e., nominals in nominative singular, and particles occurring without suffixes. These are the morphologically unmarked instances. This possibility seems to be highly utilized. Groups 2 and 3, which are unmarked from the point of view of inflectional morphology, make up as much as 36.3% of all incorporations. Some of these incorporations, as we have seen, have been Finnicized by using other morphological markers. In some instances, there are no morphological markers at all. The unmarked cases seem to be open for different degrees of phonological integration. When using morpho-syntactic criteria, most of them can be considered to be integrated, but when using phonological criteria, both integrated and non-integrated singleword incorporations are found within this group. Furthermore, it seems that if difficulties in attaching suffixes on nominals arise, the problem can be solved by using the unmarked form and giving it the status of nominative singular. Then the syntactic context can be adjusted in a way that makes the unmarked case form appropriate. Doing this, the narrow possibilities for morphological neutrality between Finnish and Swedish can be extended widely, which results in increasing intra-linguistic neutrality. Morpho-syntactic integration into English As we have seen above, the set of Swedish incorporations in English is strongly dominated by nouns. In about half of the incorporations in the English data (92 out of 207 cases), no morphological marking is needed. The cases where morphological marking is needed, will be discussed in two stages: morphological marking other than definiteness (27 cases), and definiteness marking (88 cases), which shows a more complex pattern. Table 3 Morphological integration of Swedish incorporations into English other than definiteness Possibilities of integration N % Swedish morphology 21 77.8% English morphology 4 14.8% No marking 2 7.4% Total 27 100%

< previous page

page_264

next page >

< previous page

page_265

next page > Page 265

In Table 3, morphological integration other than definiteness is shown by using three categories: Swedish morphology, English morphology and no marking. This affects only a small number of examples, primarily nouns in the plural and a few tense markings on verbs. There are only 27 items of the 207 which ought to have received some sort of morphological marking, other than definiteness. Table 3 shows a clear tendency towards not integrating morphologically: 77.8% of the items receive a Swedish marker and only 14.8% receive an English marker. In Swedish, definiteness of single nouns is marked by a suffix (-en or -et , ex: huset 'the house'). In English, it is marked by a pre-posed article. A total of 88 cases of semantically definite Swedish-origin single nouns occur in the corpus. In Table 4, the different ways of marking definiteness are shown. Table 4 Incorporation of Swedish nouns in definite contexts into English Possibilities of integration N % examples 22 25.0 the FOLKHÖGSKOLA English def marking only 40 45.5 FOLKHÖGSKOLAn Swe def marking only 3 3.4 the FOLKHÖGSKOLAn Both Swe and Eng marking 7 7.9 FOLKHÖGSKOLA Demonstrative without suffix 14 15.9 this FOLKHÖGSKOLA Demonstrative with suffix 2 2.3 this FOLKHÖGSKOLAn Total 88 100 'folk high school' There is not such a clear pattern with the definite nouns as with other cases of morphological integration. The indefinite Swedish origin nouns were categorically marked with the English indefinite article. Swedish has, like English, a preposed indefinite article. Nevertheless, here, too, there seems to be a tendency towards Swedish marking, but not as strong as for other morphological marking. In 45.5% of the occurrences, only a Swedish definite article is used. Cross-linguistic comparison of patterns of morpho-syntactic integration On the basis of the discussion above, the processes of morpho-syntactic integration in Finnish and English are summarized in Table 5. The figures derived from Tables 2, 3 and 4 are converted into four categories.

< previous page

page_265

next page >

< previous page

page_266

next page > Page 266

Table 5 Morpho-syntactic integration of Swedish incorporations in a cross-linguistic perspective Finnish data English data Finnicized 65% Anglicized 13% Neutralized 20% Neutralized 47% Conflict 4% Conflict 7% Swedish 11% Swedish 33% 100% 100% 'Finnicized/Anglicized' refers to the incorporations marked with native morphological markers. We can compare the proportion of 65% of incorporations showing Finnish morphology with only 13% of incorporations marked with English morphological markers in the English data. These are the items with English definiteness markers plus some examples of Swedish nouns with English plural -s. These figures reflect the suffixal structure of Finnish and the analytic structure of English, but also, in relation to the proportion of maintained Swedish morphology, the Finnish speakers' preference for integration and the English speakers' preference for non-integration. 'Neutralized' refers to the instances where morphological markers are not needed. In the English data, 47% of the instances are accurate as morphologically unmarked. The English speakers can take advantage of the structural similarities between the languages when using Swedish items. In the Finnish data, 20% of the instances are accurate as morphologically unmarked. The relatively limited possibilities of morphological unmarkedness are highly utilized by the Finnish speakers. 'Conflict' refers to the cases where a morphological marker could be expected but is missing, or where there is a double marking of a certain feature. In the Finnish data, we find the 25 instances of nominals, marked with 0-suffix for nominative singular when another case form could be expected. In the English data, we find the 21 instances of missing or double definiteness markers, and two instances of a missing plural marker. Compared to the other categories, these instances are not very frequent. Finally we have instances where Swedish morphology and syntax are maintained 11% of the incorporations in the Finnish data and 33% in the English data belong to this category.

< previous page

page_266

next page >

< previous page

page_267

next page > Page 267

Conclusions On average, Finns incorporate Swedish elements more frequently and they show more variation as to word and phrase types than Americans do. Americans incorporate almost exclusively Swedish nouns and especially compounds. The general pattern is that Finns incorporate Swedish items in a manner approaching ideal borrowing, i.e., morpho-syntactically integrated, while the Americans' incorporations approach single-word code-switches. The notion of linguistic neutrality, referring to similarities between languages, has been discussed in relation to the types of incorporated items and in relation to morpho-syntactic integration of these items. Typological similarities between languages provide large areas of neutrality on different levels, as in the case of Swedish and English. Between English and Swedish, a large area of lexical similarities seems to reduce the need of incorporations to nouns which have no direct equivalence in English. The morpho-syntactic similarities between these languages allow for the preference of morphologically unmarked incorporations. Morphological unmarkedness, then, might be the factor that allows for phonological non-adaptation. When phonology is taken into account, these incorporations must definitely be characterized as nonintegrated (Boyd, 1990). The Americans seem not to have productive rules for integrating Swedish incorporations. The network structures of American immigrants in Göteborg do not support integration of the incorporations either. Every individual must create her/his own rules, for example, to handle the problematic case of definite nouns. Typological distance between Finnish and Swedish predicts narrow areas of lexical and morpho-syntactic neutrality. Nevertheless, there is another source of lexical neutrality between present-day Finnish and Swedish. It is the long-term crosslinguistic influence from Swedish into Finnish. This influence is seen in a great number of loan words from different periods of time, showing different degrees of phonological and morphological integration, in productive rules for the integration of loan words, and in the large number of loan translations that reflect their Swedish sources. The Finnish immigrants in Sweden can take advantage of the lexically based similarities between Finnish and Swedish. The fact that no sharp boundaries between Finnish and Swedish vocabulary can be drawn might provide a relatively safe ground of neutral vocabulary for

< previous page

page_267

next page >

< previous page

page_268

next page > Page 268

the bilingual speaker, and an avenue for additional incorporations from Swedish. The integration rules are productive and need only to be activated when incorporating more elements from Swedish. The relatively dense networks among the Finnish immigrants is an additional factor supporting integration of Swedish incorporations. Acknowledgements The present study has been carried out, in cooperation with Sally Boyd, at the Department of Linguistics, University of Göteborg, within the project Variation and Change in Language Contact. The project is financed by the Council for Research in the Humanities and Social Sciences. References APPEL, R. & P. Muysken (1987), Language contact and bilingualism. London: Edward Arnold. BOYD, S. (1990), Attrition or expansion? Changes in the lexicon of Finnish and American adult bilinguals in Sweden. Paper presented at the conference Progression and Regression in Language, Botkyrka August 1990. CLYNE, M. (1987), Constraints on code switching: How universal are they? In: Linguistics 25, 739-764. DI SCIULLO, A-M., P. Muysken & R. Singh (1986), Government and code-mixing. In: Journal of Linguistics 22, 1-24. GRÖNHOLM, M. (1988), Ruotsalaiset lainasanat Turun murteessa. Åbo: Åbo Akademis förlag. HAKULINEN, L. (1979), Suomen kielen rakenne ja kehitys. Keuruu: Otava. JAAKKOLA, M. (1984), Immigrant life: Networks, life styles and ethnic mobilization of Finnish immigrants in Sweden. Vammala: Vammalan Kirjapaino Oy. KARLSSON, F. (1976), Finskans struktur. Lund: Liber Läromedel. LAINIO, J. (1989), Spoken Finnish in urban Sweden. In: Uppsala Multiethnic Papers 15. Uppsala: Centre for Multiethnic Research. POPLACK, S. (1980), 'Sometimes I'll start a sentence in Spanish y termino en Español': Toward a typology of code-switching. In: Linguistics 18, 581-618. POPLACK, S., S. Wheeler & A. Westwood (1987), Distinguishing language contact phenomena: Evidence from FinnishEnglish bilingualism. In: P. Lilius & M. Saari (eds.), The Nordic languages and modern linguistics 6, Helsinki: University of Helsinki Press. RAVILA, P. (1967), Lainasanamme. In: L. Hakulinen, O. Ikola & P. Ravila, Kirjoituksia suomen kielestä. Porvoo/Helsinki: SKS.

< previous page

page_268

next page >

< previous page

page_269

next page > Page 269

RINGBOM, H. (1987), The role of the first language in foreign language learning. Clevedon, Philadelphia: Multilingual Matters. RUKE-DRAVINA, V. (1969), Språk i kontakt. Aldusserien 263, Lund. SANKOFF, D., S. Poplack & S. Vanniarajan (1986), The case of the nonce loan in Tamil. Technical Report 1328 Centre de recherches mathématiques. Montreal: Montreal University. SCOTTON, C. (1987), Code switching and types of multilingual communities. In: P. Lowenberg (ed.), Language spread and language policy: Issues, implications, and case studies. Georgetown University Round Table on Languages and Linguistics 1987. Washington, D.C.: Georgetown University Press, 1-15.

< previous page

page_269

next page >

< previous page

page_ii

next page >

< previous page

page_271

next page > Page 271

PART 4 LANGUAGE MAINTENANCE AND LANGUAGE LOSS

< previous page

page_271

next page >

< previous page

page_ii

next page >

< previous page

page_273

next page > Page 273

13 Immigrant minority languages and education in Sweden Sally Boyd Prior to World War II, Sweden was an unusually homogeneous country, both ethnically and linguistically (Hammar, 1985: 22). Since that time, this situation has changed dramatically, so that at the beginning of the 1990's over a million persons, officially considered to be immigrants, now reside in the country. This means that about 12% of the total population of the country are born abroad, or have one or both parents born abroad (re. Statistiska meddelanden). Table 1 How to make a million immigrants in Sweden (Source: Sveriges officiella statistik Folkmängd 3: 1990) First generation (born abroad) 738 454 240 036 Second generation (under 18 yrs.) Second generation (over 18, est.) 170 000 TOTAL 1 148 490 Immigrants have come to Sweden as labour migrants, refugees and members of the families of natives and of immigrant residents already in the country. Provisions have been made for the education of immigrants and their children at virtually all levels. Sweden has had a reputation for being generous in its immigration policy in comparison with other European countries. Despite the country's good reputation, many problems remain in this area, including in the field of education, and the trends for the future are not all encouraging. This chapter will sketch briefly the history of post-war migration to Sweden, as well as the linguistic and educational consequences of immigration to Sweden and of Swedish immigrant policy, especially in the field of education. Finally, I will discuss some of the problems with the education of immigrant children, especially in the compulsory school, and present a

