Narrative Development in a Multilingual Context (Studies in Bilingualism)

Narrative Development in a Multilingual Context Studies in Bilingualism (SiBil) Editors Kees de Bot University of ...

Author: Ludo Verhoeven | Sven Stromqvist

100 downloads 1042 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Narrative Development in a Multilingual Context

Studies in Bilingualism (SiBil)

Editors Kees de Bot University of Nijmegen

Thom Huebner San José State University

Editorial Board Michael Clyne, Monash University Kathryn Davis, University of Hawaii at Manoa Joshua Fishman, Yeshiva University François Grosjean, Université de Neuchâtel Wolfgang Klein, Max Planck Institut für Psycholinguistik Georges Lüdi, University of Basel Christina Bratt Paulston, University of Pittsburgh Suzanne Romaine, Merton College, Oxford Merrill Swain, Ontario Institute for Studies in Education Richard Tucker, Carnegie Mellon University

Volume 23 Narrative Development in a Multilingual Context Edited by Ludo Verhoeven and Sven Strömqvist

Narrative Development in a Multilingual Context

Edited by Ludo Verhoeven, University of Nijmegen Sven Strömqvist, University of Lund

John Benjamins Publishing Company Amsterdam/Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Narrative development in a multilingual context / edited by Ludo Verhoeven, Sven Strömqvist. p. cm. (Studies in Bilingualism, issn 0928–1533 ; v. 23) Includes bibliographical references and index. 1. Language acquisition. 2. Discourse analysis, Narrative. 3. Multilingualism in children. I. Verhoeven, Ludo Th. II. Strömqvist, Sven, 1954- III. Series. P118.N367 2001 401’.93--dc21 isbn 90 272 41341 (Eur.) / 1 58811 0982 (US) (Hb; alk. paper)

2001037887

© 2001 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Contents

1. Development of narrative production in a multilingual context Ludo Verhoeven and Sven Strömqvist

1

2. Temporality and language contact in narratives by children bilingual in Norwegian and English Elizabeth Lanza

15

3. Reference continuation in L2 narratives of Turkish adolescents in Norway Ingvild Nistov

51

4. Age-related and L2-related features in bilingual narrative development in Sweden Åke Viberg

87

5. Sociocultural aspects of bilingual narrative development in Sweden Sally Boyd and Kerstin Nauclér

129

6. The development of co-constructed narratives by Turkish children in Germany Carol W. Pfaff

153

7. Influence of L1 Turkish on L2 French narratives Mehmet-Ali Akinci, Harriet Jisa and Sophie Kern 8. Development of temporal relations in narratives by Turkish–Dutch bilingual children Jeroen Aarssen 9. Temporality issues in Moroccan Arabic and Dutch Petra Bos 10. Bilingual narrative development in Papiamento and Dutch Ria Severing and Ludo Verhoeven

189

209 233 255

vi

Contents

11. Linguistic features of Spanish-Hebrew children’s narratives Judy Kupersmitt and Ruth A. Berman

277

12. Narrative development in Hebrew and English Dorit Kaufman

319

13. Trilingual children narrating in Hebrew, English and Spanish Anat Stavans

341

14. Logic and mind in Spanish–English children’s narratives Barbara Zurer Pearson

373

15. From affect to language: development of evaluation in narratives in spoken English and American Sign Language Judy Reilly

399

16. Narrative development in multilingual contexts: A cross-linguistic perspective Ruth A. Berman

419

Index

429

Acknowledgements

The purpose of this volume is to open new perspectives in the study of narrative development in a multilingual context by bringing together research findings from psychology, linguistics, and education. The editors would like to thank the Dutch Science Foundation (NOW) and the Faculty of Social Sciences of the University of Nijmegen for their support of the initiative for this volume. The editors also wish to thank all contributors to this volume for their cooperation and Karin van der Weijden-van der Laan for her editorial support of this volume. Ludo Verhoeven Sven Strömqvist

Chapter 1

Development of narrative production in a multilingual context Ludo Verhoeven and Sven Strömqvist

The language patterns of children living in a multilingual context can be quite complex. Most bilingual children belong to a minority group and learn their two languages in a successive manner: They learn the ethnic group language in the home and immediate community; the second language gradually enters their lives via television, contact with peers, and occasionally daycare. When these same children enter school, the language input is almost exclusively L2. Such minority children thus take part in lessons in the second language to some extent but to a greater extent acquire the language naturally through interactions with teachers and peers. Studies of first language acquisition have made clear that children typically command the grammatical principles and rules governing their native language by the age of four years (see, for example, Slobin 1985; Radford 1995). Several studies have also shown language development to continue into the school years at both the levels of grammar and language-use preferences (Karmiloff-Smith 1992; Nippold 1998; Strömqvist to appear). The later language development of children can be characterized by a growing command of discourse and according to Karmiloff-Smith (1986), discourse can be seen as the most significant domain of later language acquisition. Around the age of five years, critical developmental shifts take place from the use of intrato inter-sentential devices for discourse purposes, and from basic structures to additional functions. With respect to the bilingual development of school-age children, it is still unclear what sorts of operating principles they use to separate the two languages. The conditions under which processes of language transfer typically occur also remain unclear. The studies conducted so far have been rather limited in their scope in that the languages under consideration were highly related (cf. Grosjean 1982; McLaughlin 1985; de Houwer 1995). It can therefore



Ludo Verhoeven and Sven Strömqvist

be expected that data from children learning two typologically unrelated languages will provide a new perspective on the role of structural linguistic properties in bilingual language development (see Extra and Verhoeven 1994, 1998). In the present volume, the narrative development of children in a multilingual environment will be explored. The process of bilingual development will be studied from an interdisciplinary point of view, combining insights from linguistic theory and developmental theory. According to a Chomskian nativist perspective, bilingual development is an ‘‘instantaneous’’ process when it occurs in an ideal situation placing all of the principles and parameters of universal grammar at the child’s disposal along with the two sets of input data necessary to fix the relevant parameters for each language. Given the obvious fact that languages in general are not acquired instantaneously, developmental theory must explain the various ‘‘delays’’ found to characterize children’s first and second language development. Any theory of language acquisition covering the school age must also address the developing pragmatic competence of children in addition to their expanding grammatical competence. The analyses in the present volume are thus focused on the organizing processes used by children to cover spans of connected utterances. The narrative, in its broader sense, is taken to be the relevant form of extended discourse.

The study of narrative structure Personal experience narratives can be defined as a method of ’’recapitulating past experience by matching a verbal sequence of clauses to the sequence of events which (it is referred) actually occurred’’ (Labov 1972: 359–60). Hopper (1979) has argued that narratives tend to divide into two sections: the foreground or sequence of completed events defining the plot or narrative line and the background comprising the remainder of the narrative which introduces elements of the plot , supplements elements of the plot, or concludes the plot. The foreground can also be regarded as the reason for telling the narrative. From a variety of studies, it has also become clear that the thematic structure of a narrative is usually marked linguistically (Chafe 1980; Francik 1985), and that listeners/readers use thematic markers to organize their comprehension of a narrative (Kintsch and van Dijk 1978; Morrow 1985). There is clear evidence showing the manner of representing the major characters in a narrative and the anchoring of the narrative plot in time to be crucial to its organization. With respect to the representation of main charac-

Development of narrative production in a multilingual context

ters the continuation of the narrative topic has been studied extensively (see Hickman 1995). Protagonists can be described in various ways: using a full noun phrase, a reduced noun phrase, or zero-marking. On the basis of crosslinguistic research, Givón (1983) has concluded that the three aforementioned coding devices express increasing levels of topic continuity in the order mentioned. Karmiloff-Smith (1981, 1985) has shown anaphoric pronominalization to be the default marking of the thematic subject across a span of sentences, with thematic subject defined as the preferential preemption of the slot for reference to the main character. When referring to the main protagonist in a narrative, pronominalization may also not occur for a number of reasons (Maes 1991). After a protagonist has been (re)introduced, subsequent full identifying descriptions may serve to establish the character more fully in the mind of the listener (Clancy 1980). A nominal expression can also be used to mark the onset of a new episode in the narrative (Marslen-Wilson, et al. 1982). Furthermore, an identifying description may be used when a referent has not been mentioned for some time as the repeated mention of a character can help reactivate the relevant concept in the listener. The need to reintroduce referents thus depends on such factors as the number of candidates for reference and the number of previous pronominal expressions for the same referent (Givón 1983; Redeker 1987). To explore linguistic reference to time, three fundamental categories of temporality can be distinguished: basic temporal relations, aspect, and internal temporal features. Tense is a deictic category and, in narratives, temporal relations are used to, among other things, anchor events to the narrative time line. Aspect refers to the various perspectives that can be taken on an event (e.g., a perfective vs. imperfective perspective). Internal temporal features refer to the quasi-objective temporal characteristics of an event, such as durativity or transformativity. Tense and aspect oppositions in narratives not only serve to locate events relative to the moment of speech but also organize the structure of a narrative. Schiffrin (1981) and Fleischman (1985) have both shown the use of the present tense to refer to past events (historical present) and the use of the past tense to be regularly alternated in narratives. The conclusion of these authors is that the organization of a narrative delimits the area in which the historical present can occur and that various structural and functional constraints determine the switches between the two tenses. Wallace (1982) has shown the present tense (as opposed to nonpresent tenses) and perfective aspect (as opposed to imperfective aspect ) to mark the main points or foreground knowledge in a narrative.






In their ground-breaking volume, Relating Events in Narrative (1994), Berman and Slobin show different linguistic ways of relating events to exist and to include various means for narrative encoding and emplotment. The basis for this research is the re-telling of the so-called Frog story, a series of pictures representing a number of dynamic interactions between animate beings over time and in different physical settings by different aged narrators speaking different languages. Throughout this cross-linguistic study of narrative development, filtering and packaging were taken to be the guiding principles. The filtering principle holds that experiences are filtered into verbalized events via the choice of perspective taken on the experiences and the set of linguistic operations provided by the language in question. The packaging principle holds that a narrative does not involve a linear chain of successive events but a hierarchical set of events located in time and space with various event phases subordinated and interrelated. Within this typologically cross-linguistic framework, the developmental relations between form and function were explored with respect to the following functional categories: temporality, event conflation, perspective, connectivity, and narrative style. Across the different functional categories, the grammatical forms that are accessible and obligatory within a language were primarily found to channel the attention of the learner towards particular functions. That is, frequently used forms direct learner attention to the functions of those forms. Finally, the acquisition of a form to express a particular function was also found to provide an opening wedge for acquisition of the more advanced functions associated with that form.

Development of narrative construction In the process of narrative development, children must learn that linguistic forms are multifunctional and that the use of any particular form is multiply defined. Berman (1997) has explored what preschool children know about language structure and language use. By the age of five, children are adept at combining clauses and have mastered a great deal of complex syntax; they can also construct sequentially well-organized narratives and express different perspectives on events. In other research reviewing a large number of crosslinguistic studies on discourse organization in the domains of person, space, and time, Hickman (1995) concluded that the progression in narrative organization involving interrelations between the sentence and discourse levels is


relatively late. The shift from exophora to endophora appears to be the main process acquired by school-aged children to displace reference and, more generally, use discourse as a new kind of context to anchor utterances when necessary. Several studies have addressed the question of how topic continuity in narratives is acquired. Bamberg (1985, 1986) examined how German-speaking children in the age range of 3 to 10 years establish reference to two main characters in the beforementioned Frog story. He found that the youngest children, as well as some of the children in the middle age range followed a global anaphoric strategy by matching the main protagonist of the story with the third person pronoun, irrespective of whether the reference to the character is maintained or the character is reintroduced. At a more advanced stage of development, a more adult-like anaphoric strategy is followed by the children with nominal expressions used to reintroduce characters and pronouns used to maintain the reference to characters. According to Bamberg, the reorganization of the linguistic devices used by children for text cohesion involves a shift from a global system referring to the text as a whole to a more local system taking both the text as a whole and the immediate discourse environment into account. Karmiloff-Smith (1981, 1985) investigated the acquisition of discourse devices in English and French with children four to nine years of age and found a developmental sequence of three stages. During stage 1, nominal referential devices are used with their deictic function for even the first mention of a referent. When protagonists are first referred to with an indefinite pronoun, they are pronominalized afterwards. During stage 2, new referents are introduced with indefinite referring expressions, pronouns function anaphorically, and the subject slot in all sentences is used for reference to the main protagonist only. During the third stage, the subject slot is no longer rigidly reserved for reference to only the main protagonist. In addition to the rules for pronominalization in connected discourse, children must also learn the contrasts between definite and indefinite reference. When the speaker does expect entities to be identifiable, noun phrases will be interpreted as being definite. However, as shown by Karmiloff-Smith (1986), the basic referential markers in many languages are plurifunctional, which means that children must learn to map the right forms and functions to make the right distinctions for reference tracking. With respect to the acquisition of the linguistic means for temporal reference, Weist (1986) has shown the opposition between present/nonpresent to develop first and the past tense marking to be acquired in primarily