< previous page

page_273

next page >

< previous page

page_274

next page > Page 274

few speculations as to the future of immigrant minority languages in Sweden. Sketch of the history of post-war migration to Sweden Post-war migration to Sweden can be divided into several periods, some dominated by labour migration, others by refugee migration, and each period having a somewhat different make-up in terms of the migrants' countries of origin. 1945 1950 1960 1970 1980 1990 Low migration Labour migration peaks Refugee migration mainly refugees family tie migration dominant Eastern Europe Nordic countries Latin America, Africa, Germany SE Europe Iran, Middle East Figure 1 Post-war migration to Sweden As in other European receiving countries, the size and composition of the influx of immigrants during most of the postwar period was due more to the supply and demand of labour than to a well-developed immigrant policy in Sweden (Hammar, 1985: 277-279). Labour and ''family tie" migration Labour migration was dominant from the late 50's up to the 70's. In the 50's and 60's, when the Swedish economy was booming, workers were recruited from other parts of Europe (particularly Southeastern Europe), and many came seeking work from the other Nordic countries (especially Finland), with whom Sweden established a free labour market in 1954. In 1970 immigration reached an annual peak of 73,000 persons. Restrictions were made on the migration of non-Nordic residents in 1967, and all non-Nordic labour migration was ended in 1973. However, the open Nordic labour market and 'family tie' migration kept the immigration figures high during most of the 1970's. The increasing importance of family tie migration during the 1970's, together with the fact that about half of the immigrants came from other Nordic countries, reinforced the official view of immigrant minority groups in

< previous page

page_274

next page >

< previous page

page_275

next page > Page 275

Sweden. Sweden never really had a guest worker policy, as for example Germany and Switzerland did, but when the families of migrants began to arrive in greater numbers, officials had to live up to the "immigrant" policy already laid down in the 1960's. As immigrants in ever increasing numbers brought their spouses and families to Sweden, or married and raised families there, it became necessary to accept the fact that many of them would never return to their countries of origin. The migration of spouses and the subsequent birth of a second generation in Sweden have also changed the demands minorities have placed on Swedish society as regards employment, education, social service, opportunity to practice non-Christian religions etc. As long as both the authorities and the migrants themselves considered the immigrants' stay in Sweden to be temporary, the provision of and demands for these services were not so great. Refugee migration Refugees have also come to Sweden during the entire postwar period. Immediately after the war, and during the early 1950's the majority came from Eastern Europe. Later, during the 60's and 70's, many came from South America, particularly Chile and Argentina. During the 1980's, both labour migration and family tie migration have decreased, while refugee migration has increased dramatically, particularly migration from Iran, as well as from the Middle East, Ethiopia and other regions outside Europe. In the late 1980's Sweden granted asylum to 15-30,000 people annually, making it the largest per capita recipient country in Europe. The migration of many individuals from non-Western, non-Christian backgrounds has put a greater strain on the service apparatus which takes care of new arrivals, as well as on the good will of the Swedes, who, during most of the 60's, 70's and early 80's were relatively well-disposed towards immigrants (Westin, 1984). The refugees arriving in recent years are better educated than the labour migrants and their families who dominated migration in the 60's and 70's, when immigrant policy was established in Sweden. The ways established earlier for taking care of immigrants do not always suit the special needs of the better-educated refugees of the 1980's, many of whom are trained for more skilled and professional occupations, or are demanding university education.

< previous page

page_275

next page >

< previous page

page_276

next page > Page 276

Immigrants currently living in Sweden Looking at the group of immigrants currently living in Sweden, it is still the case that immigrants from the Nordic region, particularly from Finland comprised almost half of all resident immigrants at the end of the 1980's. Other minorities with over 25,000 members are shown in Table 2. Table 2 Some major groups of first and second generation immigrants in Sweden, by country or region of origin, 1990 (Source: Sveriges officiella statistik Folkmängd 3: 1990) Number % of all immigrants Finland 312 402 31% Other Nordic countries 132 533 13% Yugoslavia 60 545 6% Germany 50 188 5% Poland 45 501 4% Iran 39 262 4% Turkey 36 070 4% Africa as a whole 30 480 3% Chile 28 849 3% In short, we can say that the immigrants in Sweden make up a relatively large percentage of population, that it is a heterogeneous group (in terms of language, country of origin), but that the immigrant group as a whole is dominated by a few major groups. Prior to the 80's, the dominant groups came from the Nordic region and Europe, more recently the new arrivals come predominantly from outside Europe. Swedish immigrant policygenerous on paper The official policy of the Swedish state toward immigrants is relatively generous as it stands. The policy as a whole is summarized in the slogan from the parliamentary commission on immigration and minorities

< previous page

page_276

next page >

< previous page

page_277

next page > Page 277

of the mid-70's: "Equality, freedom of choice and partnership" (SOU, 1974: 69). Once an individual is granted work and residence permits, the basic point of departure of this policy is that s/he should be granted virtually the same rights and privileges as the native population ('equality'). In addition, the state should provide extra resources to facilitate a certain linguistic and cultural 'freedom of choice' for the immigrants. The latter is intended to provide the opportunity to retain the language and to practice the culture of one's country of origin, or to adopt Swedish language and culture. The partnership goal is intended to guarantee immigrants the right to participate in political and labour union activity. As mentioned above, the official policies also apply to a relatively broad category of people, not all of whom actually consider themselves to be immigrants. Accordingly, an individual is an immigrant if s/he is born abroad or has at least one parent born abroad. Perhaps due to the fact that the country was unusually homogeneous prior to the War, for Swedes, equality is often equated with uniformity. The Swedish social welfare system is set up on the principle that everyone receives not just equal but the same benefits, whether they are employment benefits, housing, education, health care, social service, regardless of income, for example. This ideal of uniformity means that the first two goals of immigrant policy: equality and freedom of choice are often considered to be in conflict with one another, and this perceived conflict leads to significant problems in making special provisions for immigrants and their children in many different areas of life, for example in education. Education of immigrant children and adults In the compulsory school, almost 100,000 pupils (11%) report that a language other than Swedish is regularly used in the home. The number of these languages at the end of the 1980's was 126 nationwide. Schooling is provided for these pupils in the following four types of classes: (1) preparatory class, where pupils with a variety of backgrounds receive short-term intensive instruction in their own language and in Swedish as a second language (Sw2), preparatory to attending a regular Swedish class (4 below);

< previous page

page_277

next page >

< previous page

page_278

next page > Page 278

(2) mother tongue class (up to grade 6), where all pupils have the same non-Swedish first language (L1). Instruction in the first grade is almost entirely in the children's' L1, with several hours per week of Sw2. In successive years, the proportion of instruction in L1 decreases, as the proportion of instruction in Swedish increases. About 9% of all immigrant pupils in the compulsory school attend school in this type of class. (3) integrated class (up to grade 3) where pupils with a particular L1 constitute about half of the class, and receive some instruction in that language, some Sw2 and some regular instruction in Swedish. The rest of the class consists of Swedish pupils. Only about 2% of the immigrant pupils attend integrated classes. (4) regular Swedish class, plus about 2 lessons per week in the minority language ("home language instruction"), often given by a peripatetic teacher. The remaining 89% of pupils in the compulsory school who report a home language other than Swedish attend regular Swedish classes. Altogether about 60% of the immigrant pupils in compulsory school participated in some form of instruction in L1, the vast majority is in category (4) above. In 1989, about two-thirds of the immigrant pupils were judged to need instruction in Sw2, and almost all of these actually received this instruction. Sw2 instruction is provided if the school (i.e., usually the Sw2 teacher and/or the child's regular Swedish teacher) considers the child to need it, and the instruction is then obligatory for the child. On the other hand, it is the family who decides whether or not the child should receive instruction in the minority language. This instruction is only available if the minority language is actively used in the home, and is voluntary for the child. Both Sw2 and home language instruction for immigrant pupils are also available in the upper secondary school (Swe. gymnasium), though the percentage of participation is lower (45% and 30% respectively). Instruction in Sw2 and in minority languages also take place in the municipal adult education programs, at both the compulsory school and upper secondary levels. Introductory courses in Swedish for adult immigrants have been provided free of charge since 1965. Some of this training is given during working hours, on the job, some in conjunction with occupational retraining, and some in voluntary classes. It should be quite clear from the description above of the types of instruction provided in the compulsory school, that the educational

< previous page

page_278

next page >

< previous page

page_279

next page > Page 279

model is clearly transitional. Even in the mother tongue classes, the proportion of time instruction is given in the child's L1 decreases from year to year, and is reduced to two lessons per week after six years in school. Considering how home language instruction was introduced in the 1970's, it is not so surprising that the Swedish educational system provides bilingual education within a transitional model. The arguments presented at that time, and which led up to the institution of home language instruction (in forms 2, 3 and 4 above) were arguments about the value of instruction in the child's mother tongue in order to facilitate learning of the majority language, Swedish. The bugaboo of a generation of "semilingual" immigrant children (usually characterized as those with an inadequate skills in two languages, resulting from the introduction of L2 when the child entered school) was presented to frighten school officials and politicians and to convince them of the importance of mother tongue teaching. Now that the concept of semilingualism has been rejected by virtually everyone doing research in the field of bilingualism in Sweden, educational programs for immigrant children become a likely target for budget cutbacks. Despite the official policy of freedom of choice, and the provision of 'home language instruction', language shift in the second generation is widespread, and seems to be retarded only slightly by measures such as mother tongue classes (Boyd, 1985). Sweden may have the reputation for being generous in providing bilingual education in L1 for immigrant children, but on the one hand, the educational program is transitional, and on the other, as discussed above, this education is provided in the context of an unusually conformistic society. The tendency among Swedes to link equality with uniformity was probably one reason why the only argument which would convince school officials and politicians of the value of bilingual education was one in which instruction in L1 was seen as a necessary step to development of L2. The results of the programs of "bilingual education" (if the Swedish system described above can be considered as such) are thus in line with the motivation used in instituting the program, and with the expectations of the Swedish society as a whole, i.e., widespread language shift in the second generation. This picture may change slightly, however, as immigrants from more distant countries arrive in greater numbers. On the other hand, the trend at the moment is to reduce funding to immigrant education, particularly to home language instruction.

< previous page

page_279

next page >

< previous page

page_280

next page > Page 280

Problems in field of education of immigrants Although the policy towards immigrants, particularly in relation to language and education may be seen as relatively generous, several problems have remained pressing in this field. First, both home language and Sw2 instruction for the majority of pupils (model 4 above) are often organized along the lines of remedial education, requiring the pupils to leave their regular class for one or more lessons per week, as they would do to receive instruction from the special education teacher or the speech therapist As a result, this instruction takes on the quality of compensatory education, rather than that of a benefit. Many pupils do not wish to leave their regular classes to go to home language instruction, and so, since the instruction is voluntary, they drop out of the program. The regular classroom teachers are often disturbed by individual pupils' having to leave class to attend home language instruction or Sw2 instruction. The organization of instruction is also considered to make the instruction expensive for the state. In a recent report from the National Accounting and Auditing Bureau (RRV), both home language instruction and instruction in Sw2 were criticized as being "inefficient". Some of the main reasons for this were claimed to be the fact that the students must miss other instruction to attend home language or Sw2 instruction, the fact that the groups taught at the same time are small, and the fact that the teacher had to travel from school to school during the day. This type of organization, i.e., as an extra, optional subject, for which the pupil must leave his regular class, makes the programs likely targets for budget cutbacks, which is exactly what RRV (at least indirectly) suggests. They suggest further that instruction in Sw2 and in "home language" be scheduled after regular school hours, or during the pupils' breaks during the day. This organization will make the instruction less attractive both to the students and to prospective teachers. Second, teacher training courses have not met the demands for teachers in these fields, so many teachers working in the programs are inadequately trained. For Sw2, the supply of trained teachers is gradually increasing, while the supply of teachers in home languages is decreasing, despite the increasing number of immigrant pupils in the schools. One reason for the decrease in the supply of home language teachers is that the program for becoming an accredited teacher of "home language" now requires the candidate to become accredited in teaching at least one and usually two subjects other than the candidate's mother tongue (e.g., Swedish, English, social science or natural