redundant contexts. Von Stutterheim (1986) studied the acquisition of temporality in L2 German by adult Turkish workers and found tense to initially indicate aspectual distinctions and deictic distinctions only later. The repertoire of temporal expressions was gradually differentiated with respect to temporal distance. Dietrich, Klein, and Noyau (1995) report on a crosslinguistic study of the acquisition of temporality by adult immigrants in Europe. Learners start without explicit marking of structural relations, such as government, and without grammatical marking of temporality. Early attempts to mark temporality primarily involve the localization of events in time. Among the various interacting ways to make temporal constellations clear, pragmatic devices precede lexical devices, which precede grammatical devices. The structure of the acquisition process turned out to be very similar irrespective of the source language and the target language although considerable differences in the speed of development and its final success were observed. The development of narrative construction cannot be understood without considering its psycholinguistic and communicative constraints. As Berman and Slobin (1994) have argued, younger children may employ fewer expressive options during narrative construction simply because they cannot conceive of the full range of encodable perspectives from a cognitive point of view, they cannot fully assess the listener’s viewpoint from a communicative point of view, and they cannot apply the full range of formal devices from a linguistic point of view. In a similar vein, Strömqvist and Day (1993) have observed very different developmental patterns for child first and adult second language learners on a narrative task. Given the complex nature of narrative construction, the role of contextual factors in narrative development can be viewed as very important. From research on parent input to young children, we know that such interactive activities as storybook reading greatly influence children’s narrative development (see Sulzby and Teale 1991). Conditions that highlight the relevance and purpose of story telling for learners also appear to be quite important for the development of narrative comprehension and production. The most important facilitator of narrative development turns out to be the extent to which parents are sensitive to their children’s conversational attempts and the endeavors of parents to extend the conversation with such attempts as the starting point (see Snow et al. 1991). The semantic contingency of adult speech is thus a critical factor. Semantically contingent utterances include expansions limited to the content of previous child utterances, semantic extensions adding new information to the topic, questions demanding


clarification of child utterances, and answers to child questions (Snow 1995). Along these lines, Bus and van IJzendoorn (1997) have shown children 12 to 15 months of age to develop the referencing behavior supported by their mothers. The infants’ responses to the pictures in books as denotative symbols gained significance through responding together with the mother. With respect to the development of narrative construction in a multilingual context, the complicated sociolinguistic position of minority groups with respect to literacy practices should be recognized (Durgunoglu and Verhoeven 1998). Minority groups may use a variety of language codes serving at least partially distinct sets of functions. Whereas the code with the highest status will primarily be used in societal institutions, the code of the minority language will generally be used for intragroup communication and expressing one’s ethnicity. In addition, yet another language code may be used for religious identification. Verhoeven (1991) showed different socio-cultural factors to be responsible for the patterns of language development in ethnic communities. In a study of six-year-old Turkish children in The Netherlands, predictor measures related to the child, his or her family, and the educational care received by the child were examined along with grammatical and pragmatic first- and second-language proficiency measures. Grammatical proficiency in L1 was related to the extent of caretaker interaction in L1 and the cultural orientation of the children and their parents. Pragmatic proficiency in L1 was related to the children’s cultural orientation, the presence of L1-speaking peers, and parent involvement in educational care. Grammatical proficiency in L2 was related to the child’s cognitive capacity, the period of educational guidance, the presence of L2-speaking peers, the family interaction in L2, and the extent of parent involvement in educational care. L2 pragmatic proficiency was related to the children’s cultural orientation, family interaction in L2, and the extent of caretaker L2 interaction. Leseman (1994) and Leseman and De Jong (2000) examined the effects of socio-cultural context and styles of parent-child storybook reading in particular on home literacy and found the development of literacy to be determined by opportunities for literacy-related interactions, the kinds of guidance and informal instruction provided, and the affective experiences accompanying the literacy interactions. Comparison of the effects of literacy practices on children’s literacy learning in different communities in The Netherlands also clearly showed the differences between families and their children’s literacy learning to primarily arise from the sociocultural context: particularly from the everyday cultural and religious practices of the family and the content of the parental employment. In many publica-






tions the disadvantaged position of minorities in literacy has been highlighted (see, e.g., Au 1993; Durgunoglu and Verhoeven 1998), showing that the literacy achievement gap between cultural groups within a society can be largely explained in terms of linguistic and cultural practices in conflict with mainstream educational practices.

The present volume In this volume, the results of a number of empirical studies of the development of narrative construction within a multilingual context will be presented and discussed. It will be explored what operating principles underlie the process of narrative production in L1 and L2. Developmental relations between form and function will be studied across a broad range of functional categories, such as temporality, perspective, connectivity, and narrative coherence. Previous studies of bilingual development give no reason to believe that the processes of language acquisition in bilinguals and monolinguals differ with regard to their basic features. The essential difference is that bilingual children receive two sets of linguistic input and thus have the additional task of distinguishing the two language systems. The complex process of separating two linguistic systems is poorly understood, however. The sorts of operating principles children use to acquire two languages at the same time are not at all clear. Studies of the later language development of bilingual children are scarce, and the development of narrative structure among bilingual children has only been studied in a very limited number of studies (see Verhoeven 1993, 1994, 1998). Most of the relevant studies to date are fairly limited in their scope, moreover, because the languages under study were highly related. In the present volume, a variety of language contact situations will be considered with broad variation in the typological distances between the languages in order to enable cross-linguistic comparison. The analysis of learner data in various cross-linguistic settings may thus offer new information on the role of the structural properties of unrelated languages on the process of narrative acquisition. An attempt is also made to find out how transfer from one language to the other is facilitated. The conditions promoting language transfer are still unclear (cf. Verhoeven 1994). Given the fact that children construct narratives in the two languages by filtering their experiences into verbalized events through one and the same perspective, it is hypothesized that


the acquisition of a form to express a particular function in one language may facilitate the acquisition of a form expressing that function in another language. Finally, the effects of input on narrative construction in children’s first and second language will be examined in several studies. There is reason to believe that parental language use and interactional styles influence the L1 and L2 narrative development in the child. However, the exact role played by parental input is as yet far from clear. The present volume opens with several contributions on bilingual development from different countries in Northwest Europe. Elizabeth Lanza examines temporality in the narratives of Norwegian- and English-speaking children living in Norway. In order to gain greater insight into issues of language contact, she compares monolingual and bilingual narrative data. Ingvild Nistov examines the development of topic continuity in the L2 narratives of Turkish adolescents living in Norway. The focus of the study is on the marking of reference continuation in short narratives written by the second language learners. Two additional chapters address the development of narrative production among various immigrants in Sweden. Åke Viberg explores a number of age-related features of L2 narrative development. His focus is on the devices used to map forms to functions in the domains of lexical processing, clause combining, and temporal relations. Sally Boyd and Kerstin Nauclér examine the sociocultural aspects of bilingual narrative development in Turkish children living in Sweden. Of particular interest is the role of parent and teacher input on the children’s narrative construction. In the next chapter, Carol Pfaff considers how interactions with adult interlocutors appear to influence the construction of narratives in Turkish and German by young bilingual children attending a preschool in Berlin. She also goes into the strategies that adult interlocutors and older siblings use in conversations with the children as they progress from day care to elementary school. In the following chapter, Mehmet-Ali Akinci, Harriet Jisa, and Sophie Kern address the bilingual development of school-aged Turkish children in France. In a comparison of the narratives of monolingual French and bilingual FrenchTurkish children, they explore the influence of L1 Turkish on the children’s L2 French narratives. Two additional chapters focus on the acquisition of temporality in the narrative production of immigrant children in The Netherlands: Jeroen Aarssen examines the narratives of children acquiring Turkish and Dutch; Petra Bos examines the narratives of children acquiring Moroccan Arabic and Dutch.






In a subsequent series of chapters, the development of narrative construction in countries outside of Europe will be considered. Ria Severing and Ludo Verhoeven investigate the development of narrative construction in Papiamentu- and Dutch-speaking children in the former Dutch colony of Curaçao. Of primary interest are the coherence and cohesion aspects of the children’s narratives. In the next chapter, the development of narrative production by Spanish- and Hebrew-speaking children in Israel will be reported on. Judy Kupersmitt and Ruth Berman examine how bilingual children perform the task of telling a story in each of their two languages and to what extent knowledge of linguistic structure develops in tandem with an ability to use the language for narrative construction within a discourse context. Another study involving Hebrew is reported on by Dorit Kaufman, who describes the narrative development of Hebrew- and English-speaking children living in the USA. The bilingual children’s L1 narration is compared to that of monolingual speakers of Hebrew and English. The study of Hebrew in relation to other languages is completed with a contribution by Anat Stavans who explores the narratives of trilingual children speaking Hebrew, English, and Spanish in Israel. The development of the children’s narrative constructions is examined as a function of both language-related issues and language-contact issues. In the following chapter, Barbara Pearson analyzes the bilingual narrative development of Spanish-English speaking children in the USA. Of particular interest is the expression of logical relations and metacognitive knowledge in children’s narratives as a function of developmental growth. In addition, the syntactic devices used by children to express the content of story characters’ thoughts are examined. The final contribution is by Judy Reilly on the development of evaluation in the spoken narratives of normal hearing children and the sign language narratives of deaf children in the USA. Her focus is on the transition from the early utilization of emotional expressions to convey evaluation by preschoolers to the use of lexicalization by school-aged children. The volume is completed with an epilogue by a highly respected expert in the field of narrative development, Ruth Berman. On the basis of the crosslinguistic studies presented in the volume, a number of operating principles underlying the acquisition of narrative construction within a multilingual context is considered. In addition, Berman presents a new and challenging perspective on the developmental study of language production across genres, modalities, and language-contact situations.


References Au, K. (1993). Literacy instruction in multicultural settings. Forth Worth, TX: Harcourt Brace Janovich. Bamberg, M. (1985). Form and function in the construction of narratives. Unpublished doctoral dissertation, University of California, Berkeley. Bamberg, M. (1986). A functional approach to the acquisition of anaphoric relationships. Linguistics, 24, 227–84. Bamberg, M. (1997). Preschool knowledge of language: What five-year-olds know about language structure and language use. In: C. Pontecorvo (ed.), Writing development. Amsterdam: John Benjamins, 61–76. Bamberg, M., and D. I. Slobin (1994). Relating events in narrative. A cross-linguistic developmental study. Hillsdale, NL: Lawrence Erlbaum. Bus, A. G. and M. H. van Ijzendoorn (1997). Affective dimensions of mother-infant picturebook reading. Journal of School Psychology, 35, 47–60. Chafe, W. L. (1980). The Pear Stories: Cognitive, cultural and linguistic aspects of narrative production. Norwood, NJ: Ablex. Clancy, P. (1980). Referential choice in English and Japanese narrative discourse. In: W. L. Chafe (ed.), The Pear Stories: Cognitive, cultural and linguistic aspects of narrative production. Norwood, NJ: Ablex. de Houwer, A. (1995). Bilingual language acquisition. In: P. Fletcher and B. MacWhinney (eds.), Handbook of child language. Cambridge, MA: Basil Blackwell., 219–50. Dietrich, R., Klein, W. and C. Noyau (1995). The acquisition of temporality in a second language. Amsterdam: John Benjamins. Durgunoglu, A. and L. Verhoeven (1998). Literacy development in a multilingual context. Mahwah, NJ: Lawrence Erlbaum. Extra, G. and L. Verhoeven (1994). The cross-linguistic study of bilingual development. Amsterdam: North Holland. Extra, G. and L. Verhoeven (eds.) (1998). Bilingualism and migration. Studies on language acquisition.. Berlin/New York: Mouton de Gruyter. Fleischmann, S. (1985). Discourse functions of tense-aspect oppositions in narrative: Toward a theory of grounding. Linguistics, 23, 851–82. Francik, E. P. (1985). Referential choice and focus of attention in narrative. Unpublished doctoral dissertation, Stanford University. Givón, T. (1983). Topic continuity in discourse: An introduction. In: T. Givón (ed.), Topic continuity in discourse: A quantitative cross-linguistic study. Amsterdam: John Benjamins. Grosjean, F. (1982). Life with two languages. Cambridge: Harvard University Press. Hickman, M. (1995). Discourse organization and the development of reference to person, space and time. In: P. Fletcher and B. MacWhinney (eds.), Handbook of child language. Cambridge, MA: Basil Blackwell. 194–218. Hopper, P. J. (1979). Aspects and foregrounding in discourse. In: T. Givón (ed.), Syntax and Semantics: Vol. 12. Discourse and Syntax. New York: Academic 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. (1985). Language and cognitive processes from a developmental perspective. Language and Cognitive Processes, 1, 1, 61–85. Karmiloff-Smith, A. (1986). From meta-processes to conscious access: Evidence from children’s metalinguistic and repair data. Cognition, 23, 95–147. Karmiloff-Smith, A. (1992). Beyond modularity: a developmental perspective on cognitive science. Cambridge, MA: MIT Press. Kintsch, W. and T. van Dijk (1978). Toward a model of discourse comprehension and discourse production. Psychological Review, 85, 363–94. Labov, W. (1972). Language in the inner city. Philadelphia, PA: University of Pennsylvania Press. Leseman, P. (1994). Sociocultural determinants of literacy development. In: L. Verhoeven (ed.), Functional literacy: Theoretical issues and educational implications. Amsterdam: John Benjamins, 163–84. Leseman, P., and De Jong, P. (2000). How important is home literacy for acquiring literacy in school? In: L. Verhoeven and C. Snow (eds.), Literacy and motivation. Mahwah, NJ: Lawrence Erlbaum. McLaughlin, B. (1985). Second Language Acquisition in Childhood. Hillsdale, N. J.: LEA. Maes, F. (1991). Nominal anaphors and the coherence of discourse. Unpublished doctoral dissertation. University of Tilburg, The Netherlands. Marslen-Wilson, W., E. Levy and L. Tyler (1982). Introduction and maintenance of reference. In: R. J. Jarvella and W. Klein (eds.), Speech, Place and Action. Chichester. Morrow, D. G. (1985). Prominent characters and events organize narrative understanding. Journal of Memory and Language, 24, 304–19. Nipplod, M. A. (1998). Later language development. Austin, Texas: Pro-ed. Radford, A. (1995). Phrase structure and functional categories. In: P. Fletcher and B. MacWhinney (eds.), Handbook of child language. Cambridge, MA: Basil Blackwell, 483–507 Redeker, G. (1987), Language use in informal narratives. TILL-paper, 105. Tilburg: University Press. Schiffrin, D. (1981). Tense variation in narratives. Language, 57, 1, 45–63. Slobin, D. (1985). Crosslinguistic evidence of the language-making capacity, In: D. Slobin (ed.) The Crosslinguistic Study of Language Acquisition Vol. 2. Hillsdale, NJ: Lawrence Erlbaum, 1157–256. Snow, C., W. Barnes, J. Chandler, I. Goodman and L. Hemphill (1991). Unfulfilled expectations: Home and school influences on literacy. Boston: Harvard University Press. Strömqvist, S. (to appear) Language acquisition in early childhood. In: G. Rickheit, T. Herrmann and W. Deutsch (eds.), International Handbook of Psycholinguistics. Berlin: De Gruyter. Strömqvist, S. and S. Day (1993). On the development of narrative structure in child L1 and adult L2 acquisition, Applied Psycholingistics, 14, 135–58. von Stutterheim, C. (1986). Der Ausdruck der Temporalitat im ungesteuerten Spracherwerb. Berlin: De Gruyter.


Sulzby, E. and W. Teale (1991). Emergent literacy. In: R. Barr and D. Pearson (eds.), Handbook of reading research. New York: Longman, 727–57. Verhoeven, L. (1991). Predicting minority children’s bilingual proficiency: Child, family and institutional factors. Language Learning, 41, 205–33. Verhoeven, L. (1993). Acquisition of narrative skills in a bilingual context. In: B. Ketteman and W. Wieden (eds.), Current issues in European second language research. Tübingen: Gunter Narr, 307–23. Verhoeven, L. (1994). Transfer in bilingual development: the linguistic interdependency hypothesis revisited. Language Learning, 44, 381–415. Wallace, S. (1982). Figure and ground: The interrelationships of linguistic categories. In: P. Hopper (ed.), Tense-aspect: Between semantics and pragmatics. Amsterdam: Benjamins. Weist, R. (1986). Tense and aspect. In: P. Fletcher and M. Garman (eds.), Language Acquisition. Cambridge: University Press.



Chapter 2

Temporality and language contact in narratives by children bilingual in Norwegian and English Elizabeth Lanza

Recent work on the simultaneous acquisition of two languages, or bilingual first language acquisition, stresses the notion that bilingual children develop two separate morpho-syntactic systems, just like the monolingual cohorts of their respective languages (Meisel 1990; De Houwer 1990; Paradis and Genesee 1996). Although the focus in this research has been on the emergence of functional categories, we may ask how the bilingual child’s two languages develop and potentially interact once ‘‘primary language acquisition’’ (Genesee 1989) has taken place, that is, beyond the first few years of language development. The expression of temporal relations in connected discourse provides a focused area in which to explore these issues. Children acquiring two languages from birth have to learn how to express time in each of their languages by employing the conventional grammatical and lexical means available for both of the languages. An excellent means for investigating temporality in bilingual children’s speech, and thus secure comparable types of data in both languages, is the elicited narrative. The instrument that has proved very successful in eliciting such data from a crosslinguistic perspective is the picture story book, Frog, Where are you? by Mercer Mayer — a book without a text that involves many temporal sequences (see Berman and Slobin 1994). This chapter presents some analyses from a project investigating the development of morphology within a discourse perspective among children bilingual in Norwegian and English, as well as among monolingual Norwegian children, living in Norway. The Norwegian data were collected in order to secure a basis for comparing the bilingual children’s development with monolingual children; however, they are also part of an inter-Nordic study of language acquisition (see Strömqvist et al., 1995).



Elizabeth Lanza

The plan of this chapter is to (1) give an overview of the important issues in work on the simultaneous acquisition of two languages; (2) present some narrative data from three bilingual children, stories that were produced under somewhat different conditions; and finally, (3) discuss the children’s use of temporal expressions in each of their languages and interpret what this use might say about their processing and acquisition of both languages. Some data from monolingual children and from Norwegian adults who are second language speakers of English will be presented for the purpose of comparison. Hence, my focus will be on individual variation in expressions of temporality in narratives within a bilingual context in order to highlight certain language contact issues that should be given more attention in bilingual first language acquisition studies.

Issues in bilingual first language acquisition De Houwer (1990) points out that the most important issues in the study of bilingual first language acquisition are: (1) whether these children operate with one or two linguistic systems from the beginning; (2) to what extent the child’s language development is similar to what happens in monolingual development; and (3) what role the input plays, that is, how important input is for explaining bilingual development. Although these questions pertained to early development, they still hold for the child’s continuing bilingual development. My focus will be on the first issue; however, the issue of one system or two is not really the issue at play, in the sense that what is of interest is how the two linguistic systems interact. This shift in focus from a static to a more dynamic view of bilingualism is relevant not only for studies of early bilingual first language acquisition, but also for more recent work in bilingual memory research among adults. As Hummel (1993) in her review article points out, there has been a shift in research among bilingual adults from interest in storage issues (e.g., whether there is one lexicon or two) to processing issues. Ransdell and Fischler (1987) suggested that the independence/interdependence dichotomy is simplistic and proposed that attention focus instead on the particular conditions under which two languages interact. In focusing on the interaction between the bilingual child’s two languages, we may ask: What are the factors that influence the bilingual child’s language choices when rendering two accounts of the same story? What factors come

Norwegian–English children’s narratives

into play in the child’s processing of the two languages in the rendition of the stories? What are the linguistic factors, that is, the language-specific choices that are to be made? Norwegian and English are two closely related languages; however, they do differ in subtle ways which makes them interesting to examine as language pairs. Language dominance will play a role in this discussion. Although we will find some lexical/grammatical indicators of language dominance, we may ask whether there are any indicators at a more subtle level, for example, in the child’s expression of temporal relations in connected discourse. An examination of the interaction between the two linguistic systems will also necessitate a consideration of the second issue brought up by de Houwer, that is, a comparison with the language use of monolingual children. I will argue that in studies of the simultaneous acquisition of two languages, it is time that we consider how these young bilingual children’s language processing is different, albeit normal, from their monolingual peers.

The present study Participants The data that I present on bilingual development come from three children at different age levels. There are indeed individual differences in language acquisition, and hence these data are used for suggesting issues to be explored in future research. The children are: 1. Kaja — 4;6 2. Alice — 6;10 3. Nina — 8;8 Each child comes from a home in which the mother is American and the father Norwegian. The mother speaks mostly English to the child and the father mostly Norwegian. My reason for focusing on these three children is that they were active bilinguals. In my data collection, several of the children exposed to both languages in the home could readily comprehend English but preferred speaking Norwegian. When placed in an interaction in which the interviewer would attempt to negotiate a ‘‘monolingual’’ situation with the child (cf. Lanza 1992, 1997), these children could produce some English utterances. Kaja, Alice and Nina, however, readily interacted in English, the minority language in this case.





Elizabeth Lanza

Data collection Kaja’s and Nina’s stories about the search for the frog were collected under the same conditions: their parents were asked to each tell the story separately to the child the evening before I and my assistant would come to ask the child to retell the story in each language. This input from the parents was also recorded. The parents were told to tell the child the story in the manner in which they normally read books to their child. The parents did not collaborate on the stories. In both families, the resulting stories were to various degrees co-constructed by parent and child. There are several reasons for exposing the child to the story before requesting a retelling: (1) input is an important variable in bilingual language acquisition, and this allowed me to capture some of that input, including the type of storytelling strategies to which the child was exposed. Bamberg (1987) in his study with German children also tapped on the children’s input, and familiarized the children in this way with the story; (2) I wanted the stories to be retellings in order to make the situation more natural. Attempts were made to get the child to sit opposite the interviewer; (3) the very fact that the child is to tell the story from the same book in two languages means the child will be familiar with the story before the second telling, even if he/she has not been exposed to the book before the interview situation. In other words, it is impossible for the child to produce two stories with an ‘‘unfamiliar’’ text when each story is based on the same text. My underlying assumption is that there is a greater difference between familiarity and non-familiarity than between various degrees of familiarity. Also an implicit assumption in collecting narratives like this from bilinguals is that language is the only variable at play in two different language retellings. We do not know what kind of variation there is in monolingual children’s retellings of the same story (however, see Strömqvist 1996). The third child, Alice, had been interviewed in a preschool situation along with some other monolingual Norwegian children who were exposed to Eastern Norwegian (at the time Norwegian children began school at the age of 7). She had seen and told the frog story in both English and Norwegian about a year prior to the new recordings which are the objects of analysis in this chapter. At the beginning of the new recordings (at age 6;10) she was allowed to refresh her memory by looking silently through the book. She said that she remembered it, but that there were many things she had forgotten. Alice was a very verbal child.


All three children retold the story first in English and then in Norwegian. For those receiving parental input, the mothers had told the story first in English then the fathers in Norwegian. The decision to have the retelling of the story in English first was due to considerations of the children’s dominance in Norwegian, which I will discuss in light of their stories. If traces of Norwegian lexical items or structure were found in their English stories after they had told them in Norwegian, it might be difficult to ascribe this influence to language dominance as the order of telling may have had an impact. This relationship might be easier to discern if the child first tells the story in her non-dominant language. The question then is of course whether the non-dominant language has any influence on the dominant language in the second retelling. Hence the children retold the story in Norwegian immediately following a conversation in English during which they had told the story in English. Given the conversational embedding of the stories, the different language versions of the story were thus not told back-to-back. Kaja used the book in retelling her stories. Alice also used the book for her retellings but after each narrative with the book, she also retold the story from memory, a task she was eager to do. Nina used the book to retell the English story, but insisted on telling the Norwegian story from memory. See Table 1. Table 1. The bilingual children’s retellings Kaja (4;6)

Alice (6;10)

Nina (8;8) X

English

Book Memory

X

X X

Norwegian

Book Memory

X

X X

X

Since I was in possession of these stories retold from memory, I decided to use them to investigate yet another dimension for bilingual language processing.

Results Story structure Although the focus in this chapter is on temporality, it is important to first consider whether the stories told in each language by the child are similar, and thus comparable, in both length and structure, especially in regards to those of





Elizabeth Lanza

other speakers of their two languages. Hence in the following presentation of the children’s stories, I will first give a general overview of their story structure. Comparisons will be made with the stories of three Norwegian children aged 6;7 to 7;1 (two boys and a girl) who were exposed to the story before their retelling in a preschool setting, similar to Alice’s input situation. The bilingual children’s data will also be compared to the stories of four Norwegian adults (two women and two men). As English is a compulsory subject in Norwegian schools, all Norwegians possess various degrees of fluency in this language. The adults interviewed were graduate students in English at university level, and hence highly proficient in English, particularly American English. They too were interviewed in English by me and in Norwegian by a Norwegian assistant. Reference will also be made to the monolingual child and adult data reported on in Berman and Slobin (1994). The focus will be on the bilingual children’s stories; some general features of language dominance in their stories will also be presented. With a discussion of story structure and language dominance as a backdrop, the temporal relations in the bilingual children’s stories will be examined and compared with the other data. Table 2 illustrates the number of clauses each bilingual child produced in her stories. A clause is identified here according to the definition by Berman and Slobin (1994: 660): ‘‘any unit that contains a unified predicate. By unified is meant a predicate that expresses a single situation (activity, event, state). Predicates include finite and nonfinite verbs as well as predicate adjectives’’. In Table 2 we see developmental differences: the older the child gets, the longer the narratives become. Apart from Nina’s retellings, the Norwegian versions are longer than the English ones. Nina’s Norwegian story was told from memory, a factor that may have affected the story length. The greater length of the Norwegian stories can be related to the child’s preference for Norwegian; however, we could also feasibly ascribe it to the fact that the child is more familiar with the story. For Alice she has told it three times when she finally tells it in Norwegian by memory (although it was told at the end of the

Table 2. Number of clauses in the bilingual children’s narratives Kaja

Alice

Nina

Book

English Norwegian

38 51

55 60

79 –

Memory

English Norwegian

– –

56 73

– 75


conversation: approximately 10 minutes after she had told the story from the book). However, it is interesting to point out that Strömqvist (1996) found that second narratives were usually shorter than first narratives among the monolingual population in his study. Hence language preference may indeed be a strong motivation for longer second stories among the bilingual children. Kaja’s story length approximately corresponds with the mean for 5-year-old English-speaking children and German-speaking children respectively (German children had longer stories than the English speakers) as reported on by Berman and Slobin (1994: 31). Nina’s story in English exceeds the mean and the range for the 9-year-old English speakers and fits within the range but exceeds the mean for the German-speaking children of that age. Seven-yearolds are not reported on in Berman and Slobin (1994); however, we may note that Alice’s stories fill an intermediate position between those of Kaja and Nina. Let us now examine the length of the narratives produced by the monolingual Norwegian children, who approximate the age of the bilingual child Alice. Recall that the monolingual children had also been exposed to the story before the retelling in a preschool setting similar to that of Alice’s. As with the bilingual children, these children also either told the story with reference to the book, or by memory. The story length for the first two children, a boy and a girl, are similar to Alice’s stories, as noted in Table 3. The third child’s story, told from memory, however, is shorter than those of the other children. This may be related to individual differences; however, the memory factor may have played a role, especially since the child had first been told the story two days prior to his retelling from memory. Two of the children’s stories, however, do fit in the same range between the story lengths of Kaja and Nina, similar to those of Alice. The Norwegian adult second language speakers of English had considerably longer stories in both languages with more clauses in the Norwegian versions than in the English versions (apart from one speaker), as illustrated in Table 4. Table 3. Number of clauses in the monolingual Norwegian children’s narratives 6–7-year olds

Clauses

NorwCH 1B

6;7

NorwCH 2G NorwCH 3B

7;0 7;1

Book Memory Book Memory

(NorwCH 1B = Norwegian child number 1, boy, etc.)