< previous page

page_280

next page >

< previous page

page_281

next page > Page 281

science). This change implies that it is difficult for many prospective candidates to pass the entrance exams to the programs at the teachers colleges leading to accreditation in home language. The problems of a lack of trained home language teachers are exacerbated in many parts of the country, where there are only a small number of pupils of approximately the same age with the same L1 in a particular area. In these areas, home language instruction must be carried out as part-time employment for the teacher, and/or s/he must teach in a number of different schools. This situation has made these positions unattractive from the beginning of the program. If the lessons are confined to after school hours, they will become even less attractive. Third, many teachers and other personnel in close contact with minorities feel the need to incorporate a multicultural component in their own training, as well as in the school curriculum, in order to meet the needs of individuals from an increasing number of linguistic and cultural backgrounds now living in Sweden. This need has also been pointed out by virtually everyone working in the field of immigration research in Sweden. As long as migration to Sweden was dominated by persons coming from geographically and culturally less remote parts of the world, the main problem for integrating them into Swedish society was seen to be a language problem. During the 80's and possibly also the 90's, as refugees and immigrants come from further afield and from a greater diversity of cultural backgrounds, instruction in Swedish alone is no longer adequate to meet the needs of immigrants and native Swedes in an increasingly multilingual, multicultural society. On one hand, immigrants are currently making greater demands in areas other than language instruction, that their particular needs be met such as their ability to practice their religion, and live in a manner more in tune with their cultural heritage. For their part, at least some Swedes are beginning to feel the need to learn more about immigrants and refugees culture of origin, their ways of life in Sweden etc. Instruction in these areas is clearly needed in many educational programs for persons in contact with immigrants and refugees. Conclusions and discussion Two current political trends would seem to have an important impact on the education of immigrant minority children in Sweden, and thus on

< previous page

page_281

next page >

< previous page

page_282

next page > Page 282

the development of immigrant minority languages there. One has to do with Swedish domestic policy, and the other with international policy. The domestic political trend which is evident at present in Sweden is the increasing decentralization of funding decisions and of administration, and the transfer of power from the federal government and its agencies to the local municipalities. Within the municipalities, the trend is the same: decision-making is being decentralized to so-called neighbourhood boards. This decentralization will make it more difficult for immigrants to make their voices heard on a variety of issues in many municipalities and local neighbourhoods. In the area of education, it has recently led to cut-backs in home language instruction, especially the 2 hour per week model for small groups of scattered pupils. Previously, municipalities receive extra funding to provide for this instruction, and this funding is earmarked for instruction of immigrant pupils. Since 1991, this money is now incorporated into educational funding as a whole, so that the amount of instruction offered in Sw2 and home language is left up to the individual municipalities. If, in addition, home language lessons are moved to outside regular school hours, this will lead to a significant drop in the number of pupils attending, as well as to a further decrease in the already inadequate supply of home language teachers. The future is somewhat brighter for mother tongue classes, which are considered not to be so costly and also for instruction in Sw2, where the supply of teachers (who of course belong to the majority group) is greater. The need for this instruction is also regarded as more important by those who make the decisions. The second important political trend is due to the fact that Sweden applied for EC-membership in 1991. EC-"harmonizing" has already led to a revision of refugee policy, which has resulted in a reduction in new refugees arriving, as compared to the late 1980's. At the same time, there is a certain amount of support in the government for allowing a significant increase in migration from Eastern Europe, particularly from the Baltic republics. This migration may increase in the near future so that the trend mentioned earlier, for refugees to come increasingly from outside Europe may be broken, while the number of refugees, which we (due to EC-harmonizing) might otherwise expect to be reduced, may remain stable or even increase. Membership in EC will probably lead, if not to a deterioration of immigrant minority rights in Sweden, at least to a lack of improvement, as Sweden, in comparison with other EC countries has had a relatively generous policy in several areas (Hammar, 1985). There is, however,

< previous page

page_282

next page >

< previous page

page_283

next page > Page 283

one area of immigrant policy which may benefit immigrant minorities as a result of harmonizing with EC policy, and this is in the area of education. At present, in comparison with other European countries, it is relatively difficult to start independent schools in Sweden. This situation is completely in line with the linking of equality and uniformity which was mentioned above. Hopefully, this situation will change in the future, making it easier for immigrant minority groups to start their own independent schools with government support. If future cutbacks in home language instruction are as severe as they now appear to be, this is a course of action which certain immigrant groups may choose if they feel that preservation of linguistic and cultural freedom of choice is important enough. References BOYD, S. (1985), Language survival. A study of language contact, language shift and language choice in Sweden. Gothenburg: University of Gothenburg, Department of Linguistics (PhD. Thesis). HAMMAR, T. (ed.) (1985), European immigration policy. A comparative study. Cambridge: Cambridge University Press. WESTIN, C. (1984), Majoritet om minoritet. En studie i etnisk tolerans i 80-talets Sverige (The majority' s view of the minority. A study of ethnic tolerance in Sweden in the '80s). Stockholm: Publica. Government documents: RIKSREVISIONSVERKET (National Accounting and Auditing Bureau) (RRV) (1990), Invandrarundervisningen i grundskolan (Immigrant instruction in the compulsory school) Revisionsrapport 1990-09-25. Stockholm: RRV. INVANDRARUTREDNINGEN 3 (The Government Commission on Immigration, Part 3) (1974), Invandrarna och minoriteterna (Immigrants and minorities) English summary. Stockholm: Department of Labour, SOU 1974: 69. STATISTISKA CENTRALBYRÅN (The Swedish Central Bureau of Statistics): Statistiska meddelanden Hemspråk och hemspråksundervisning (Statistical reports: Home language and home language instruction) no. U72 SM 9002. Örebro: SCB. STATISTISKA CENTRALBYRÅN (The Swedish Central Bureau of Statistics) Sveriges officiella statistik Folkmängd 31 dec 1989. Del 3 (Public statistics for Sweden. Population. 31 Dec 1989) Stockholm: SCB.

< previous page

page_283

next page >

< previous page

page_ii

next page >

< previous page

page_285

next page > Page 285

14 Romani at the crossroads Donald Kenrick Romani is a language of north Indian origin, related to Hindi and Punjabi. The ancestors of the present day speakers of Romani migrated from India some time between the 3rd and 11th century AD. We do not know whether at the time of departure Romani was already split into dialects or whether all the dialects developed at a later date. My own hypothesis is that Romani actually developed outside India - in Iran - from the mixing of ethnic groups speaking a number of Indian languages or dialects. There are over 50 dialects of Romani. The total number of speakers in Europe could be as high as 4 million (see Appendix). Dialects are of two kinds. There are 'national' dialects such as Slovak Romani and Finnish Romani which are spoken by the vast majority of Gypsies who live a fairly sedentary (i.e., non-nomadic) life within a fixed geographical boundary. And there are the dialects of mainly nomadic groups such as the Kalderash and Lovari who can be found in many countries. There has not yet been an overall study of the dialects. It has been suggested that the Sinti and Manouche dialects should be classed as a separate language from Romani. However, the distinction between 'language' and 'dialect' is often not linguistic but political, and until such time as Sinti or Manouche develop a written literature I see no practical value in treating them as a separate language. There are some groups of Gypsies (and I will define my use of this word in a moment) who used to speak Romani but do so no longer. Many of them speak a variety of the majority language of the country they live in, with an 'exotic' vocabulary based largely on Romani. The most well known of these means of communication are Calo (in Spain and Portugal) and Pogadi Chib in England. Some groups have lost the Romani language in this century. For example, it seems impossible to find now anyone speaking the Welsh dialect which was the basis for the largest dictionary of Romani, that of John Sampson in the 1930's. The term 'Gypsy' is not in itself pejorative and I propose to use it sometimes instead of and as a synonym for Romani-speaker or descen-

< previous page

page_285

next page >

< previous page

page_286

next page > Page 286

dant of Romani speakers. It is an abbreviation of 'Egyptian' and one of the terms used of the Romani speakers when they first came to Europe. Finally, I will use Romani for referring to the language of Indian origin spoken by most Romanies, whereas Romany refers to a member of an ethnic group of Indian origin, or is used as an adjective appertaining to this group. I prefer this established English usage to using as do some writers the Romani word 'Rom'. The sociolinguistic situation Romani speakers are in a minority in each country where they live, even if they are in a majority in a town, such as Shuto Orizari in Macedonia or part of a town, such as the Gypsy quarter of Sliven in Bulgaria which recently voted in a Gypsy member to the Sofia Parliament. For a number of reasons, dealings with officialdom are not in Romani but in the majority language, even when the Romani speakers form a substantial minority. These reasons include the low social status of Romani even among its own speakers (hence the reluctance of officials to learn it when they might perhaps learn other minority languages such as Welsh or Hungarian), the lack of a standard written language, the lack of modern terminology, and in some cases, active discouragement. For example, in Bulgaria the use of Romani was actively discouraged until the changes in November 1989. After 1945 there was a periodical for Gypsies with the Romani title Nevo Drom (New Way). It was in Bulgarian, but there were occasional songs in Romani. The title was changed to the Bulgarian Nov Put and the songs disappeared, while leading articles preached assimilation. In 1951 the Romani theatre in Sofia was closed and its director sent into internal exile. As there is little written on the subject, I will quote two incidents from my own experience. I was talking to the librarian in the town of Kotel in Romani when his niece came in. She greeted him in Bulgarian and they conversed in that language about a book she wanted for her schoolwork. He then introduced her to me and I found she could speak Romani fluently, as could all the Gypsy children I met in that town. In a town in eastern Europe I met a young Gypsy whose Romani was rather weak - both lexically and grammatically - and he was trying to learn more of the language from his grandparents as his parents had not spoken the language with him. I met him some ten years later at an international Gypsy conference and his use of the language had

< previous page

page_286

next page >

< previous page

page_287

next page > Page 287

improved. However, he asked me if I could get a Romani grammar book for him as he wanted to learn Romanes (real Romani) by which I assume he meant either Kalderash or one of the Yugoslav dialects he had heard at the conference. Previously, he had been ashamed at not speaking Romani, now he was ashamed of his own dialect. Linguistic change Because of the need to know another language for dealings with officials, because there are - to my knowledge - no economically self-supporting communities of Gypsies in Europe, and because there are no monolingual Romani reception classes in primary schools, we will not find any monolingual Romani speaker much over the age of six. There are, however, many communities (for example in Slovenia) where, because of the social isolation of the Gypsies, the children up to six are monolingual in Romani. As a result, all the three elements of language change occur, i.e., bilingualism or multilingualism, codeswitching, and language loss. I am not aware of any scientific studies of any of these phenomena in Europe, though Hancock (1975) has written extensively about language usage amongst the speakers of the Kalderash dialect in the USA. What follows will therefore be based again on my own observations (see also Kenrick, 1989), quotation from others who have talked of linguistic change in more general studies of Gypsies, and some textual analysis. I will not attempt to cover all 50 dialects or all the countries of Europe but will give an indication of the phenomena. It will be evident, that just as in the last century many people saw the Gypsies as an opportunity for missionary work without the need to travel to Africa or India, there is a great opportunity not far from most linguists' doorstep for active and passive research amongst the Romani speakers of Europe. Bilingualism and multilingualism Muslim Gypsies in Yugoslavia speak Romani as the home language but also Macedonian and/or Serbian, Turkish and often Albanian. The writer Shaip Jusuf also speaks a second dialect of Romani as do many Gypsies in the Skopje area. On a previous visit to Holland, I visited a family who, as it happens, were being moved from place to place in their caravan and in no