60 61 61 46





Elizabeth Lanza

Table 4. Number of clauses in the adults’ narratives NorAD 1F NorAD 2F NorAD 3M NorAD 4M

English

Norwegian

67 60 74 85

95 53 97 102

(NorAd 1F = Norwegian adult number 1, female)

As for the complexity of the stories, Ragnarsdóttir (1991), based on the work of Berman and Slobin and their colleagues, lists seven key components of the plot and indicates a developmental difference with the Icelandic children in her study. These components are set out in Table 5. Table 5. Key components of the plot (Ragnarsdóttir 1991) 1. 2. 3. 4. 5. 6. 7.

The frog leaves the jar. The boy (and the dog) discover that the frog has gone. Initial search inside the house. The search continues outside (at least one attempt). The search continues throughout the story. The boy finds or takes a frog. The finding of the frog in the end is explicitly linked to the loss in the beginning.

The narrators were given a score of 0 to 7 dependent upon how many of these components they referred to explicitly. In Table 6, we see which components, if any, were missing form the bilingual children’s narratives. As for the monolingual Norwegian children, elements #5 and #7 were those generally missing. In the adult stories, none of the components were missing. Table 6 shows that the bilingual children performed similarly across their two languages, with more components missing, however, in the memory mode. Tying the beginning and the end of the story is not accomplished by Kaja in either of her stories, nor by Alice in her Norwegian book version. In Alice’s stories told from memory, no explicit reference to a search for the frog is even made in the two retellings in either language; however, she does tie up the story. Recall that she had a larger number of clauses during the memory stories, and it actually appears then that she has interpreted the task as a memory task, merely listing as many aspects of the story as she could remember without linking them to the search for the frog. Tannen (1979) noted cross-cultural differences in the stories of American and Greek adults who told a story based upon a film without talk (cf. Chafe


Table 6. Missing components from the bilingual children’s narratives BOOK MEMORY

English Norwegian English Norwegian

Kaja

Alice

Nina

#7 #7 – –

#5 #7 #4,#5 #4,#5

0 – – #1,#5

1980). Americans generally completed the task as a memory task filling in as many details as possible while the Greeks provided many interpretations and moral judgments. This idea that it was a memory task is perhaps clear in Alice’s retelling in Norwegian when she stopped to backtrack and talk about a prior event in the boy’s search for his frog:1 ALICE (Norwegian retelling from memory): =Og så – og så datt han ned, And then – and then he flopped down, og da glemte jeg å fortelle noe. and there I forgot to tell about something Og det var det And that was that at etter han hadde hoppa ut after he had jumped out så våkna de, they woke up så var det det at – gutten lette i støvlene then it was that – the boy looked in the boots og så hunden lette i glasset and then the dog looked in the glass der hvor frosken hadde vært where the frog had been

In Nina’s English story told from the book, she continually weaves into the boy’s adventures his motive for the search. In the Norwegian version from memory, she doesn’t explicitly state that the frog left the jar or escaped, nor does she explicitly state the motive for all of the boy’s adventures. Much more is left implicit in the retelling from memory. It is important to point out that Nina was distracted along the way during this retelling, playing with some play doh, and intermittently referring to that; however, that could have been a stalling technique in order to recall the story. Nonetheless, Nina’s stories are richer than the other children’s in the sense that she does not limit her narratives to actions and objects but integrates the protagonists’ thoughts and emotions, a development also evident in the Icelandic data. Nina also uses dialogue through different voices in her stories, particularly at the end which is the high point (cf. Schiffrin 1981) when the boy and dog find the frog. This is similar to her mother’s storytelling strategy. Notice in the following excerpts from the closings of Nina’s story and the story which was her input from her





Elizabeth Lanza

mother. We see that Nina adopts her mother’s storytelling style of including voice changes and dialogue, including approximations of several of the direct quotations. Ending of Nina’s English story: They looked over, And what did they see?!

‘‘Oh I just went out to see my family!’’ ‘‘Your family?!’’ ‘‘One, two, three, four, five, six, seven, eight, nine.’’ ‘‘Yes’’, he said – really proud, Freddie.

‘‘Here – you can have Ragnar with.’’ ‘‘Bye!’’ ‘‘Thank you!

‘‘We’ll maybe come and see you one time!’’ The end.

Ending of Nina’s mother’s story: So Bobby went over and peeked over. … ‘‘Ah,’’ said Bobby. ‘‘There you are. I’m so glad to see you. I’m so glad to find you!’’ And Freddy says, ‘‘Oh, I just had to come and see my family.’’ ‘‘Your family?!’’ said Bob- Bobby. ‘‘Yes,’’ said a very proud Freddy. ‘‘Look at all my little froggies.’’ . . . One, two, three, four, five, six, seven, eight, nine. Nine little froggies. ... And then Freddy said, ‘‘You can take- take George here. And he can stay with you for a while,’’ said Freddy. ‘‘Oh, thank you, Bo-’’ said Bobby. ‘‘That will make me so happy’’. ... ‘‘I’ll be seeing ya,’’ said Bobby. ‘‘It was good to see you. Maybe I’ll come back and see you again.’’ .. N: The end. M: The end.

The mother explicitly states who the author of each quotation is, employing the simple past tense, but shifts to the present when indicating Freddy the


frog’s reason for his escape, thus marking a high point in the story. Nina essentially uses voice modifications to contextualize the quotations. Despite other similarities with her mother’s story, Nina also incorporates the Norwegian name of the frog which is offered to the boy — Ragnar, instead of George. In Nina’s father’s story, Nina co-constructs the ending by providing dialogue, a strategy similar to that of her mother’s story. Note the following ending: Ending of Nina’s father’s story: F: Og så titter han over kanten på stokken N: Og hva ser de? (makes frog noises)

F: And then he peeks over the edge of the log. N: And what do they see? (makes frog noises) F: Og der ser de to frosker, F: And there they see two frogs, mamma- og pappa-frosk… Mama and Papa frog… Og en hel haug med småfrosker. Og så - And a whole heap of little frogs. And thenN: Ni stykker! N: Nine of them! … … N: De kan ta med seg Ragnar hjem. N: They can take Ragnar home with them. F: Skal han få med seg Ragnar hjem? F: Is he going to get to take Ragnar home? N: (Laughs) N: (Laughs) … … F: Når dem da har sett froskene F: When they have seen the frogs, så fikk dem lov til å ta med they were allowed to seg en babyfrosk hjem. take a baby frog home. //Men frosken, og ]// But the frog, and ]N: //’’Kanskje vi ser deg] igjen.’’ N: //‘‘Maybe we’ll see you ] again.’’ F: Froskene sitter igjen, unntatt én, F: The frogs stay behind sitting, except one, og ser etter dem når dem går- går bort. and watch them when they go- go away. Og gutten vinker. And the boy waves. N: ‘‘Ha det! Ha det, frosker!’’ N: ‘‘Bye! Bye, frogs!’’ Slutt! End!

In Nina’s retelling to the research assistant, she incorporates much of the dialogue which originated from her mother’s input, and was expressed in her co-construction with her father.





Elizabeth Lanza

Ending of Nina’s Norwegian story: De gikk over stubben og så. Hm – frosken hadde gått ut til fifamilien hans. De hadde ni barn. Og så … ‘‘Jeg skulle bare ut til familien min. –Joda – jeg skulle bare på besøk.’’ ‘‘Jada! Hvorfor skulle du det da?’’ ‘‘Fordi! Ja – Du får ta med deg Ragnar hjem du da.’’ ‘‘Åja! Tusen takk! Vi kommer kanskje og besøker deg igjen.’’ ‘‘Greit.’’ ‘‘Ha det!’’ Slutt!

They went over the tree stump and looked. Hm – the frog had gone out to his family. They had nine children. And then… ‘‘I was just going to go to my family. –Yep – I was just going to visit.’’ ‘‘Yes! Why were you going to do that then?’’ ‘‘Because! Yes – You can take Ragnar home with you then.’’ ‘‘Oh yeah! Thanks a lot! We ‘ll maybe come and visit you again.’’ ‘‘O. K.’’ ‘‘Bye!’’ End!

Kaja also uses different voices at one point. For example, she makes reference to the picture in which the frog is found in the same style as in her mother’s storytelling — by posing a question and then providing the answer. She does this in both language versions, as illustrated below. Kaja

(English) And what does he find there? Frog! The frog. (Norwegian) Hva ser de der da? What do they see there? Frosken! The frog!

Here we see how the child integrates the input storytelling strategies into both language retellings. Similar to the Icelandic 5- and 7-year-olds, both Kaja and Alice use a large number of sequential markers in their stories, that is, and then, then and the Norwegian equivalents og så, så. The percentage of sequentials reaches a peak among the 7-year-olds in the Icelandic data, and this also happens with Alice who used up to 70 per cent in some retellings. Nina does not resort to this very often in either language. We will return to the use of temporal connectors in the bilingual children’s narratives.


In sum, we see some developmental differences and some which may be related to the interpretation of the task — as a retelling or memory task. The story structure in each of the bilingual children’s language version is similar to that of monolingual peers, both as concerns the number of clauses and key components of the plot. As for storytelling strategies, we find elements from the input the children received from their parents. Some of these strategies are applied across languages, as noted in Nina’s stories.

Signs of language dominance Although all three girls are active bilinguals, there are clear signs of language dominance in their stories, that is, of transfer from Norwegian to English. Although there is minimal language mixing, we see the girls’ dominance in, for example, their language structure.

Kaja There is only one time in which Kaja actually mixes Norwegian into her story and this is in reference to the scene in which the deer throws the boy and the dog over the cliff — a scene which she quite clearly perceives as dramatic, and which represents a high point in her story. Although she does switch back into English, after having begun some utterances in Norwegian, Norwegian intonation prevails throughout the utterances. Up until this point, there were distinct traces of her mother’s regional American accent. (1)

And now the reindeer is running after the dog, — and now they both fall down the cliff! → Men de er very safe! See here. → De fell in the water, → men de er both safe. See here.

But they are . . . They . . . but they are . . .

All of the Norwegian items that she mixed in are function words — part of the grammatical structure of Norwegian. There is no such mixing in her Norwegian story, of English grammatical structure. In line with Petersen’s (1988) dominant language hypothesis, we can interpret this directionality of mixing as a sign of her language dominance. This hypothesis predicts a directionality in mixing with grammatical morphemes from the dominant language cooccurring with grammatical and lexical morphemes from the non-dominant





Elizabeth Lanza

language, but not the opposite situation. What is interesting in this particular example is that it seems to be triggered by her excitement, that is, we see affect triggering her resorting to Norwegian grammatical items, even though she has the corresponding items in English. A similar switch, during an emotional moment, was also registered in a two-year-old Norwegian-English bilingual girl (Lanza 1993). Also in Kaja’s data we find an example of Norwegian word order involving inversion, peculiar to Norwegian which is a verb-second (V2) language. (2)

And then the beehive fell down, → And now is all the bees coming out

Alice In Alice’s stories too we see some interaction between languages from one direction — from Norwegian to English. Alice does not mix any Norwegian into her stories; however, there are some detectable influences, as illustrated in the following: (3)→ ‘t was a time a little frog → It was a time a little glass

(book) (memory)

Once upon a time is rendered in Norwegian by det var en gang, literally ‘it was a time’, which Alice appears to transfer into her English stories. In Examples 4 and 5, we find influence from the Norwegian expletive det which is apparently translated as it, when appropriate usage would require there. This use of it instead of there is also a typical error for Norwegian second language speakers of English. (4)

And then he looked in the little hole, → and there it was a other animal (book)

(5) → Then it was water down there

(book)

In Example 6, we see a calque from Norwegian: (6) → And then they could take with a frog home

(memory)

Norwegian ta med ’take with’ is the equivalent of bring along. In Norwegian, the expression equivalent to be angry with/get mad at is bli sint på; the particle på can be translated as ‘on, at’. In 7, Alice chooses the incorrect form. (7) → then they got mad on the dog

(memory)


Alice does, however, use the correct expression mad at him in the story told from the book. Hence it is only the memory version that indicates dominance. (8) → and then the bees got uh — mad at him, at the dog

(book)

Nina Nina too basically only shows interaction from one language to the other, in her English story. She only once mixes in a Norwegian word, as in Example 9. (9)

And the frog was smiling. → He jumped out! . . . mens (=’while’) Fido and Bobby was sleeping.

Just as in Alice’s case though we see influence in lexical choice and structure. In 10, we also find influence from the Norwegian expletive det, as was the case with Alice. (10)

Once upon a time — in the evening, → it was a boy named Bobby and Freddie — a — the frog and Fido — the — the — dog.

The Norwegian verb miste also includes the meaning ‘ drop, lose’, as well as ‘miss’, which Nina uses in Example 11. (11) → So he missed balance, and fell down.

In 12, Nina rephonologizes the Norwegian verb riste, ‘shake’; although she has not marked it for the past tense, it may actually be marked as a no change verb similar to hit, that is, a verb ending in a dental. (12) → He rist the tree really hard.

In 13, Nina marks the same verb for the past participle. (13) → And when the doggie had risten — shaked — the tree, the bee houses fell down

Although English has a verb miss, as in example 11, this is not the case for rist. Her inflectional ending is interesting but cannot be ascribed to Norwegian. Another example with the same inflectional ending occurs in 14. (14) → ‘‘You could have hurten yourself!’’ Bobby said to the dog.





Elizabeth Lanza

There is one example of an English lexical mix; however, Nina rephonologizes the word, adding Norwegian word accent 2 (indicated by the superscript; cf. Plunkett and Strömqvist 1992): (15) Så så’n i mole-hull.

And then they looked in the mole ([2mɔlə]) hole.

Nina’s father had used the Norwegian equivalent muldvarp. As this is not a very common word, the transfer from English in the Norwegian version told from memory is perhaps predictable. Given this noted dominance, or transfer, from Norwegian to English, we may predict that in marking temporal relations in the stories, the bilingual children will also indicate dominance in Norwegian, in areas of the grammar in which the two languages differ. Let us now turn to examining more closely the temporal relations in the elicited narratives.

Temporal relations in the stories A narrator can choose to present events as past or present; moreover, events can be contoured in time by means of aspectual markings of various sorts. How does the child express ‘‘before’’, ‘‘after’’ and ‘‘simultaneity’’? Important here is the notion of foregrounding and backgrounding in discourse for as Berman and Slobin (1994) point out, foreground and background are not given by the pictures, but are rather constantly constructed by the narrator. The use of the progressive, for example, within a discourse perspective is a backgrounding technique rather than a narrowly semantic use of progressive aspect. Tense forms (simple present/past) are foregrounding techniques and are plot advancing. The successful narrator will interweave both foreground and background information, as is shown in the following excerpt from Nina’s story. (16) And the frog was smiling. He jumped out! . . . mens (=while) Fido and Bobby was sleeping.

Despite the non-adult language use, we see that Nina marks backgrounding information through the use of the present progressive and foregrounds the frog’s jumping out of the container with the use of the simple past tense. The acquisition of these verb forms takes more than just matching form and meaning. The child has to learn to use these forms effectively in discourse, a feat which takes more time. Another backgrounding device is the perfect, considered an aspectual contour by some and a relative tense by others. The perfect is used to relate a


prior event to a present circumstance. Norwegian and English are two closely related languages. They both have grammatical means to express present, past, and perfect, although their distribution of these grammatical means varies in subtle ways. Moreover, progressive aspect is not expressed grammatically in Norwegian. Generally, Scandinavian languages do not mark aspectual distinctions through verb inflections (cf. Haugen 1987; Plunkett and Strömqvist 1992).

The use of temporal connectors in narrative discourse An interesting study of bilingual children’s narratives, though not with the frog story, is Schlyter’s (1996) article on the acquisition of narrative skills in French and Swedish by four children from the ages of 2 to 7. As for markings of temporality, French has a contrast lacking in Swedish, namely that between the passé composé and the imparfait (IMPF). These correspond grammatically to the preterite or simple past in Swedish. As Schlyter points out, the passé composé serves ‘‘to mark foregrounded events, that is, those that are important for the main line of the story, whereas IMPF is used for backgrounded events, that is, for local and temporal settings, explanations, descriptions, etc.’’ (p. 1061). Since the temporal contrast encoded grammatically in French is not encoded grammatically in Swedish, how do the bilingual children accomplish this contrast? What Schlyter found was that in Swedish, the foregrounded clauses in the narratives, which move the temporal reference point, were practically always introduced by a temporal connector då/ (och) sen/ (och) så (similar to and then) followed by the inverted construction VS (Swedish similar to Norwegian is a V2 language). Hence the pattern employed in Swedish for foregrounding clauses was that of temporal connector + V + S, as noted in Table 7. In French, the passé composé and SV word order was used to this end. Hence temporal connectors (+ VS word order) are a kind of compensatory device in Swedish for marking what is accomplished by aspectual opposiTable 7. Textual devices for foregrounding and backgrounding in French/Swedish bilingual children (adapted from Schlyter 1996: 1082) Foregrounding Backgrounding

French

Swedish

passé composé SV word order imparfait SV word order

preterite då/(och) sen/(och) så/VS word order preterite SV (non-inverted) word order





Elizabeth Lanza

tion in French. And as Schlyter hypothesizes, we would expect to find more temporal connectors in the Swedish narratives than in the French. In three of the four children this was in fact the case. Just like French, English has such an aspectual opposition. English has the progressive -ing in addition to the simple past, while Norwegian similar to Swedish only has the preterite. So we can similarly predict that the bilingual children will introduce more clauses in Norwegian with (og) så (‘(and) then’) and similar devices than they do in English, as they use this especially for foregrounding information. In this calculation, we only include clauses that have finite verbs. In Table 8 we see that the prediction is met in all of the cases except in Alice’s narratives without the book (noted in italics). In other words, temporal connectors are used more often in the Norwegian stories compared with the English stories. Here we see how the bilingual children differentiate between their languages in their use of narrative devices. We will return to Alice’s narratives. Table 8. Use of temporal connectors by bilingual children in Norwegian and English Kaja

Alice

Nina

Norwegian No. of clauses % of clauses

34 (of 51) 67%

Book 42 (of 57) 74%

Memory 42 (of 72) 58%

20 (of 60) 33%

English No. of clauses % of clauses

18 (of 36) 50%

Book 33 (of 53) 62%

Memory 39 (of 52) 75%

3 (of 63) 5%

What about monolingual children’s use of temporal connective devices? Berman and Slobin (1994) point out that and then is the single most favored connective device among the 5- and 9-year-olds, used as a chaining device. On the whole the 5-year-olds in their study employed it in 26 per cent of the clauses while the 9-year-olds only used it 15 per cent of the time. Hence we see a tapering down with age and it was only marginal among the adults. This can be ascribed to the fact that with greater linguistic resources, the children start using other devices such as subordinators like when, nonfinites and other introductory markers of sequentiality. This was also the case for Nina. If we contrast the narratives of the Norwegian monolingual children in this study with those of the English monolingual children, we find that in fact they do use this connective device more often: from 50 per cent to 69 per cent


of the clauses contain them. Hence this analysis shows how the bilingual children differentiate between the languages at a discourse level.

The use of tense/aspectual forms in narrative discourse Table 9 shows the different tense/aspectual forms used by the children. Note that although time relations can be expressed by other means including adverbials, the emphasis here is on the development of the grammatical means for indicating temporal relations. Table 9 merely lists the forms used by the children whereas we need to examine how these forms are used in discourse. Work on the frog story has shown that even adults will vary as to whether or not they narrate in the present or the past. For those narrating in the past, a switch to the present can be functional, as in the use of the ‘‘historical present’’ (cf. Schiffrin 1981). Both Alice and Nina established a narrative time line in the past but Nina does shift to the present essentially for functional reasons, especially in the form of dialogue, as discussed above. Kaja, on the other hand, seems to shift back and forth, a developmental phenomenon documented in Berman and Slobin’s (1994). Very young children often performed the task as a picture description task while integration of the pictures into a narrative structure developed later. One interesting use of temporal/aspectual forms here is of the perfect. As Johansson and Lysvåg (1991) point out, when there is no overt specification of a definite time in the past, Norwegian is more apt to use the perfect while English more frequently uses the past tense. American English in turn tends to use the past tense more often than British English (cf. Elsness 1991; Gathercole 1986). Slobin and Berman (1994) note that in the English narratives in their

Table 9. Use of tense/aspectual forms

Pres Pres Prog Past Past Prog Pres Perf Past Perf

Kaja (4;6)

Alice (6;10)

Eng Nor

Eng Nor Book Book

Eng Nor Memory Memory

Eng

Nor

14 5 14 2 0 0

1 0 45 6 0 1

2 0 42 6 0 2

11 1 54 4 0 4

7 – 60 – 0 5

31 – 20 – 0 0

0 – 54 (3) 0 0

Nina (8;6)

0 – 66 (2) 0 4





Elizabeth Lanza

data, the verbs that were marked for perfect aspect expressed the related functions of marking anteriority of one event over another, and securing a sequence of tenses. Both related functions can be characterized as ‘‘expressing relative tense’’ (p. 142). Slobin and Bocaz (1988) point out that there were no uses of the perfect in the stories told by the English-speaking children in their study. Berman and Slobin (1994) note that it occurred among some of the 9-year-olds in their sample, but rarely. In Table 9, we see that Alice uses the past perfect less frequently in her English than in her Norwegian in the stories told from memory, but then she does not use it in her Norwegian story told from the book. Nina has one more occurrence in Norwegian than in English. Of these two girls’ combined past tense forms, 4 per cent in English involve the perfect while 5 per cent were involved in the Norwegian stories.2 We may ask whether this use of the perfect in English can be attributed to an influence from Norwegian, given the noted refraining from the use of this aspect among the monolingual children. This may be a subtle interaction between the languages in the child’s bilingual processing. Hence although the grammatical means is available in both languages, the propensity to use it is greater in Norwegian. This may influence the child’s language processing as she tells the story in English. Before examining the distribution of forms in the Norwegian monolingual children’s stories, and in both language versions of the Norwegian adults, let us discuss each bilingual child’s use of particular tense and aspectual forms in her stories.

Kaja In Table 9 we see that Kaja uses a larger percentage of present tense forms than the other two children. It is her mixing of both past and present, however, that is of interest. Berman and Slobin (1994) point out that the youngest children in their study used forms of present and past tense to mark the aspect of the pictured events, rather than to locate them on a narrative time line. These were children that had not constructed ‘‘narrative time’’ as distinguished from ‘‘speech time’’ — they basically described what they saw in the pictures. In example 17 from Kaya’s data we see that she described a process with a present tense form, and a state with a past tense form in her reference to the picture in which the beehive is lying on the ground. The same is done in her Norwegian version (18). (17) And then the beehive fell down, And now is all the bees coming out


(18) og så — nå falt den ned, og så kommer alle biene ut

and then — now it fell down, and then all the bees come out

This tendency was common in the younger English monolingual children’s frog stories reported on in Berman and Slobin (1994). Although Kaja uses the sequential marker and then and the Norwegian equivalent to move the story along, she also uses now and the Norwegian equivalent nå with narrative clauses. This reflects the tension in her stories between following a narrative time line and describing pictures.

Alice Alice also uses temporal conjunction to move the narrative along, as previously mentioned. In her Norwegian story (from memory) we find attempts at expressing simultaneous activities, as noted in her reference to the activities of both the boy and the dog: (19) og så begynte han å leke med det og sånn and then he began to play with it imens gutten klatra opp i treet while the boy climbed up in the tree

If we compare this with the English story from memory, there is no such attempt to express simultaneity. Here she appears to express more sequentiality in her English story: (20) And then he started playing with it, and then the boy was climbing in a tree.