< previous page

page_287

next page >

< previous page

page_288

next page > Page 288

mood for linguistic research. I think they were Lovari rather than Kalderash. Between the parents, their children and the son and daughter-in-law, they could speak Russian, Bulgarian, Serbian, Hungarian (as far as I could check), Italian, German, French, Swedish and of course Romani - the common language for them all. If they are still in Holland, I am sure at least the younger generation will have added Dutch to their repertoire. Most of the Yugoslav families who come to England each summer can communicate adequately in Romani, Serbian, Italian, German and French. I get the impression that these migrant Yugoslavs tend to marry other migrant Yugoslavs in the country where they have residence and that Romani is not always the language of communication for young couples. With Lovari and Kalderash often Romani is the only common language known fluently by husband and wife. In a recently arrived family in London, the grandmother speaks Russian, Polish and French. The husband, born in Canada, and the wife, born in France, use Romani as their usual language of communication. The husband at least speaks Italian and Swedish. Messing (1988) in his study of Athens Romani, found that the older generation spoke Turkish, Greek and Romani - the younger generation Greek and Romani. Code-switching Here are examples of sentences from one side of a telephone conversation between a Kalderash woman whose family moves between Australia and England and a relative who lives mainly in Canada and Australia. So both know English as well as Romani. The notes are from about five years ago and the speaker was then aged around 30. The Romani is italicised: I don't blame M. te marela les (R = if he fights him) It's too late to keras wareso (R = do anything) I L darel, (R = L. is afraid) she's scared of trouble, why not, te marena pes le Roma (R = if the Gypsies quarrel) Won zan * (R = They are going) and I couldn't get them to den duma o D. (R = to talk to D.) Maybe sai* mangela (R = he wants to) Monday I don't know what to do e savensa* (R = with the children) Wo dia ma bis* liri ha lem o TV (R = He gave me twenty pounds and I took the TV) back

< previous page

page_288

next page >

< previous page

page_289

next page > Page 289

In the following examples, the English is affected by Romani syntax: I don't want M. to get the suspicion to run I can't interfere to that Also loan words occur. The nine year old daughter of that speaker was telling a story one day to her younger sister. There is no code switching but two loan words, for both of which I would think she knew the Romani word. (The English words are italicised): asun ekh (R = listen to a) story. Ekh sajori * xaljas ash. Aviljas ekh covaxano*, phendja laki. Av mansa (R = A little girl ate ash. A wich came and said to her: come with me.) Language loss Intermarriage seems usually to lead to language loss in the next generation. On one day in Czechoslovakia I met four couples. They were not a random sample but couples where the men were involved in Romani cultural activities. Only one couple spoke Romani to the children. Of the others: (a) a Romany married to a Czech woman, where the children spoke Czech; (b) a Romany married to a woman from a fairground family, where the children spoke Czech; (c) a Romany who spoke the Slovak dialect married to a Romany who spoke the Carpathian dialect; the family was living in Prague; the couple spoke Slovak to each other and Czech to the children. The dialect of the Finnish Gypsies is currently in a state of change, under the strong influence of Finnish, and may well be dying out. That is to say, parents in both Finland and Sweden are not passing it on to their children, but are using Finnish when they speak to them. The knowledge of Finnish Romani held by many young persons is so little that it cannot even serve them as a secret code. Many Sinti parents too are not passing the language on to their children. From the sparse references that exist in the literature, Soravia (1977), referring to the Romani of Reggio Calabria, states: 'I giovanissimi non lo conoscono e usono piu'. (The young people no longer know it or use it.).

< previous page

page_289

next page >

< previous page

page_290

next page > Page 290

Loss of phonemes As a result of contact with other languages many of the phonemes of Indian origin are disappearing. The retroflex consonants were the first to go. A retroflex l or r is found still in Bulgarian dialects, and some other dialects mark the lost phoneme in other ways (an uvular r in Kalderash). Of the aspirates, ch * disappeared some years back again except in Bulgaria and some Yugoslavian dialects. The remaining aspirates (kh, ph, th) are being lost by the younger generation and they are being replaced by non-aspirated consonants. There are references to this loss in many descriptions of individual dialects, for example, in Italy in the dialects of Sinto piemontese, Sinto lombardo and the Romani of Abruzzo (Soravia), and in Athens. Younger speakers tend to omit aspiration of ph, th, kh (Messing, 1988). He also notes the substitution of Greek phones for Romani, e.g., dz for (dzav) or ts for (tsavo). Loss of gender distinction The neuter gender was probably lost before the Gypsies reached Europe as I have not come across any trace of it. The feminine and masculine are coalescing in Finnish Romani and Hungarian Lovari under the influence of the majority languages. This has meant that in Hungarian Lovari voi (which means 'she' in most dialects) is used for 'he' and 'she'. This makes it difficult to use texts in Lovari with speakers of the otherwise closely related Kalderash dialect. Calques, as they were called by the scholar Gilliat Smith (in his many articles in the Journal of the Gypsy Lore Society), based on the majority languages, abound, e.g., in Bulgaria dui drom (literally 'two roads' used meaning 'twice'). This is a translation or calque of Bulgarian dva puti (id.). Historically, Romani has the word duivar meaning 'twice'. Not strictly a calque is the replacement in Athens dialect of the pronouns for 'he' and 'she' by kava, kaja (originally 'this') on the model of Greek avtos (Messing, 1988). In view of the little research on language change one may wonder what scholars from or close to the Gypsy community have been doing. There are some dictionaries (not all of which have been published), descriptions of a particular dialect, collections of folktales, translations, teaching material for children and some work on standardisation and orthography. I would like to mention the names of a few Gypsies who have written on linguistics and whose works I have listed in the bibliography: Kochanowski (1963), Jusuf (1978), and Manush (1990).

< previous page

page_290

next page >

< previous page

page_291

next page > Page 291

The creation of a literary language The existence and use of a standard written language acts as a brake to linguistic change. This is my justification for a few remarks on this topic. The major efforts to create a written literature in Romani since 1945 have been in Sweden and Yugoslavia. In Sweden as part of a policy to encourage all immigrant children to learn their home language, a number of schoolbooks and other material have been published, largely in Kalderash but some in Lovari. In Yugoslavia there have been a considerable number of books and magazines (often duplicated and short-lived) for adults. Also in Hungary there has been some activity. Parts of the New Testament have been translated and published, particularly in Kalderash dialect. However, as the level of active literacy in the majority language of the countries where the Gypsies live is low and the ability to read Romani even lower, I do not feel that so far the existence of written Romani has had any effect on either holding up language change or on standardising dialect speech. Towards a standardised literary language Linguists have always been less concerned with written language than with spoken language. I feel, however, that the reader should be informed on what has been done in an attempt to create one standardised language for literary purposes. The First World Romany Congress established a standardised list of symbols to be used for writing Romani. The meeting of the Language Commission at the Second World Congress modified this list slightly and took the decision to replace loan words of European origin by words from Hindi and Punjabi. A number of Yugoslav poets have in fact done this, as has the Skopje writer Shaip Jusuf in his grammar and translation of the standard biography of Tito. In practice, the only Indian word which seems to have had any widespread acceptance amongst writers is 'lekh' (write). There was no meeting of the Language Commission at the Third Congress. Between the Third and the Fourth Congress there was an important conference on the Romani language at Sarajevo (Sipka *). Following this a Language Conference was held prior to the Fourth Congress in Poland (1990). Language standardisation was the major theme of this Conference. A new alphabet (based on proposals by Cortiade, 1990) was adopted incorporating

< previous page

page_291

next page >

< previous page

page_292

next page > Page 292

morphophonemic principles and seeking to produce a standard spelling for Kalderash and Balkan dialects. It remains to be seen whether this will be used in practice. There was also at this conference a move away from loans from India towards using international words based on Latin and Greek, and a proposal to use computers to produce a dictionary of the most common words from the major dialects. Conclusions and discussion It will have become evident either that there are very few hard facts to be discovered in print about language change among Romani speakers, or perhaps there is literature on the subject that I have not found. Romani is at the crossroads. The language could die out everywhere. One or two dialects could become vehicles for literature and education, while others will wither away through neglect. Romani offers an open field for research and creative activity. It needs dictionaries, collections of oral literature and school textbooks based on at least regional standardizations. It has a unique role as the first Indian language to be spoken in Europe. I hope that most people will feel that the need is not so much for time, energy and money to be put into charting the death of Romani but into helping it live. References CORTIADE, M. (1990), Les latérales en romanes; La langue romani; The so called seven cases of Romani; Léxicologie rom. In: Materials for the Language Conference, Warsaw (published by the author). HANCOCK, I. (1975), Problems in the creation of a standard dialect of Romanes. Austin, Texas: SW Educational Development Laboratory. JUSUF, S. (1978), Amen sam e Titoske. Ljubljana: Univerzum. KENRICK, D. (1989), L'évolution d'une langue standard pour le romanes. Où en sommes-nous? In: P. Williams (ed.), Tsiganes: identité, évolution, Paris. KOCHANOWSKI, V. (1963), Gypsy Studies (2 pts). N. Delhi: Sata-Pitaca Series. MANUSH, L. (1990), Ramayana. Special number of Roma. Chandigarh: Roma Publications. MESSING, G. (1988), Glossary of Greek Romani. Columbus, Ohio: Slavica Inc.. SIPKA *, M. (ed.) (1989), International Symposium Romani Language and Culture (Papers and summaries in several languages). Sarajevo: Institut za Proucavanje* nacionalnih Odnosa. SORAVIA, G. (1977), Dialetti degli Zingari Italiani. Pisa: Pacini.

< previous page

page_292

next page >

< previous page

page_293

next page > Page 293

The journals Journal of the Gypsy Lore Society, Études Tsiganes, Lacio Drom (Rome) carry articles on the languages.

< previous page

page_293

next page >

< previous page

page_294

next page > Page 294

Appendix Major dialects of Romani in Europe: a tentative classification. International Kalderash and related dialects (e.g. Churari)

The Kalderash lived in Roumania from approximately 1500 to 1800 after which they migrated throughout the world. Over 500.000 speakers. Also lived in Roumania and migrated during the 19th century. There was also apparently a dialect switch from Carpathian to Lovari by many Romanies in Hungary during the 19th century. Some 20.000 speakers (excluding Lovari Hungary, see below). The Sinti and Manouche lived in the German Empire perhaps from 1550 onwards. Apart from Germany and Austria there are large numbers in Sinti and France and smaller numbers in Belgium, Manouche Holland and Italy. About 50.000 speakers. Reached West Europe around 1400. Settled in Finland, North Wales and Spain. The three dialects became mutually incomprehensible because of the large number of loan words and the phonological influences of the majority languages. Only survives in Finland Kalé and Sweden. About 6.000 speakers. Yugoslav Since 1960 groups speaking a variety of dialects (outside from Yugoslavia have migrated into Italy and Yugoslavia)Western Europe. Over 50.000 speakers. Romani Originally in England and France. Extinct in chal. England. Few if any speakers in France.

< previous page

page_294

next page >

< previous page

page_295

next page > Page 295

Territorial Finland (Kaale), Sweden 6.000 Two long established dialects in Italy Calabria and Abruzzo. 4.000 Two main dialects. Sedentary and nomadic (within Greece). Greece 50.000. Yugoslavia Three major dialect clusters: * Sedentary Christians of Serbia * Sedentary Muslims of Bosnia, Kosovo, Macedonia * Nomadic groups e.g. Burgudzia Between half and one million. Slovakia Czechoslovakia. 300.000 Hungary, South Slovakia, Poland, Carpathian (Romungro). USSR 30.000 Two dialects: Highland and Poland Lowland. 30.000 Bulgaria Over 30 dialects. 300.000 Albania Two dialects. 50.000 Roumania Several varieties. 500.000 Hungary Lovari. 250.000 Baltic states and European USSR 30.000

< previous page

page_295

next page >

< previous page

page_ii

next page >

< previous page

page_297

next page > Page 297

15 Methodological issues in language shift research Koen Jaspaert & Sjaak Kroon One of the consequences of (international) migration movements is the emergence of language contact situations. Language contact generally leads to changes in the language behaviour of the individuals and groups that are involved. At a certain point in time, in certain contexts, individual immigrants may opt for the language of the immigration country instead of their mother tongue. Such individual language choice processes, be it or not in combination with processes of first language loss (Andersen, 1982) or incomplete learning of the mother tongue (Gonzo & Saltarelli, 1983), at group level, over generations, are likely to lead to a situation of permanent language shift. Referring to the design and some of the results of a large scale research project in language shift and language loss that we have been working on in the last few years, we will mainly discuss some aspects of the methodology of language shift research and, additionally, formulate some suggestions for crosslinguistic research into language shift. The sociolinguistic research project which we carried out, is mainly focusing on the social conditions that influence the processes of language shift and language loss in Turkish and Italian immigrants in The Netherlands and Flanders, the Dutch speaking part of Belgium. In this project some 800 subjects were interviewed by means of a rather lengthy questionnaire. Questions were included on the choice of language in different situations, on the social, cultural and educational background of the subjects, on assumed language proficiency and attitudinal factors. To this questionnaire a number of language proficiency tests in Turkish and Italian were added. In this context we will limit ourselves to the language shift part of the project. As to the language loss part, we refer to, e.g., Jaspaert & Kroon (1989).