(memory)

Alice does, however, attempt temporal subordination in both her English and Norwegian stories : (21) → And when they went into the forest to look then they — then — the dog was following the bees (memory) (22) → Men når biene kom — etter hunden for hunden gikk dit, og så — og så kom elgen over . . . ikke elgen da, jeg mener ugla (book) But when the bees came — after the dog for the dog went there, and then — and then the moose came over not the moose then, I mean the owl





Elizabeth Lanza

What is interesting with Alice’s data is her use of an optional lexical construction in Norwegian which can convey progressive aspect, that is, the conjunction of two verbs. Haugen (1987) refers to this construction as a ‘‘durative expression’’; it involves the combination of a verb of motion or position, followed by and, then the main verb, with both verbs in the same tense: for example, jeg står og snakker, lit., ‘I stand and speak’ meaning ‘I am speaking’. The idiomatic Verb-and-Verb constructions essentially involve the verbs ‘to sit’, ‘to lie’, and ‘to stand’.3 Alice, however, uses the Norwegian verb drive (/2dri:ve/) in this construction which means ‘be engaged in doing (something)’ which further accentuates the notion of durativity. Of the five occurrences of the Verb-and-Verb construction in her data, four are with the verb drive. Hence her constructions are less idiomatic, yet do express progressivity. There are three occurrences in the story with the book, and two from the story from memory (noted in Table 9 in parentheses), as in the examples below: (23) og så sto de og ropte ut av vinduet (book) and then they were calling out from the window (lit., and then they stood and called out from the window) (24) og så drev de og løp etter ham (book) (referring to picture 12) and then they were running after him (lit., and then they were engaged in and ran after him or they were (busy) running after him)

As Alice has already told the story twice in English before telling it in Norwegian, and has used the progressive aspect in the English, we could hypothesize that this is motivating her to use this construction in Norwegian in which it is not obligatory. In examining the distribution of these five progressive constructions in her Norwegian stories in comparison to her English stories, we find, however, that there is only one which appears to be used in narration for the same picture or event, as illustrated below from the Norwegian story told from memory : (25)

og så satt de i vannet, og så så de en · stokk → som drev og — som var — var i land. and then they sat in the water, and then they saw a log which was (prog.) and — which was- was on land.


In this clause, Alice self-interrupts and reformulates the relative clause with the verb to be, possibly because of a lexical accessing problem. In the English story told from the book, we find the following excerpt in narration of the same picture: (26)

they sat in the water, and then — then they saw · a little branch → floating in the water, no- eh — on land.

In the English version, Alice does express an ongoing activity through the nonfinite participial use of an -ing form. With the other occurrences, since there is no overlap between the distribution of the progressive constructions in Norwegian and those used in English, it is difficult to claim that Alice’s use of this construction is simply to be attributed to the fact that she had told the stories in English before Norwegian. In other words, it appears that Alice senses a need to express progressivity in her Norwegian stories and to mark it through the verb, despite the fact that she appears to be dominant in Norwegian. In that case, this would be an interesting interaction between a nondominant language and a dominant language. This may be parallel to her greater use of the perfect in her English stories than used by English monolingual peers. We will return to this aspect of her language use once we have compared the bilingual children’s use of temporal forms with those of the monolingual Norwegian children and the Norwegian adults.

Nina Nina does not use the double verb construction as Alice did. She did, however, use present tense forms like Kaja, but these were essentially in clauses in direct speech, which also moved the plot along. By shifting to the present tense in her narrative in reported speech, she presents internal evaluation for the story. Labov (1972) has pointed out that narratives are often told to illustrate general propositions and to establish a central affective point. As Schiffrin (1981) noted in reference to tense variation in narrative, direct quotes can provide an internal evaluation in a narrative. Direct quotes allow the narrator to convey the point of the narrative through the exact words of someone present during the experience itself, so that the audience can appreciate for itself the authenticity of the evaluation. Hence Nina’s use of the present tense forms is quite functional. She uses more in the English version than the Norwegian one; this may be due to the fact that she is involved in a memory task in the Norwegian version. We cannot, however, dismiss the fact that her mother’s narrative





Elizabeth Lanza

input in English also employed direct quotes in present tense forms, as discussed above, and this may have had some influence on Nina’s performance. Nina also uses time adverbials to sequence events as in her use of først. (27) Så måtte de ut og lete da men først så ropte de ut av vinduet

Then they had to (go) out and look then but first then they called out of the window

Discussion Before comparing the bilingual data with the monolingual data and that of the Norwegian second language speakers of English, let us summarize the picture which is suggested by these bilingual children’s construction of temporal relations in their stories. First of all, we see a developmental difference across the three girls in both stories. The youngest child, Kaja, uses simple present and past tense forms, as well as present and past progressive forms, alternating between narration and picture description. She does use some temporal conjunction to follow a story line. Both Alice and Nina use narration in the rendition of their stories. Both also use temporal conjunction, and especially Nina, the oldest child, uses temporal subordination. Both girls also use time adverbs to move the story along. All three girls show signs of language dominance in Norwegian, yet Alice incorporates durative expressions in her Norwegian stories, thus marking events through the verbs as durative in Norwegian, a language which does not have grammatical coding for such an aspect. Moreover, she and Nina together seem to employ the perfect in English more often compared to monolingual peers. A greater use of the perfect is a distribution common to Norwegian in comparison with English. The monolingual Norwegian children Let us now examine the data from the three monolingual Norwegian children, as shown in Table 10. Recall that the Norwegian child 1 is a boy aged 6;7, the second a girl aged 7;0, and the third a boy aged 7;1. In Table 10, we see that the younger boy also mixes present and past tense forms in his story told from the book, similar to Kaja, while this is not exactly the case for the story told from memory. Contrary to Kaja, however, his entire story is narration. What happened is that he started narrating the story in the


Table 10. Use of tense/aspectual forms by Norwegian children: finite verbs

Pres Pres Prog Past Past Prog Pres Perf Past Perf

NorCH 1B

NorCH 2G

NorCH 3B

Book

Memory

Book

Memory

29 – 28 – 1 0

4 – 55 – 0 1

0 – 55 (1) 0 5

2 – 42 – 0 1

past, then switched to the present, only to conclude the story once again in the past. In other words, there is no alternation between tense forms. The few present tense forms in the story told from memory are basically direct quotes, and hence functional uses of this form. The other two children prefer narration in the past; the two uses of present tense by the second boy are in direct quotes. The bilingual children Alice and Nina are similar to the Norwegian children, (who approximate Alice in age), preferring narration in the past, although they incorporate present tense forms in quotes. The 7-year-old Norwegian girl uses one form which is an idiomatic durative expression: (28) → og så satt han og så på den frosken langt utover kvelden og så ble han så trøtt at han sovna.

and then he was looking (lit., he sat and looked) at that frog late into the evening and then he became so tired that he fell asleep.

Apart from this example, there are no other instances of durative expressions among the Norwegian children. Hence Alice’s usage is not similar to that of her monolingual Norwegian peers here as far as durative expressions are concerned. Although both monolingual boys do use one instance of the perfect in each of their stories, the monolingual girl has greater use of this form. Overall for the Norwegian monolingual children, 4 per cent of the past tense forms involved the perfect4 compared to the 5 per cent usage by the two older bilingual children. Here we see that the bilingual children’s use of Norwegian compares to that of the monolingual children, as far as the use of the perfect in narratives is concerned. There is, however, individual variation.





Elizabeth Lanza

Table 11. Adults’ use of tense/aspectual forms: finite verbs Pres Pres Prog Past Past Prog Pres Perf Past Perf

Eng

Nor

117 5 129 2 8 4

137 (1) 178 (1) 6 7

The Norwegian adults Let us now examine the verbal forms used by the Norwegian adults/second language speakers of English for marking temporal relations on finite verbs. Table 11 illustrates the distribution of such forms. Recall that there are four adults in the sample, two women and two men. The two women preferred the present tense in narration which was their ‘‘dominant tense’’ while the two men preferred narration in the past, with occasional use of the present in direct quotes. Similar to the monolingual Norwegian children, the Norwegian adults employed the perfect in 4 per cent of the past tense forms; moreover, they also used the perfect in 4 per cent of the present tense forms. There are, relatively speaking, few instances of either the present progressive or past progressive in the English sample, despite the number of narrative clauses. There are, however, two examples of ‘‘durative expressions’’ in the Norwegian stories involving combinations of two verbs: (29) NorAD 1F: og de sitter der og koser seg and they are enjoying themselves (lit., and they sit there and enjoy themselves) (30) NorAD 4M: en kveld satt de og pratet sammen one evening they were chatting together (lit., one evening they sat and chatted together)

Given the paucity of such constructions in the adults’ Norwegian stories, and in the monolingual children’s stories, it is interesting to consider Alice’s use of these constructions. Given that she did not mark the exact same verbs across languages, we hypothesized that she appeared to feel the need to mark progress-


ivity lexically through the verb, despite the fact that Norwegian does not have any grammatical means for doing so. This then can be interpreted as a sign of the influence, albeit subtle, of English over her dominant language, Norwegian.

Distribution of -ing forms in the English stories Let us now examine the total use of -ing forms in the English stories of the bilingual children and the adult L2 speakers of English. Berman and Slobin (1994) investigated the use of verb forms with suffixal -ing in the English narratives in their sample, and discovered that they accounted for approximately 25 per cent of all lexical verbs across the different age groups, that is, verbs other than the copular be. The youngest children in the sample, the three-year-olds, used many ‘‘bare’’ progressives, that is, -ing forms with no overt auxiliary be. Although there are no ‘‘bare’’ progressives in the bilingual child data, we see in Table 12 that apart from four-year-old Kaja, the children’s stories generally do not show preference for the use of -ing forms to the same extent as the monolingual English child and adult population. However, Alice’s pattern resembles more the monolinguals’ than does Nina’s (the English monolingual 5-year-olds and 9-year-olds used 19 per cent and 20 per cent respectively, cf. Berman and Slobin, p. 138). Similar to the monolingual English-speaking children in Berman and Slobin (1994), we note that the youngest preschooler, Kaja, used progressive forms in the present more often than in the past; as the two older bilingual girls narrated in the past, they used the past progressive. Berman and Slobin (1994) found developmental differences in the use of finite progressive aspect, with the youngest children preferring to anchor their stories in the present and Table 12. Distribution of -ing forms in bilingual children’s English narratives

Total lexical verbs % of -ing forms Total -ing forms ‘‘Bare’’ progressives Present progressive Past progressive Nonfinite -ing forms

Kaja

Alice

Book

Book

Memory

Book

27 26%

48 15%

52 15%

58 9%

7 0 5 2 0

7 0 0 6 1

8 0 0 6 2

5 0 1 4 0

Table adapted from Berman and Slobin (1994: 138)

Nina





Elizabeth Lanza

using a great deal of -ing forms to mark progressive aspect, often in the process of picture-description. As they point out, progressive aspect was often used by a few older children ‘‘as a bridge between picture-description and fully narrative mode’’ (p. 141). There was some variation in the adults’ use of these forms; however, unlike the youngest children, they never anchored their narratives in progressive aspect. Rather they used these forms ‘‘for events which form a durative background to the plot-advancing sequentially unfolding course of events described in simple narrative present or past’’ (p. 142). In Table 13, we find that the second language speakers of English also use -ing forms less frequently than the monolingual speakers. There was, however, considerable variation among these Norwegian adults (18 per cent and 23 per cent for the two women, and 3 per cent and 9 per cent for the two men). Table 13. Distribution of -ing forms in Norwegian adults’ English narratives All Adults Total lexical verbs % of -ing forms

221 12%

Total -ing forms ‘‘Bare’’ progressives Present progressive Past progressive Nonfinite -ing forms

27 0 5 2 20

One salient difference between the adult second language speakers of English and the bilingual children concerns the use of nonfinite -ing forms. Berman and Slobin (1994: 139) define ‘‘an -ing form as nonfinite if it occurs with a verb in present or past tense, other than the auxiliary be, within the same simple clause or in an adjacent dependent clause.’’ Although the adults use fewer finite -ing forms than the bilingual children, they do use more nonfinite -ing forms, as shown in Table 14. The difference in distribution of the -ing forms as shown in Table 14 may actually be a developmental one in that the monoligual English-speaking adults Table 14. Distribution of -ing forms in English narratives: Adults (L2) and bilingual children Finite Non finite

L2 English adults

Bilingual children

7 (26%) 20 (74%)

24 (89%) 3 (11%)


perform similarly to the L2 English-speaking adults in their preference for nonfinite -ing forms. Among the adult English monolinguals, 48 per cent of the -ing forms were finite, compared with 52 per cent nonfinite. However, the L2 speakers of English do on the whole use fewer -ing forms than the monolingual adults (12 per cent compared to 25 per cent). We may ask whether this can be interpreted as a strategy of avoidance by at least some of the adults since it is an area of grammar which differs markedly from Norwegian. The bilingual children, on the other hand, performed similarly to their age cohorts among the English monolingual sample in their propensity to use finite -ing forms over nonfinite. However, unlike their monolingual peers, they rarely used nonfinite verb forms. The English monolingual 5- and 9-year-olds used 4 per cent and 38 per cent respectively (see Berman and Slobin, p. 138, Table 2). Neither Kaja nor Nina used any such non-finite forms, and only 3 per cent of a total of 100 lexical verbs used by Alice were nonfinite -ing forms. The preferred use of the nonfinite form among the Norwegian adults was as a complement to an aspectual verb (often start, keep), as in the following examples: (31) → they start looking around for the frog — frantically

(NorAD2F)

(32) → the moose kept running faster and faster

(NorAD4M)

This usage was common among the preschoolers in the English monolingual sample. Of the three uses of nonfinite -ing forms by Alice, two involve aspectual verbs, and one is a nonfinite participial functioning as a relative clause. The -ing form in English is used to express progressive aspect; the expression of durativity or simultaneity of events can also be rendered by other means, for example, by adverbials of aspect. Similar to German, Norwegian can use sentence-modifying adverbials and subordinating conjunctions for expressing simultaneity (see Aksu-Koç and von Stutterheim 1994). We find two examples of the subordinating conjunction mens ‘while’ in the Norwegian adult data, and none in the Norwegian monolingual child data, as in the following: (33)

NorAD2F: → og bikkja klarer å rive ned vepsebolet, eh- mens gutten driver på med dette her. and the dog managed to tear down the wasp nest, eh- while the boy was engaged with this here.