< previous page

page_297

next page >

< previous page

page_298

next page > Page 298

Defining language shift A first question to be answered is what we consider to be language shift, i.e., how language shift is to be operationalized. Although under the heading of language maintenance and language shift a multitude of different processes are investigated, some basic characteristics can be observed. Language shift usually refers to the gradual disappearance of a language in a community where it used to be spoken. The main distinction between different approaches lies in the choice of linguistic facts that are used as an index for this disappearance. The candidates for this position can be divided into two groups: language use and language proficiency. Our position in this discussion is that we refer to changes in language use as language shift, and to changes in language proficiency as language loss. The fact that we opt for language choice as the basis for language shift implies that we consider language use as the more fundamental element in the gradual disappearance of a language. This implies that we believe a diminishing language proficiency to be a consequence of the fact that a language loses (some of) its functions rather than vice versa. In order to explain the way in which we operationalized language shift in our research we will take a closer look at language contact situations. Imagine, for the sake of argument, a group that migrated to an area in which another language than their mother tongue is the dominant language and that stops using its own language in a period of three generations. The disappearance of the ethnic group language is really the outcome of a number of changes that have occurred in the ethnic community. Being confronted with another dominant language probably meant in the first place being confronted with speakers who are monolingual or at least do not speak the immigrant group's language. This implies that there is a problem when people of the immigrant group want to communicate with the speakers of the dominant language. This problem is typically one of coordination, i.e., the type of problem that instigates the creation of social norms (Ullmann-Margalit, 1977; Bartsch, 1985). In this context members of ethnic groups have different options which have, in turn, different chances of becoming the norm. We will discuss four different options. Option I The ethnic group chooses not to communicate with people who do not speak their language, thus avoiding the emergence of communication

< previous page

page_298

next page >

< previous page

page_299

next page > Page 299

problems. As far as intergroup contacts are unavoidable these can be structured in a way that only a small number of ethnic group members are involved in the communication process. An example of this option is the relative isolation of the Pennsylvania Dutch or German in the USA (Huffiness, 1980). An important factor here is the dominant group's reaction. As long as the idea exists that the ethnic group will one day return to its country of origin the option of isolation is likely to be favoured by the dominant group. Examples of this position are to be found in Western Europe with respect to the early groups of migrant workers (cf. Boos-Nünning & Hohmann, 1989 for a discussion of this policy in the Federal Republic of Germany). But as soon as groups have settled permanently in another language area, as is the case with the groups we are studying, the choice for abstinence of communication becomes an unlikely option for both the dominant group and the ethnic group. Option 2 The ethnic group tries to establish communication in its own language. This option, too, needs the consent of the dominant group to become the norm and it is clear that this consent will not be easily obtained. Although segregation or integration policies may lead a dominant group to accept to communicate with the minority group in the minority group language in a limited number of well defined situations, it is very unlikely that minority language use becomes the norm in less formalized surroundings. Option 3 Members of the two groups choose to communicate in a third language that they both know. Although in the intergroup communication between Italian immigrants and the Flemish majority in Flanders, French for some time has played this role of a third language, generally speaking this seems to be an exceptional choice since in most cases there will not be such a language available (Vanvolsem, Jaspaert & Kroon, 1991). Option 4 Given the limited chance for the options discussed so far to become the norm in intergroup communication, the most likely norm for communication with members of the dominant group is, of course, the dominant group language. If a member of the minority group wants to ask direc-

< previous page

page_299

next page >

< previous page

page_300

next page > Page 300

tions in the street from a member of the dominant group, or when he is asked directions from somebody from this group, the conversation will almost certainly take place in the dominant group language. For more formalized situations the dominant group language will also be the norm, except in a few instances as described above, where an exception to this norm is allowed. The above implies that when migration is followed by a more or less permanent settlement, and both sides choose for integration rather than segregation, members of the minority group almost unavoidably shift towards the use of the dominant language in most of their contacts with the dominant group. The extent of the shift process will be determined by the extent of the interethnic communication that is established. Returning to our previous example of the ethnic group losing its language in the course of three generations, it has to be remarked, of course, that this choice for the dominant language does not necessarily explain the shift that occurred in intragroup communication. As long as we are dealing with a group, i.e., as long as the group does not lose its group status, there is intragroup communication which does not necessarily follow the same norms as intergroup communication, and which therefore could go on taking place in the ethnic group language. Such a process would lead to a language situation in the community which could be described as a form of stable bilingualism: the minority language is used in intragroup communication and the majority language in all other instances (Fishman, 1972). Language death only occurs as a result of the disappearance of intragroup communication and this can normally only happen when the group itself dissolves due to demographic causes (Dorian, 1980). If the minority language disappears without group dissolvement, as in our example, this means that the members of the minority group have chosen to use the dominant language in intragroup communication. This form of shift is clearly different from the one in intergroup communication. The use of the dominant language in intragroup communication in general is no answer to a coordination problem. There might be instances in which coordination problems do occur, e.g., in the case of a large amount of linguistic diversity in the home country or a declining proficiency in the ethnic group language in the second generation which causes intragroup communication problems in the immigration situation. However, intragroup language shift is not primarily a question of the need to understand and the need to be

< previous page

page_300

next page >

< previous page

page_301

next page > Page 301

understood. It is not motivated by the need to establish communication but by the wish to establish communication in a certain way. The difference between the two forms of shift can be expressed in a different way. In the first case norms have to be established as a result of the new situation. Once they have been established, they do not change anymore. What does change is the situation to which they apply. If, for instance, a minority group family moves from an area predominantly populated by members of the minority group to an area where no other members live, the amount of use of the minority language will drop drastically. The norm, however, i.e., speak the minority language with minority group members, will not change. With the second form of shift, the shift process does not only imply the creation of new norms, but the adaptation of old ones. This change of norms is an essential part of the shift process. If in the example of the family moving to an area where no group members live, the situation changes again so that other group members come to live in that area, the use of the group language will rise again. But if in the meantime the norm itself has undergone change, the new migration may not lead to increased use of the group language for that family. In other words, the choice of language in interaction with group members should be considered the form of behaviour which is central to questions of language maintenance. As long as there is a minority group and, as long as this minority group is not demographically broken up, the use of the minority language will not disappear unless the norms for language use within the group are changed. Whereas the first form of shift creates a stable bilingual situation, the second form destabilizes such a situation to the possible extent that bilingualism may disappear altogether. This discussion on the nature of shift leads to the conclusion that in order to understand processes of language maintenance and language shift, it is imperative to study changes in language choice in intragroup communication. Only when we understand the mechanisms that govern these changes, will we be able to interpret correctly processes of maintenance and shift in demographically stable situations. In situations that are demographically unstable we will also need to study the demographic changes and the interaction between these changes and the normative changes.

< previous page

page_301

next page >

< previous page

page_302

next page > Page 302

Measuring language shift As to establishing the extent to which group members use the group language for intragroup communication, two different options seem to be of importance. One can either take the position that shift is determined by the frequency with which the group language is used. Any choice of the dominant language is then a next step in the process of shift, regardless of the situation in which it occurred. Or one can take the position that functionality of a language much more than frequency of its use, is a key concept in determining the importance of language choice for language shift. In the latter case the more important question becomes in which situations or domains which language is chosen (Fishman, 1970). The frequency of use of a language may change without the functionality changing. An example may clarify this point. Take, for instance, two individuals who interact with other members of their community in two domains. In domain 1 they almost always use the majority language, and in domain 2 they almost always use the minority language. One individual, however, interacts most of the time in domain 1, whereas the other hardly ever finds himself in situations that do not belong to domain 2. The functionality of the two languages in question is equal for both individuals, the frequency of use, however, differs considerably. The question now is whether one should measure the amount of group language used within a community in order to determine the scope of the shift process, or concentrate on how functional the language is within that community. Although most researchers in language shift research work with the domain concept (e.g., Pauwels, 1986), i.e., base themselves on measures of functionality, we decided to include both, functionality and frequency of use in our study. This means that on the one hand we gathered information on the language choice people make in a certain number of domains. On the other hand, we collected information on the relevance of these domains in intragroup communication, and we asked some general questions on the amount of group language use, without reference to domains. We distinguished eight domains: family, neighbourhood, clubs and meeting places, work, religion, shopping, friends and family outside the community, and other contacts. Of all these domains (and subdomains within these) the informant was asked whether he or she came into situations which belonged to that domain. When the informant affirmed this, he/she was asked whether he/she met group members in that situa-

< previous page

page_302

next page >

< previous page

page_303

next page > Page 303

tion. When this was again affirmed, the informant was requested to indicate on a five point scale whether he/she met in this situation: (1) (2) (3) (4) (5)

almost exclusively group members more group members than others group members as frequent as others more others than group members almost exclusively others.

Next the informant was asked which language he/she usually employed with group members in that situation. After having chosen from the possible alternatives (Dutch, Turkish/Italian, Italian dialect, other), the interviewer asked whether the interviewee sometimes used another language with group members in that situation. When that was the case, the informant was invited to tell something more about the circumstances in which this happened. The interviewer was instructed to listen to this story and pick out of it to which circumstances this occasional use related, thereby using four categories to structure these circumstances: (1) (2) (3) (4)

characteristics of the interaction partner in terms of generation persons present during the interaction the place of interaction the topic of conversation.

As to the frequency of use the informants were asked to generally consider their contacts with group members during the course of one week, and to indicate on a five point scale to what extent they used Italian with certain categories of group members. The five point scale consisted of the following possibilities: (1) (2) (3) (4) (5)

almost always Italian/Turkish more Italian/Turkish than Dutch an equal amount of Italian/Turkish and Dutch more Dutch than Italian/Turkish almost always Dutch.

The different groups the question was asked for were: (1) (2) (3) (4)

group members of the first generation group members of the second or third generation when group members of different generations are present when non-speakers of the group language are present.

< previous page

page_303

next page >

< previous page

page_304

next page > Page 304

Language shift data As to the results of the language shift part of our research we will limit ourselves here to some global shift figures on the Italian and the Turkish group respectively, without discussing the social correlates of these data (cf. Jaspaert & Kroon, 1991). The data show that in the Italian group, roughly speaking, intragroup communication takes place in Dutch in about 40% of the instances we recorded. The differences between the domains that were included appear to be lower than was expected: there is no domain that is especially resistant to Dutch and there are no domains that have almost completely been affected by shift. An interesting point is that the family is certainly not the stronghold of the ethnic minority language since communication with siblings is the situation most favourable to the use of Dutch (about 60%) and even for communication with the parents Dutch has become the usual language in about 40% of the cases. The domains that seem to be most resistant to Dutch are neighbourhood (27%) and church (19%). The figures concerning occasional use of Dutch in intragroup communication in the Italian group are on the whole rather low. Less than 20% is found to use Dutch occasionally. The few switches that occur are usually triggered by the characteristics of the persons present at the interaction, with the presence of non-speakers of Italian being the most important incentive. In the Turkish group the number of subjects claiming Dutch as usual language in intragroup communication is much smaller and in quite a number of domains even negligible (less than 10%). The number of subjects claiming occasional use of Dutch is much larger than in the case of the Italian group (almost 48%). Resemblances can be found in the order of Dutchification of domains. As is the case with the Italians, neighbourhood, church and visits are the domains least affected by shift. Sportsclubs, work and communication with siblings are the domains in which Dutch is used most often. If we compare the language shift data of the two groups, the resemblances in the order of domains as to the importance of use of Dutch suggest that we are dealing with a more general process of shift. In this process the two groups have clearly reached different stages. The Italian group has reached the stage in which Dutch is the usual language for an important part of the group. Within the Turkish group, there are hardly any individuals for whom Dutch is the usual language,

< previous page

page_304

next page >

< previous page

page_305

next page > Page 305

but an important part of the group occasionally uses Dutch. Splitting up the data of the two groups into tables for each of the specific places in The Netherlands and Flanders where data were gathered, shows that we are in fact not dealing with two homogeneous groups of Italians and Turks. Without going into much detail here, it is clear that different settlements have reached different stages in more or less the same process of language shift. Some elements of the way in which this process proceeds are clear reminders of what has been called the wave theory (Bailey, 1973). In a first step Dutch becomes the occasional language in some domains for a limited group of people. Next the occasional use of Dutch spreads to other domains and people, and at the same time a number of occasional users in a number of domains switch to Dutch as a usual language. This wave movement continues, turning certain domains into all Dutch domains, and gradually Dutchifying others. The final outcome, if the process keeps going, is the total Dutchification of the intragroup communication. In this sense language shift is only a special form of language change. These different stages in such a wave process can also be observed in the data related to frequency of use of Italian and Dutch. The frequency figures show that for communication within the ethnic group, Dutch is used with a larger frequency by the Italian group than by the Turkish. For both groups the minority language is used more often with members of the first generation than with members of other generations, and more when no Dutch speakers are present than when Dutch speakers are present. Explaining language shift On the basis of these figures two language shift measures were constructed, one for functionality and one for frequency. These measures were then used in a series of statistical analyses that aimed at establishing the social correlates of the shift observed. Neither the construction of the shift measures, nor their social correlates will be discussed (cf. Jaspaert & Kroon, 1991). They are mentioned here for the sole purpose of introducing the conclusion that statistical correlates between language variables such as shift on the one hand and social factors on the other are insufficient tools for the understanding of the processes underlying this form of language change. This type of correlational analysis falls within the tradition of the exploration of the importance of what is called distance variables influencing

< previous page

page_305

next page >

< previous page

page_306

next page > Page 306

language change (Extra & Vallen, 1989). Research has been done, for example, into the influence of communicative, social, economic, juridical, linguistic, cultural, religious and demographic distance. In this research, it is remarkable that these factors do not in all situations consistently work in the same direction. In one group of subjects the factor inter-ethnic marriage, for example, favours language shift, whereas the same factor in another group counteracts to these processes. Furthermore, in many research projects it is not at all clear to what extent these factors are related to each other, to what extent, in other words, they have individual explanatory power. At the moment it is generally accepted that the factors that were mentioned before, and possible other sociological factors, do not directly influence the processes of language choice and language shift, but exert their influence through intervening or intermediating variables or concepts (Appel & Muysken, 1987). To construct these concepts it is necessary to work on the basis of a more general theory of sociocultural contact and sociocultural change (Fishman, 1972). In this respect, several proposals have been put forward in the literature. Depending on the theoretical backgrounds of the researchers, central importance is given to concepts like attitudes (Giles, 1979), ethnolinguistic vitality (Giles, Bourhis & Taylor, 1977), social networks (Tandefelt, 1992), ethnic identity (Fishman, 1985), or symbolic exchanges (Jaspaert & Kroon, 1991). An important limitation of the models that have been proposed on the basis of these central notions is that in most cases they have not been empirically tested on a large scale with different groups of subjects. Even more importantly, they cannot be tested on the basis of (rather superficial) census data that are available (Fishman, 1992). In our study, groups are incorporated that differ among each other as to their immigration history, culture, religion, language background, and social, economic, juridical, and demographic position. We ultimately aim at testing the models that have been proposed. Conclusions One of the main insights that we gained from the research that we have been working on is, that the development and empirical validation of a theoretical, explanatory model of language choice and language shift cannot be limited to just studying two groups in a particular language area in which processes of language shift occur. We therefore decided

< previous page

page_306

next page >

< previous page

page_307

next page > Page 307

to expand our study to other ethnic minority groups in different circumstances, countries and language environments. Until now we managed to have data collected on language shift in Italian immigrants in London (n=100) and in Panjabi speaking Sikhs in Birmingham (n=100). The data of these and other groups will make crosslinguistic and crossnational analyses of language shift processes possible. The envisaged international and crosslinguistic expansion of our study seems to provide a good guarantee for the acquisition of a more profound insight in possible sociolinguistic consequences of language contact for the patterns of language use of individuals and groups involved. The acquisition of such insight is also relevant from a societal perspective, for example, in view of questions of the development of national and international policies in the field of language and education. Acknowledgements This research was partially supported by the Foundation for Linguistic Research, which is funded by the Netherlands organisation for research (NWO), and by the Belgian Fund for Fundamental Collective Research (FKFO). References ANDERSEN, R. (1982), Determining the linguistic attributes of language attrition. In: R. Lambert & B. Freed (eds.), The loss of language skills, Rowley, Mass.: Newbury House, 83-118. APPEL, R. & P. Muysken (1987), Language contact and bilingualism. London: Edward Arnold. BAILEY, C.-J. (1973), Variation and linguistic theory. Arlington, Va.: Center for Applied Linguistics. BARTSCH, R. (1985), Sprachnormen. Theorie und Praxis. Tübingen: Niemeyer. BOOS-NÜNNING, U. & M. Hohmann (1989), The educational situation of migrant workers' children in the Federal Republic of Germany. In: L. Eldering & J. Kloprogge (eds.), Different cultures same school: Ethnic minority children in Europe, Amsterdam: Swets & Zeitlinger, 39-59. DORIAN, N. (1980), Language shift in community and individual: The phenomenon of the Laggard Semi-speaker. In: International Journal of Sociology of Language 25, 85-94.

< previous page

page_307

next page >

< previous page

page_308

next page > Page 308

EXTRA, G. & T. Vallen (1989), Second language acquisition in elementary school: Cross-national perspective on the Netherlands, Flanders and the Federal Republic of Germany. In: L. Eldering & J. Kloprogge (eds.), Different cultures same school: Ethnic minority children in Europe, Amsterdam: Swets & Zeitlinger, 153-188. FISHMAN, J. (1970), Sociolinguistics. Rowley, Mass.: Newbury House. FISHMAN, J. (1972), Language maintenance and language shift as a field of inquiry: Revisited. In: J. Fishman, Language in sociocultural change, Stanford: Stanford University Press, 76-132. FISHMAN, J. (1985), The rise and fall of ethnic revival: Perspectives on language and ethnicity. Berlin: Mouton. FISHMAN, J. (1992), Conference summary. In: W. Fase, K. Jaspaert & S. Kroon (eds.), Maintenance and loss of minority languages, Amsterdam: John Benjamins, 395-408. GILES, H. (1979), Sociolinguistics and social psychology: An introductory essay. In: H. Giles &R. StClair (eds.), Language and social psychology, Oxford: Basil Blackwell, 1-20. GILES, H., R. Bourhis & D. Taylor (1977), Towards a theory of language in ethnic group relations. In: H. Giles (eds.), Language, ethnicity and intergroup relations, London: Academic Press, 307-348. GONZO, S. & M. Saltarelli (1983), Pidginization and linguistic change in emigrant languages. In: R. Andersen (ed.), Pidginization and creolization as language acquisition, Rowley, Mass.: Newbury House, 181-197. HUFFINESS, M. (1980), Pennsylvania German: Maintenance and shift. In: International Journal of Sociology of Language 25, 43-58. JASPAERT, K. & S. Kroon (1989), Social determinants of language loss. In: ITL-Review of Applied Linguistics, 83-84, 75-98. JASPAERT, K. & S. Kroon (1991), Social determinants of language shift by Italians in The Netherlands and Flanders. In: International Journal of Sociology of Language, 90, 77-96. (= K. de Bot & W. Fase eds., Migrant languages in Western Europe) PAUWELS, A. (1986), Immigrant dialects and language maintenance in Australia. The case of the Limburg and Swabian dialects. Dordrecht: Foris Publications. TANDEFELT, M. (1992), Some linguistic consequences of the shift from Swedish to Finnish in Finland. In: W. Fase, K. Jaspaert & S. Kroon (eds.), Maintenance and loss of minority languages, Amsterdam: John Benjamins, 149-168. ULLMAN-MARGALIT, E. (1977), The emergence of norms. Oxford: Claredon Press. VANVOLSEM, S., K. JASPAERT, & S. Kroon (1991), Erosione e perdita di lingua presso emigrati Italiani in Belgio e in Olanda: primi risultati. In: I. Baldelli & B. Da Rif (eds.), Lingua e Letteratura Italiana nel mondo oggi II, Firenze: Leo S. Olschki Editore, 677-691.

< previous page

page_308

next page >

< previous page

page_309

next page > Page 309

16 Lexical aspects of language attrition and shift Kees de Bot and Bert Weltens In earlier publications, we distinguished types of attrition research on the basis of two factors, namely whether the language studied is a native language (L1) or a non-native language (L2), and whether or not this language is the dominant variety in the speech community in question, i.e., whether the language under investigation is functioning as an L1 or as an L2 in the environment we are studying (De Bot & Weltens, 1985). This subdivision results in four main types of language attrition research (see Figure 1).

Figure 1 Four types of attrition research. Although this is a somewhat crude division of reality into four main categories that certainly needs further refinement on some points, we still consider it a useful tool for establishing what we are talking about. The present chapter is in principle concerned with type 2 research, i.e., the development of the native languages of migrants who moved to a country where these native languages are not the dominant language. Nevertheless, we will also mention some findings from research that falls into category 3, namely the attrition of second language skills after re-migration, and the attrition of foreign language skills learned in

< previous page

page_309

next page >

< previous page

page_310

next page > Page 310

school, as some of the findings are particularly relevant to our line of argument. Language attrition and shift In the sociolinguistic tradition, language maintenance and shift have been important issues in the last decades. Classical studies in this field are Kloss (1966) on German in the US, Clyne (1967) on German and Dutch in Australia, and Haugen (1969) on Norwegian in the US. In these studies specific groups of immigrants are treated in considerable detail both historically and linguistically. Typically, these early studies concern languages in contact that are typologically closely related. Only in recent years have less related pairs of languages been studied, such as Finnish and Serbo-croatian vs. Swedish (Boyd, 1986) and Gujarati vs. English in Britain (Linguistic Minorities Project, 1985). As far as indigenous minorities are concerned, there are a great number of possible relations between the dominant language and the minority language variety. Cheshire et al. (1989: 2-3), for example, distinguish four types: (1) the minority speaks a standard language that is the national language in a neighbouring country, e.g., French in Switzerland; (2) the minority speaks an adapted variety of a neighbouring standard language, e.g., Swiss German; (3) the minority speaks a dialect of a neighbouring standard language, e.g., the German dialect spoken in the French Alsace, or Arvanitika, the Albanian dialect spoken in Greece; and (4) the minority speaks a dialect of the national standard language. Although the dividing line between some of the types is a gradual rather than a discrete one, the list certainly shows that research into indigenous minorities is an area dealing with quite distinct types of multilingualism. Another distinction that divides this field is the emphasis lent to functional vs. structural changes in the language contact situation, i.e., whether one focusses on changes in the domain distribution of the two language varieties, or whether one focusses on changes in the linguistic systems in question. This dichotomy is quite clearly reflected in the recent volume on language obsolescence edited by Nancy Dorian (1989); she ordered the contributions under two headings, namely ''focus on context" and "focus on structure" respectively. A rare