Elizabeth Lanza

(34)

NorAD3M: → ens gutten og hunden sov, fant frosken ut at nå var det nok på tide å gå sin vei. while the boy and the dog slept, the frog decided (found out) that now it was about time to go its own way.

Interestingly, it is only Alice among the children, both bilingual and monolingual, who uses a similar form. In her data, there are three occurrences of the form imens ’meanwhile’, as illustrated in example 19 above. In sum, the picture that emerges concerning the expression of temporal relations in the elicited narratives is that the bilingual children resemble the English monolingual children in some respects, but not in all. Moreover, especially in the case of Alice, we see that the bilingual children do not exactly perform as their Norwegian monolingual peers either. In a very subtle way then, the bilingual children appear to fall ‘‘in between’’, as in the words of Hernandez, Bates and Avila (1994).

Conclusions and discussion In this exploratory study, some interesting patterns have emerged. With a focus particularly on the expression of temporal relations in elicited narratives by bilingual children, comparisons were made with the narratives of monolingual children and adult speakers. Some developmental differences were noted in the three bilingual children aged 4;6, 6;10, and 8;8, particularly in comparison with the extensive work done on the monolingual English-speaking children reported on in Berman and Slobin (1994). Although the bilingual children did perform like the monolingual children of their respective languages in certain domains, they also seemed to perform differently from their monolingual peers in other domains. A case in point is Alice’s use of durative expressions in Norwegian, a language which does not mark progressive aspect like English. This may be just an individual difference given that neither of the other two bilingual girls performed this way. However, from a bilingual processing point of view, it is nonetheless quite interesting that she should sense the ‘‘need’’ to mark events as durative in Norwegian. Moreover, both of the older bilingual girls also used the perfect in their English narratives more than is reported for monolingual children.


Although studies of early bilingualism in children report on how young bilinguals resemble their monolingual peers in their language use, there is a need to consider how they may in fact differ. Understandably, emphasis has been on how bilingual children resemble their monolingual peers so as to stress the idea that their language development is normal. However, there is reason to anticipate that their development may be different albeit normal. The bilingual is, after all, not two monolinguals in one and the same person, as Grosjean (1985) has so strongly argued. Hernandez, Bates and Avila (1994) have convincingly demonstrated how Spanish-English bilinguals, who acquired English during primary language acquisition, appeared to fall ‘‘in between’’ the two monolingual control groups in an on-line sentence interpretation task. These scholars presented four potential profiles of bilingualism in a task in which the subjects were to choose the agent of a sentence: differentiation, or the use of separate strategies for each language; forward transfer, or the use of L1 strategies in processing L2; backward transfer, or the use of L2 strategies in processing L1; and amalgamation, in essence a combination of forward and backward transfer, that is, the use of a set of strategies merging those employed by monolinguals in each of the bilingual’s languages (see Hernandez, Bates and Avila, 1994: 421). We cannot speak of an L1 and an L2 in the simultaneous acquisition of two languages; however, as for direction of transfer, we may see a parallel between the L1 and the dominant language. Although the three young bilingual children in this study show signs of differentiation in their expression of temporal relations, there is also some indication of amalgamation. With supporting evidence from other bilingual studies, Hernandez, Bates, and Avila (1994) point out that amalgamation has been the dominating pattern among bilinguals who have acquired both languages by an early age and who use both languages on an everyday basis. Moreover, they point out that the strategy of amalgamation suggests that ‘‘bilinguals are not really on the road to monolingualism’’ (p. 441). The evidence presented in Hernandez, Bates, and Avila (1994) is from psycholinguistic testing and hence synchronic evidence for bilingual language processing. Other synchronic evidence of language change based on natural language use and which reveals ‘‘covert influence’’ (Romaine 1989a) from one language to another in a contact situation is reported on in Klein (1980; referred to in Romaine 1989a). Parallel with Alice’s case, the grammatical markings in question are the use and distribution of the simple present and the progressive in a language contact situation involving English and Spanish in the U.S. Both languages mark progressivity; however, the distribution of the





Elizabeth Lanza

tense and aspect categories differ. Klein found that compared to monolingual Spanish speakers, the bilinguals used the progressive relatively more frequently; furthermore, they used the simple present less frequently than their monolingual counterparts in reference to time which included the specific moment of speech. A semantic shift was, therefore, in progress in the language system of these speakers since the simple present changed its meaning due to its more narrow range of use. We find other diachronic evidence of what Hernandez, Bates, and Avila (1994) term ‘‘amalgamation’’ in the form of language change through language contact. Talmy (1982) discusses the effects upon the semantic system of one language in contact with another. Focusing on Yiddish in contact with Slavic, Talmy presents ‘‘a framework for understanding the structured interaction of semantic systems’’ (p. 248). Yiddish developed initially in the Middle High German-speaking Rhineland and some centuries later extended its range into Slavic-speaking territories. It was under the influence of Slavic that the Yiddish semantic system underwent many accommodations; many of these can be observed in the verbal prefix system and constructions associated with that system. Talmy points out quite astutely a variety of ways in which Yiddish has borrowed from Slavic, including many subtle ways. He concludes by abstracting a set of principles from the properties of semantic change that Yiddish prefixes have undergone under the influence of Slavic. A particularly interesting principle is one for which we find a parallel in the data from the bilingual child Alice. Talmy states that if a language borrows from an unparalleled donor morpheme class, the borrowing language does not assume the syntactic category but the meanings, and expresses these meanings with a native construction which is already semantically consonant with the donor language meaning. Yiddish did not borrow the Slavic suffix which indicated semelfactive aspect, but expressed it with its native periphrastic construction which already contained some of the donor language meaning. Similarly, Alice did not import progressive aspect marked on the verb directly into her Norwegian, rather she expressed similar meanings through the use of native durative expressions. Romaine (1989b) discusses the role children play in the overall communicative structure of their speech community. As she notes, during language acquisition, ‘‘children play an active role in elaborating, constructing and (re)-creating their language’’ (p. 218). While there is aged-graded linguistic behavior, children learning language do employ processes which have been noted applicable to language history. More work is needed, however, as Romaine points out, to


ascertain the relationship between language acquisition by children and language change in order to understand why some changes eventually become part of the restructuring of a language while others do not. A particularly fruitful area for doing such work should be stable bilingual communities. Investigating the child’s acquisition of two languages could provide insights into language change as a result of language contact. Central issues in the study of bilingual first language acquisition include the extent to which there is interaction between the bilingual child’s two languages, that is, how they interact. Relevant to this issue is that of the extent to which the child’s language development is similar to the linguistic development in monolingual children acquiring each of the bilingual child’s languages. Furthermore, we may ask what role the input plays in the bilingual child’s language development. Among young bilingual children, it is particularly in such domains as the expression of temporal relations that subtle differences in language use may be detected. Hence this domain has been the focus in this chapter for addressing the central issues in the study of bilingual first language acquisition. Kaja and Nina received bilingual input in the stories, that is, input in Norwegian and English from the mother and the father, respectively. As noted particularly in the case of Nina, the children incorporated some of the input, both structures and storytelling strategies, in their retellings in each language. This indicated that the input stories were still in their memory. Indeed the relationship between the retellings from the book and those from memory is an issue which will require further exploration. As for the processing of the two languages, the retellings from the bilingual children suggest that several factors come into play. First, there is the psycholinguistic factor of having to reproduce in the one language, a task accomplished or processed in the other language. Memory constraints will also play a role, including the factor of familiarity with the story. The child must, furthermore, make language-specific choices where the two languages differ, and in making these choices the psycholinguistic factor of dominance may play a role. Another dimension of the retelling of a story that has not been investigated is the amount of variation that can occur in several retellings from one child. This dimension will include sociolinguistic/interactional aspects of interpersonal communication as well as psycholinguistic ones. As noted in the discussion, among the bilingual retellings, there is an underlying assumption that language is the only variable at play. This assumption must be tested, also among monolingual children.





Elizabeth Lanza

The data presented in this chapter suggest that the bilingual child’s language development and use are similar to that of their monolingual peers, in some respects; however, there were noteworthy differences. Whether these differences are purely individual, or indicative of bilingual language development, will need to be explored in larger populations. In studies of language contact, more work is needed on how children actually may function as agents of language change. And in studies of bilingual first language acquisition, there is a need for exploring the subtle ways in which the bilingual child’s linguistic development may in fact be different from that of the monolingual child. The narrative provides an excellent means for addressing these intriguing questions.

Appendix: Transcription conventions (Adapted from Berman and Slobin 1994: 659) Default (no indication) = a steady or sustained intonation raised dot ( · ) = very slight pause, break A dash ( — ) = a short pause/self-interruption Three dots (or more if needed) = a longer pause . . . Comma ( , ) = partially falling intonation Period ( . ) = fully falling intonation Question-mark (?) = end of question-type utterance Colon ( : ) = lengthening of a vowel (e.g. ru:::n) Underlining = stressed element (e.g. that tree) Overlap = // placed at beginning of overlap ] placed at end of overlapped utterances e.g. Child: the boy found // the frogs] Adult: // look at them] → = the element/clause in focus

Notes . See the appendix for some transcription conventions in excerpts from the stories. . Seven of the 164 past tense forms were in the perfect in English while nine of the 194 Norwegian past tense forms were in the perfect. . This construction is also found in Swedish and Icelandic; however, in a crosslinguistic study of 15-year-olds, Swedish speakers used the Verb-and-Verb construction far more often especially for encoding progressive than their Icelandic peers. Icelandic speakers


tended to prefer present and past progressive (Sven Strömqvist and Hrafnhildur Ragnarsdóttir, personal communication). . Seven instances out of 188 past tense forms. Note that the girl has 8 per cent.

References Aksu-Koç, A. and C. von Stutterheim (1994). Temporal relations in narrative: Simultaneity. In: R. Berman and D. Slobin (eds.), Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Bamberg, M. (1987). The acquisition of narratives. Learning to use language. Berlin: Mouton de Gruyter. Berman, R. and D. I. Slobin (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Chafe, W. (1980). The pear stories: Cognitive, cultural, and linguistic aspects of narrative production. Norwood: Ablex. De Houwer, A. (1990). The acquisition of two languages from birth: A case study. Cambridge: Cambridge University Press. Elsness, J. (1991). The perfect and the preterite. The expression of past time in contemporary and earlier English. Doctoral dissertation, University of Oslo. Gathercole, V. (1986). The acquisition of the present perfect: Explaining differences in the speech of Scottish and American children. Journal of Child Language, 13, 537–60. Genesee, F. (1989). Early bilingual development: One language or two? Journal of Child Language, 6, 161–79. Grosjean, F. (1985). The bilingual as a competent but specific speaker–hearer. Journal of Multilingual and Multicultural Development, 6, 467–77. Haugen, E. (1987). Danish, Norwegian and Swedish. In: B. Comrie (ed.), The world’s major languages. London: Croom Helm. Hernandez, A. E., E. A. Bates and L. X. Avila (1994). On-line sentence interpretation in Spanish-English bilinguals: What does it mean to be ‘‘in between’’? Applied Psycholinguistics, 15, 417-46. Hummel, K. (1993). Bilingual memory research: From storage to processing issues. Applied Psycholinguistics, 14, 267–84. Johansson, S. and P. Lysvåg (1991). Understanding English grammar. Part II: A closer view. Oslo: Universitetsforlaget. Klein, F. (1980). A quantitative study of syntactic and pragmatic indications of change in the Spanish of bilinguals in the U. S. In: W. Labov (ed.), Locating language in time and space. New York: Academic Press, 69–82. Labov, W. (1972). The transformation of experience in narrative syntax. In: W. Labov, Language in the inner city. Philadelphia: University of Pennsylvania Press. Lanza, E. (1992). Can bilingual two-year-olds code-switch? Journal of Child Language, 19, 633–58. Lanza, E. (1993). Language mixing and language dominance in bilingual first language acquisition. In: E. V. Clark (ed.), The proceedings of the Twenty-fourth Annual Child Language Research Forum. Stanford University, 197–208.