< previous page

page_310

next page >

< previous page

page_311

next page > Page 311

example of a study that explicitly incorporated both aspects is Münstermann & Hagen's investigation of dialect loss in the Dutch town of Maastricht (see Münstermann & Hagen, 1986; Münstermann, 1989). Quite a lot of the research on indigenous minorities deals with language varieties that are about to disappear (language death; for a recent overview, see Dorian, 1989). There are clear differences between language shift in migrant groups and language death, the most important being that for the migrant language there is a source of new input, both in terms of newly arriving migrants and in terms of on-going processes of language change in the country of origin. Joshua Fishman and his colleagues report on the situation in New York, where the level of bilingualism is constant in the Spanish group, which appears to be the result of a constant flow of new immigrants rather than the maintenance of two languages within individuals (Fishman et al., 1968). In recent years it has become clear that the processes of language shift and language attrition should be treated independently. In their description of the sociolinguistic situation of migrant languages, Gonzo & Saltarelli (1983) suggest that it is a combination of the two processes which determines the fate of migrant languages: in the first generation there is attrition within the individual members of the group, partly through non-use, partly through interference from the dominant contact language. The resulting reduced version of the language is then the model for the second generation learners of the language. Due to a number of factors, the second generation acquires only part of the reduced variety. Within the second generation there is attrition again, and consequently reduced transmission of the migrant language to the next generation. Gonzo & Saltarelli claim that this cascade-type of process leads to the disappearance of the migrant language within three or four generations. There is, however, not very much empirical evidence to support such a model. In their 1986 study on L1 acquisition in an L2 environment, Verhoeven & Boeschoten present what seems to be a different point of view. They argue that attrition should be distinguished from what they call "stagnation". This distinction is graphically represented in Figure 2. Verhoeven & Boeschoten reserve the term attrition for an actual decrease in knowledge and/or performance, as shown in Figure 2c, and introduce the term "stagnation" for the development shown in Figure 2b. Their argument is that the acquisition of L1 does continue in an L2 submersion environment, but at a slower rate than is the case with

< previous page

page_311

next page >

< previous page

page_312

next page > Page 312

children living in a monolingual L1 community. Their data confirm that what occurs is indeed "stagnation" rather than attrition, but we have to remember that their subjects were 4- to 8-year-old children. Our contention is that Figure 2b represents what we have labelled the acquisition of a reduced model; in other words, this is the partial shift that has occurred between the first and second generations. Figure 2c, then, shows what will happen to thesealready reducedskills in L1 in later life. Interpreted in this way, the perspective they present is only seemingly different.

Figure 2 Potential L1 acquisition patterns in an L2 submersion environment (from: Verhoeven & Boeschoten, 1986: 242). Lexical attrition and shift There are several publications that mention aspects of the linguistic system that are relatively more 'vulnerable' in the process of attrition or shift than others (e.g., Andersen, 1982). Vocabulary is supposed to be lost fairly easily, as are syntactic aspects of a language which are acquired relatively late. At any rate, the notion of 'vocabulary' needs to be specified further. Even if we distinguish between productive and receptive know-

< previous page

page_312

next page >

< previous page

page_313

next page > Page 313

ledge of a word, a distinction we will return to later, there are many aspects to 'knowing a word'. An attempt at listing the relevant aspects is made by Nation (1990), which is reproduced in Table 1. Table 1 Aspects of word knowledge (from Nation, 1990: 31; P = productive; R = receptive). Form Spoken form Written form

RWhat does the word sound like? P How is the word pronounced? RWhat does the word look like? P How is the word written and spelled?

Position Grammatical patterns Collocations

RIn what patterns does the word occur? P In what patterns must we use the word? RWhat words or types of words can be expected before or after the P word? What words or types of words must we use with this word?

Function Frequency Appropriateness

RHow common is the word? P How often should the word be used? RWhere would we expect to meet this word? P Where can this word be used?

Meaning Concept Associations

RWhat P What RWhat P What

does the word mean? word should be used to express this meaning? other words does this word make us think of? other words could we use instead of this one?

As Cohen (1986: 146) pointed out, "just as there is a lot to learn in learning words, there is much that can be forgotten in the forgetting process". In other words, talking about vocabulary per se is a gross oversimplification. When we look at research into the attrition of foreign language skills, vocabulary knowledge is typically defined as knowing the meaning of words, as becomes clear from the examples in Table 2. An

< previous page

page_313

next page >

< previous page

page_314

next page > Page 314

exception to this rule is the work by Trapman (in prep.), who investigated the attrition of case marking in German. He found that this syntactic aspect of lexical items was lost, without the lexical items themselves being affected. Table 2 Examples of foreign language attrition studies into word knowledge Bahrick (1984) recall: 'llamar'.... recognition: 'feliz' 1 happy, 2 fault, 3 feet, 4 new, 5 clean Weltens (1989) J'ai nettoyé l'évier. Ik heb de ..... schoongemaakt. [I have cleaned the .....] On the other hand, foreign language attrition on the lexical level is defined quite differently in studies such as Olshtain & Barzilay (1991) and Weltens & Grendel (1989). In the latter case, a pilot experiment is reported on, in which the lexical decision paradigm was used for testing French word knowledge in Dutch students. In a lexical decision experiment letter strings are presented one by one, and the subjects have to indicate for each of those letter strings, as quickly as they can, whether it forms an existing word in a particular language or not. In other words, vocabulary knowledge was defined in this particular investigation as knowledge of the orthographic representation of words, orto use Nation's terminology as shown in Table 1the "written form" of the words. Strictly speaking, since the reaction times were used as the principal data, it was defined as the speed with which this information could be retrieved. The importance of such a different definition of word knowledge becomes clear when we compare, on the one hand, the results obtained by Weltens (1989) by means of traditional vocabulary tests (in which word meanings were required as answers), and the reaction time data obtained in a lexical decision experiment. The traditional vocabulary tests suggested that word knowledge in Dutch learners of French was virtually unaffected over a period of four years following course completion, whereas the lexical decision experiment showed that, already after two years, the accessibility of French word forms was slowed down significantly (cf. Weltens & Grendel, 1989).

< previous page

page_314

next page >

< previous page

page_315

next page > Page 315

In a recent study, Hakuta & d'Andrea (1992) investigated changes in the vocabulary of Spanish speaking secondary school students in the US. In this (cross-sectional) study they looked both at attrition by comparing groups that differed in length of residence in the US, and at shift by comparing groups with US-born and foreign-born parents respectively. The results of this study support the model proposed by Gonzo & Saltarelli: there is some attrition within generations, and considerable shift between generations. Another interesting aspect of this study was that part of the population was tested using two experimental techniques: picture naming and word-picture matching. The first task is typically a retrieval task, while the second is a recognition task. The data from this experiment show that retrieval is affected much more severely by processes of attrition or shift than recognition is. Similar findings have been reported for the attrition of second language skills after re-migration. Cohen (1989), for instance, reports that his two children, who were fluent in Portuguese after living in Brazil for one year, showed progressive stages of attrition in their productive lexicon, nine months after their departure from Brazil. In an oral comprehension task, however, they were able to "readily identify these words" (p. 147). All this supports one of Andersen's (1982) hypotheses: 'active' vocabulary is lost more easily than 'passive' vocabulary. This also confirms a hypothesis put forward in earlier research on language attrition (e.g., Weltens, 1989) that (lack of) availability rather than loss of lexical information is the central issue in most cases of attrition. In accordance with current thinking in memory research, it is assumed that knowledge is not lost once it has been acquired. On the other hand, it is plausible that non-use of a language leads to reduction of availability of the knowledge of that language. Several researchers have tried to find attrition of vocabulary by looking at spontaneous speech (Cohen, 1986; Olshtain, 1986). Although they did find differences in measures such as the number and types of words used in story retelling tasks when comparing an attrition group with a control group, it is not clear how these differences should be interpreted. It does not necessarily follow that all aspects of the words in question have been erased from memoryand as we have just argued, this is in fact highly unlikely. It may simply be the case that speakers tend to avoid words that are 'hard to find'. Research on lexical changes in processes of attrition and shift apparently needs to be done with tasks that imply some sort of time pressure, and that are ideally

< previous page

page_315

next page >

< previous page

page_316

next page > Page 316

'on-line'. It has become clear in research on language attrition, that language 'attriters' are very skillful in using all sorts of compensatory strategies in order to avoid problems in their language use. Only in well-controlled on-line experiments can informants of this type be forced to apply rules that they are uncertain about. Let us take a concrete example. If a particular word does not appear in a given speech sample, what does that mean? If we look back at Nation's analysis of word knowledge, this might mean (1) that the speaker was unable to link the meaning he wanted to express to a word form in the target language; (2) that the speaker did find a word form, but was unable to determine whether it was the right word in this context and therefore did not utter it; or (3) that the speaker found the word, knew it was appropriate in the context, but was unable to pronounce it. And finally, it may have been the case that the search process took so long that the information did not become available in time, with the result that the speaker had to express his idea in quite another way. In other words, spontaneous speech data can be useful to suggest possible problem areas, but they are rather problematic if we try to determine where exactly in the language system attrition or shift has occurred. The organization of the bilingual lexicon If loss of lexical skills can be evidenced, this has to be accounted for in terms of a psychological model of bilingualism. One of the questions of interest is to what extent this is reflected in a change in the organization of the bilingual lexicon, in other words, how the items of different languages are stored in memory. In an overview, Paradis (1981) examined four hypotheses with respect to the bilingual lexicon (see Figure 3). According to the extended system hypothesis, languages are undifferentiated in their representation, and elements of the various languages are processed as 'allo-elements', i.e., different surface manifestations of the same underlying concept. The dual system hypothesis states that each language is represented independently, while the tripartite system hypothesis proposes that those items that are identical in both languages are represented in one single underlying neural substratum common to both languages, and those that are different each have their own, separate representation.

< previous page

page_316

next page >

< previous page

page_317

next page > Page 317

Figure 3

< previous page

page_317

next page >

< previous page

page_318

next page > Page 318

Finally, there is the subset hypothesis, which assumes the use of a single storage system where links between elements are strengthened through continued use. This implies that, in general, elements from one language will be more strongly linked to each other than to elements from another language. The result is the formation of subsets which appear to consist of elements from the same language, and which can be retrieved separately. At the same time links between elements in different languages will be just as strong as links between elements in one language in bilingual speakers who employ a 'code switching mode', and who live in a community where code switching is a normal conversational strategy. There is quite some research to support the subset hypothesis. In a number of investigations different variants of the lexical decision paradigm have been used in order to assess the semantic relationship between words, both within and between languages. One of these variants is the so-called priming technique, in which pairs of words are presented to subjects; the first word, the prime, is either semantically related or unrelated to the second word, the target. Subjects have to indicate as quickly as they can whether the target is an existing word or not. If the prime is related to the target, as in the case of 'doctor' and 'nurse', the response time will be reduced. The difference in response time between targets preceded by unrelated primes and targets preceded by related primes, is called the priming effect There appear to be consistent priming effects between and within languages, but, interestingly enough, the magnitude of the effect differs depending on the direction of the priming effect. For instance, Kerkman (1984) found a difference in priming between his two language conditions English (primes)-Dutch (targets) and Dutch-English. Comparable effects are reported by Kirsner et al. (1984) and Brown et al. (1984). This difference in the size of the priming effect suggests that the amount of activation spreading is not a characteristic of a given pair of words, but rather that it is dependent on the direction of the activation spreading. This supports the notion of different subsets that can be activated. In research on the bilingual lexicon it is not always clear what is actually stored. In recent psycholinguistic models of the language user (cf. Levelt, 1989), lexical items consist of two parts: the lemma, which contains a meaning component and a syntactic component, and the lexeme, which contains morphological and phonological characteristics

< previous page

page_318

next page >

< previous page

page_319

next page > Page 319

of the item. It is not inconceivable that this type of decomposable lexical items play a role in both production and perception. As yet, it is not completely clear how decomposable a lexical item is. Research on language production suggests that the meaning and syntax part of the lemma function as a whole, as do the morphological and phonological part of the lexeme. The connection between the lemma and the lexeme appears to be less strong. For theories on the bilingual lexicon we have to take into account the possibility that lemma and lexeme are not always represented and stored in the same way for different languages. In research in this field, quite some experiments were carried out in which different types of cognates were investigated (Caramazza & Brones, 1979; Cristoffanini, Kirsner & Milech, 1986; Kerkman & De Bot, 1989; De Groot & Nas, 1991). This research clearly shows that there is a difference in storage between words that are the same in form and meaning (English 'hand' and Dutch 'hand') vs. words that have the same form and a different meaning (English 'list' and Dutch 'list'). The work by Kerkman in particular supports the view that only real cognates (i.e., words with the same form and meaning) share one single entry in the mental lexicon. For so-called 'false friends', such as 'list/list', it may well be the case that there are two lemmas, one for each language, while there is one shared lexeme. Apparently we have to look at two levels in the organization of the lexicon: a lemma level and a lexeme level. Lemmas in two languages may differ with respect to the meaning part and with respect to the syntactic part. When a Dutch/ German cognate like 'Antwort/antwoord' is retrieved, different lemmas may be called upon because of differences in syntax (gender), despite the great similarity in form. Usually these different lemmas will be connected with different lexemes, but this is not necessarily the case. When two lexical items differ only syntactically, then a reference to the same lexeme is conceivable. In the same way as ambiguous words in a language are connected with the same form from different lemmas, reference to cognates may take place from language-specific lemmas to forms that are not language-specific. The theoretical considerations given have important implications for research on lexical attrition. In fact it is quite meaningless to test (lack of) lexical knowledge by simple tests such as naming an object or a picture. Taking together the different aspects of a word that have to be acquired and, consequently, may be lost (Nation, 1990), and the psycholinguistic modelling presented above that has to be used as a framework for fundamental research, the various aspects of lexical

< previous page

page_319

next page >

< previous page

page_320

next page > Page 320

knowledge and representation have to be studied on the basis of different measuring techniques. Perspectives We will discuss two recent projects which attempt at incorporating these considerations. A project already well underway is the one carried out by Marjon Grendel at the University of Nijmegen. She investigates orthographic and semantic aspects of word knowledge in Dutch students and former students of French by means of lexical decision tasks. An additional interesting aspect of this project is that it not only investigates attrition of knowledge, but also reactivation of that knowledge (cf. Grendel, 1990). Another project, carried out by El Aissati & Schaufeli from 1990-1994, concerns the attrition of lexical skills in first and second generation migrants from Turkey and Morocco. The central research question in this project is: to what extent can attrition of lexical skills be evidenced in first and second generation adult informants in a setting in which the first (migrant) language has to compete with a higher prestige language. More specifically, the main research questions are: a. to what extent is language attrition or shift reflected in a decrease in availability of lexical information? b. to what extent is language attrition or shift reflected in a change in the organization of the bilingual lexicon? The emphasis will be on lexical skills for two reasons. The first reason is that the experimental techniques used in the lexical experiments have proven their relevance and effectiveness in past research; the second reason is that attrition is more likely to be found with lexical skills than with any other skills. With respect to type of loss, both attrition and shift will be investigated. Attrition will be studied by comparing informants from the same generation, in particular migrants from the first generation, while shift will be studied by comparing informants from different generations, in particular first and second generation. This means, that we will not look at attrition within the second generation, which would be problematic for several other reasons. With respect to type of activity, both production and perception data will be gathered. Perception data will be gathered using lexical

< previous page

page_320

next page >

< previous page

page_321

next page > Page 321

decision techniques, while production data will be gathered using a lexical substitution task (Cross, 1981). Given what we have argued before, we have good hopes that these investigations will make a substantial contribution towards our understanding of the processes of language attrition and shift. References ANDERSEN, R. (1982), Determining the linguistic attributes of language attrition. In: R. Lambert & B. Freed (eds.), The loss of language skills, Rowley, Mass.: Newbury House, 83-118. BAHRICK, H. (1984), Fifty years of second language attrition: Implications for programmatic research. In: Modern Language Journal 68, 105-118. BOT, K. DE & B. Weltens (1985), Taalverlies: Beschrijven versus verklaren. Handelingen van het het 38e Nederlandse Filologencongres, 51-61. BOYD, S. (1986), Using the present to predict the future in language contact: The case of immigrant minority languages in Sweden. In: B. Weltens, K. de Bot & T. van Els (eds.), Language attrition in progress, Dordrecht/Providence: Foris, 99-115. BROWN, H., N. Sharma & K. Kirsner (1984), The role of script and phonology in lexical representation. In: The Quarterly Journal of Experimental Psychology 36A, 491-505. CARAMAZZA, A. & I. Brones (1979), Lexical access in bilinguals. In: Bulletin of the Psychonomic Society 13, 212-214. CHESHIRE, J., V. Edwards, H. Münstermann & B. Weltens (eds.) (1989), Dialect and education: Some European perspectives. Clevedon: Multilingual Matters. CLYNE, M. (1967), Transference and triggering. The Hague: Martinus Nijhoff. COHEN, A. (1986), Forgetting foreign-language vocabulary. In: B. Weltens, K. de Bot & T. van Els (eds.), Language attrition in progress, Dordrecht/Providence: Foris, 143-158. COHEN, A. (1989), Attrition in the productive lexicon of two Portuguese third language speakers. In: Studies in Second Language Acquisition 11, 135-149. CRISTOFFANINI, P., K. Kirsner & D. Milech (1986), Bilingual lexical representation: The status of Spanish-English cognates. In: The Quarterly Journal of Experimental Psychology 38A, 367-393. CROSS, D. (1981), Tests of substitution: A simple alternative to cloze. In: System 9, 51-53. DORIAN, N. (1989), Language obsolescence: Studies in language contraction and death. Cambridge: Cambridge University Press. FISHMAN, J., R. Cooper & R. Ma (1968), Bilingualism in the Barrio. Bloomington: Indiana University Press. GONZO, S. & M. Saltarelli (1983), Pidginization and linguistic change in emigrant languages. In: R. Andersen (ed.), Pidginization and creolization as language acquisition, Rowley, Mass.: Newbury House, 181-197. GRENDEL, M. (1990), Verlies en herstel van Franse woordkennis. In: Gramma 14, 2, 127-137.

< previous page

page_321

next page >

< previous page

page_322

next page > Page 322

GROOT, A. DE & G. Nas (1991), The lexical representation of cognates and non-cognates in bilinguals. In: Journal of Memory and Language 1, 90-123. HAKUTA, K. & D. d'Andrea (1992), Some properties of bilingual maintenance and loss in Mexican background high-school students. In: Applied Linguistic 13, 1, 72-99. HAUGEN, E. (1969), The Norwegian language in America: A study in bilingual behavior. Bloomington: Indiana University Press. KERKMAN, H. (1984), Woordherkenning in twee talen. In: A. Thomassen, L. Noordman & P. Eling (eds.), Het leesproces, Lisse: Swets & Zeitlinger, 139-152. KERKMAN, H. & K. de Bot (1989), De organisatie van bet tweetalige lexicon. In: Toegepaste Taalwetenschap in Artikelen 34, 115-121. KIRSNER, K., M. Smith, R. Lockhart, M. King & M. Jain (1984), The bilingual lexicon: Language-specific units in an integrated network. In: Journal of Verbal Learning and Verbal Behavior 23, 519-539. KLOSS, H. (1966), German-American language maintenance efforts. In: J. Fishman (ed.), Language loyalty in the United States, The Hague: Mouton, 207-251. LEVELT, W. (1989), Speaking: From intention to articulation. Cambridge, Mass.: MIT Press. LINGUISTIC MINORITIES PROJECT (1985), The other languages of England. London: Routledge and Kegan Paul. MÜNSTERMANN, H. (1989), Dialect loss in Maastricht Attitudes, functions and structures. In: K. Deprez (ed.), Language and intergroup relations, Dordrecht/Providence: Foris, 99-128. MÜNSTERMANN, H. & T. Hagen (1986), Functional and structural aspects of language loss: A research plan and some first results. In: B. Weltens, K. de Bot & T. van Els (eds.), Language attrition in progress, Dordrecht/Providence: Foris, 75-96. NATION, I. (1990), Teaching and learning vocabulary. London: Newbury House. OLSHTAIN, E. (1986), The attrition of English as a second language with speakers of Hebrew. In: B. Weltens, K. de Bot & T. van Els (eds.), Language attrition in progress, Dordrecht/Providence: Foris, 187-204. OLSHTAIN, E. & M. Barzilay (1991), Lexical retrieval difficulties in adult language attrition. In: H. Seliger & R. Vago (eds.), First language attrition: Structural and theoretical perspectives, Cambridge: Cambridge University Press, 139-150. PARADIS, M. (1981), Neurolinguistic organization of a bilingual's two languages. In: J. Copeland & P. Davis (eds.), The seventh LACUS forum, Columbia, SC: Hornbeam Press, 420-431. TRAPMAN, H. (in prep.), Taalverwerving en taalverlies: Een onderzoek naar regressieverschijnselen in het Duits. The University of Nijmegen (PhD. Thesis). VERHOEVEN, L. & H. Boeschoten (1986), First language acquisition in a second language submersion environment. In: Applied Psycholinguistics 7, 241-256. WELTENS, B. (1989), The attrition of French as a foreign language. Dordrecht/ Providence: Foris. WELTENS, B. & M. Grendel (1989), Taalverlies en woordherkenning in het Frans als vreemde taal. In: Toegepaste Taalwetenschap in Artikelen 34, 76-80.

< previous page

page_322

next page >

< previous page

page_323

next page > Page 323

AUTHORS AND AFFILIATIONS Jeroen Aarssen Tilburg University Dept of Language and Minorities P.O. Box 90153 5000 LE Tilburg The Netherlands Safder Alladina The Polytechnic of North London School of Teaching Studies Prince of Wales Road London NW5 3LB Great Britain Paula Andersson University of Göteborg Dept. of Linguistics Renströmparken 41298 Göteborg Sweden Ad Backus Tilburg University Dept. of Language and Minorities P.O. Box 90153 5000 LE Tilburg The Netherlands Kees de Bot Nijmegen University Dept. of Applied Linguistics P.O. Box 9103 6500 HD Nijmegen The Netherlands

< previous page

page_323

next page >

< previous page

page_324

next page > Page 324

Sally Boyd Stockholm University Center for Research on International Migration and Ethnic Relations 10691 Stockholm Sweden Guus Extra Tilburg University Dept. of Language and Minorities P.O. Box 90153 5000 LE Tilburg The Netherlands Koen Jaspaert Leuven University Dept. of Migration Studies Blijde Inkomststraat 7 3000 Leuven Belgium Lars Johanson J. Gutenberg Universität Seminar für Orientkunde P.O. Box 3980 6500 Mainz Germany Donald Kenrick 61 Blenheim Cresc. London W 11 2EG Great Britain Sjaak Kroon Tilburg University Dept. of Language and Minorities P.O. Box 90153 5000 LE Tilburg The Netherlands

< previous page

page_324

next page >

< previous page

page_325

next page > Page 325

Jarmo Lainio University of Stockholm Dept. of Finnish 10691 Stockholm Sweden Sirkku Latomaa University of Helsinki Dept. of Finnish Sovsapuisto 1 I 137 14 33500 Tampère Finland Jacomien Nortier Vlieland 158 3524 AD Utrecht The Netherlands Andrina Pavlinic University of Zagreb Inst. for Migration and Nationalities Trnjanska b.b. P.O. Box 88 41000 Zagreb Croatia Carol Pfaff FU Berlin JFK Institut Lansstrasse 7-9 1000 Berlin Germany Jan Jaap de Ruiter Tilburg University Dept. of Language and Minorities P.O. Box 90153 5000 LE Tilburg The Netherlands

< previous page

page_325

next page >

< previous page

page_326 Page 326

Anneli Schaufeli Nijmegen University Dept. of Applied Linguistics P.O. Box 9103 6500 HD Nijmegen The Netherlands Ludo Verhoeven Tilburg University Dept. of Language and Minorities P.O. Box 90153 5000 LE Tilburg The Netherlands Bert Weltens Nijmegen University Dept. of Applied Linguistics P.O. Box 9103 6500 HD Nijmegen The Netherlands

< previous page

page_326