Elizabeth Lanza

Lanza, E. (1997). Language mixing in infant bilingualism: A sociolinguistic perspective. Oxford: University Press. Meisel, J. (ed.) ( 1990). Two first languages: Early grammatical development in bilingual children. Dordrecht: Foris. Paradis, J. and F. Genesee (1996). Syntactic acquisition in bilingual children: Autonomous or interdependent. Studies in Second Language Acquisition, 18, 1–25. Petersen, J. (1988). Word-internal code-switching constraints in a bilingual child’s grammar. Linguistics, 26, 479–93. Plunkett, K. and S. Strömqvist (1992). The acquisition of Scandinavian languages. In: D. Slobin (ed.), The crosslinguistic study of language acquisition. Vol. 3. Hillsdale, NJ: Lawrence Erlbaum, 457–556. Ragnarsdóttir, H. (1991). Episodic structure and interclausal connectives in Icelandic children’s narratives. Proceedings of Colloquium Paedolinguisticum Lundensis. Lund: Sweden. Ransdell, S. E. and I. Fischler (1987). Memory in a monolingual mode: When are bilinguals at a disadvantage? Journal of Memory and Language, 26, 392–405. Romaine, S. (1989a). Bilingualism. Oxford: Blackwell. Romaine, S. (1989b). The role of children in linguistic change. In: L. E. Breivik and E. H. Jahr (eds.), Language change. Contributions to the study of its causes. Berlin: Mouton de Gruyter. Schiffrin, D. (1981). Tense variation in narrative. Language, 57, 1, 45–62. Schlyter, S. (1996). Bilingual children’s stories: French passé composé/imparfait and correspondences in Swedish. Linguistics, 34, 1059–85. Slobin, D. I. and A. Bocaz (1988). Learning to talk about movement through time and space: The development of narrative abilities in Spanish and English. Lenguas Modernas (Santiago, Chile), 15, 5–24. (Circulated as Berkeley Cognitive Science Report No. 55, Institute of Cognitive Studies, University of California, Berkeley, January 1989). Strömqvist, S. (1996). Discourse flow and linguistic information structuring in speech and writing. Gothenburg Papers in Theoretical Linguistics. Vol. 78. Strömqvist, S., H. Ragnarsdóttir, et al. (1995). The Inter-Nordic study of language acquisition. Nordic Journal of Linguistics, 18, 1, 3–29. Talmy, L. (1982). Borrowing semantic space: Yiddish verb prefixes between Germanic and Slavic. Proceedings of the Eighth Annual Meeting of the Berkeley Linguistics Society. Berkeley Linguistics Society, University of California, Berkeley. Tannen, D. (1979). What’s in a frame? Surface evidence for underlying expectations. In: R. Freedle (ed.), New Dimensions in Discourse Processes. Norwood, NJ: Ablex Publishers. Also in Tannen (1993), Framing in Discourse. New York: Oxford University Press.

Chapter 3

Reference continuation in L2 narratives of Turkish adolescents in Norway Ingvild Nistov

The scope of the overall study which this chapter1 is derived from is the investigation of how L2 learners of Norwegian whose source language is Turkish approach the target language conventions for referential management. The focus here is the L2 learners’ expression of continuation of reference in short written narratives. Referential management in decontextualised narrative discourse has been associated with cognitive and/or academic and literary skills (cf. Bartlett 1984; Brown and Yule 1983; Cummins 1983; Lindberg 1995). The narrative in that respect is characterised as language use which requires from the learner to convey a coherent message with little available contextual support, in which all the information needed for comprehension should be included in the words of the story (Lindberg 1995: 9). This would apply even more to written than to oral narratives. The focus on written narratives in this study can thus be seen as relating to their suitability for the analysis of the acquisition of decontextualised language, considered crucial for the communicative needs of the informants since they are school pupils. Hatch (1984) called for more attention to text rather than to syntax in research on second language learning, holding that ‘‘The acquisition of text types as discourse should become part of our description of interlanguage competence’’ (op.cit.: 197). Hatch further noted that ‘‘[. . .] even if much effort has been spent on the identification of cohesive ties, we know little about their distribution, scope or frequency [. . .]’’ (ibid.). In child language research considerable progress has been made in the past 10–15 years in this research area, where the narrative has become a frequently studied text type in acquisition research with a functional perspective, encompassing the acquisition of referential management through nominal cohesive devices. Child L1 acquisition research in this domain has now been carried out on various languages



Ingvild Nistov

yielding a wide range of crosslinguistic data.2 Some of these works also include analyses of children’s bilingual development. Investigations of adolescent and adult L2 acquisition and referential management in narratives are more rare however, and the studies that have been carried out have largely had their foundation in child L1 acquisition research. The seminal works of e.g. Karmiloff-Smith (1979, 1981, 1985), Hickmann (1982, 1987) and Bamberg (1987) are frequently referred to in discussions of the development of referential management also in adolescent and adult L2 learners’ language (cf. e.g. Pfaff 1987; Broeder 1991; Strömqvist and Day 1993; Lindberg 1995; Hendriks 1998). Command of discourse is regarded as a late L1 development, and adultlike solutions in reference are found to emerge around the age of ten (cf. Bamberg 1987). For adult and adolescent L2 learners, we may expect that they already have the cognitive capacities of obeying universal discourse principles (cf. Hendriks 1998). What one perhaps may question is whether the sociocommunicative skills needed in decontextualised language use are fully developed in adolescents — and for that matter also in adults — with only a few years of schooling. This question should be seen as pertaining to lack of social practice with discourse types (Clancy 1992) rather than to explanations in terms of cognitive deficits. We may assume however, that the main task of the adolescent L2 learner is a linguistic one. This task with regard to referential management involves not only the acquisition of the necessary language specific devices, but also appropriateness in the choice of the available referential expressions.

Referential devices in Turkish and Norwegian There are some basic differences between the Turkish and the Norwegian nominal systems that could influence the distribution of referential devices used for continuing reference in a narrative in the two languages. These differences pertain to the possibility of using zero-marking, the pronominal system and the morphological marking of indefiniteness and definiteness on the noun. From a syntactic perspective, Turkish, in contrast to Norwegian, is categorised as a so-called pro-drop language, as person and number are marked by verb-suffixes, and the grammatical subject does not have to be expressed. Pronouns in subject position in Turkish are generally used if some kind of

L2 narratives of Turkish adolescents in Norway

contrast or emphasis is wanted (Erguvanlı-Taylan 1986: 210; Johanson 1992: 254). In Norwegian the use of zero subjects is limited to finite coordinate clauses; in other contexts the subject is obligatory (cf. Mac Donald 1997: 141).3 It should be noted that in Norwegian overt marking of the grammatical subject is also obligatory in coordination containing conjunctive adverbs, where the subject is placed in postverbal position (cf. Nistov 1999). Another difference between Turkish and Norwegian is the fact that in Turkish grammatical objects can be omitted as well. To account for this phenomenon we have to go beyond the intrasentential explanation related to agreement (cf. Kerslake 1987: 97). Objects may be omitted with an anaphoric function if the antecedent can be easily identified from the discourse context. Moreover, third person singular pronouns as anaphoric expressions in Norwegian and Turkish incorporate differing amounts of information about their referents; in Norwegian: number, gender and to some extent animacy and human/non-humanness; in Turkish: only number. This fact accounts for a difference in the referential capacities of the third person singular pronouns in the two languages. The use of definite nouns in Norwegian as an option for continuation of reference can only be regarded as partly having a parallel in Turkish where the status of definite is only morphologically marked in the accusative case. In Turkish the distinction between new and given information is mainly expressed by word order.

Research questions The present chapter is concerned with L2 learners’ referential choice in the light of source and target language conventions. With a focus on contextdependent aspects of language, the above account of linguistic resources does not suffice; descriptions of the referential expressions in use in the specific narrative discourse contexts under scrutiny should be added. This is the rationale for including an analysis of reference material consisting of narratives from native Turkish and native Norwegian storytellers. The first question to be analysed in what follows is thus: I. What is the pattern for reference continuation in the source and the target languages as realised in the narratives of native and Turkish–Norwegian storytellers?





Ingvild Nistov

The answer to this question will in turn serve as a point of departure for the analysis of the main research question concerning the L2 learners’ referential solutions: II. What is the pattern for reference continuation in the narratives of the L2 learners? – How do the L2 learners approach the target language conventions for reference continuation to story characters? – Does the learners’ choice of referential expressions suggest transfer of L1 conventions? – Can any developmental tendencies be detected?

Method The analysis is based on a longitudinal case study where written narratives were elicited from three L2 learners. In addition, data collected at one point of elicitation from the two reference groups consisting of native speakers are analysed.

Informants The three main informants (henceforth called core informants) Ayfer, Birhan and Cafer are immigrants to Norway from Mid-Anatolia, Turkey. The majority of the immigrants from Turkey living in Norway, Turks as well as Kurds, are workers who came to Norway between the mid-sixties and mid-seventies,4 others came later as political refugees.5 In many cases the wives and the children of the early labour immigrants remained in Turkey and were only later reunited with their husbands and fathers. This is the case for the three core informants of the present study, who immigrated to Norway around 1991 at the age of 13–14. The core informants were chosen from two schools in Oslo which offered a special program for immigrant pupils from Turkey. The classes were for newcomers and were organised as language homogeneous groups in order to make a bilingual program possible. The pupils were offered tuition in Norwegian as a second language and Turkish, and other subjects such as social sciences, mathematics and science were taught on a bilingual basis — with both a bilingual teacher and a native Norwegian teacher. In addition to language


background, the other most decisive variables in selecting the informants were age, the duration of the pupils’ stay in Norway and their educational and social background in Turkey. Some of this information is given in Table 1. Table 1. Core informants Stay in Norway Ayfer (F) 12 months Birhan (M) 10 months Cafer (M) 10 months

School in Norway

School in Turkey

Age

9,5 months 8 months 8 months

5 years 5 years 5 years

15 15 14

At the first elicitation the three core informants were 14–15 years old and had been 10–12 months in Norway. All three informants chosen had attended the obligatory five years of primary school in Turkey. Afterwards they had been away from school for 1.5–2 years before coming to Norway. One of the reference groups consists of Turkish pupils in Turkey. These informants were of the same age as the core informants and were pupils at the beginning of the eighth school year at a state school in Istanbul. The school is situated in an area of Istanbul with a socio-economically mixed population, but the majority of the pupils come from lower middle class and working class families, most of them with their origin in Anatolia. The other reference group consists of Norwegian pupils from the school in Oslo where also one of the core informants was enrolled. This school is situated in an old working class area, where the population is, however, quite mixed as to socio-economic status at the present time. The elicitation of data took place at the end of the ninth school year. The two reference groups are referred to as TST (Turkish Speakers in Turkey) and NSN (Native Speakers of Norwegian).

Data and data collection Data from the core informants were collected in five elicitation sessions (henceforth also referred to as data points) over a period of 16 months, at intervals of four months. The elicitation of narratives from the core informants as well as from the reference groups took place in the classroom, where I was present as ‘‘a participant observer’’. Hence it was possible to control that the time spent on the task was approximately the same for all the pupils concerned, and that each pupil alone was responsible for his/her own production. The set of elicitation materials used consisted of The pear film and three short cartoon series.6 The stories elicited were given the following names: The





Ingvild Nistov

Pear Story, The Beach Story, The Shop Story and The Bird Story. A short description of the elicitation material with respect to the type of story characters involved in each set of pictures and the film is given in Table 2, which is inspired by Karmiloff-Smith’s (1985: 65) description of story types. The presence of characters of different sex or of the same sex is considered a crucial variable in this study. It should be noted that the presence of a twogender context in a story does not exclude the presence of a same-gender context in the same story and vice versa. However, some of the stories are more dominated by the one type of context than the other. A two-gender context prevails in The Beach Story, The Bird Story and The Shop Story, whereas in The Pear Story a same-gender referential context is more predominant. The same elicitation material was used throughout the elicitation period. This was done to keep the variable of linguistic (referential) context as constant as possible The elicitation procedures were identical both for the collection of the core and the reference data. The task was presented as an ordinary school task that the pupils were expected by their teacher to carry out. The cartoon series were given one by one to the pupils, and they were instructed to write a story based on the pictures, while having the pictures in front of them on their desks. The average time spent on each cartoon series was about 20 minutes. The film was shown to the pupils in a video room. Subsequently they returned to the classroom and were asked to write a story about what they had seen. The narratives used for analysis in this chapter were elicited through writing. Some linguists have pointed out that the distinction between written Table 2. Referential contexts in terms of referent types The Beach Story

The Shop Story

The Bird Story

The Pear Story

Number of characters

2 human 1 animal

3 human

3 human 1 animal

7 human 1 animal

Number of major characters

2

1

1 + 1 animal

2

Are there characters of different sex in interaction? (two-gender context)

yes

yes

yes

yes

Are there characters of the same sex in interaction (same-gender context)

no

yes

no

yes


and oral production may not always be the most relevant one to make, and that it may be a question of continua of variation rather than discrete poles (Biber 1989). The present data could for example be related to a continuum of unplanned vs. planned discourse (cf. Ochs 1979). As the written narratives used in this study are a result of rush-writing, they could be characterised as ‘‘unplanned spontaneous first draft output’’ and the question of their ‘‘writtenness’’ could be raised. With this in mind, some references for comparsion are made to studies based on oral narratives. This does not mean that the problems of making comparisons of results from oral and written production should be overlooked. Strömqvist (1996), who reports from a study on discourse flow and linguistic information structuring in spoken and written narratives, points to the fact that the production process as evidenced in writing on the level of on-line events is still poorly examined, and that more research is needed to be able to say how differences on the on-line level in speech and writing influence the spoken and written production in terms of linguistic information structuring (op.cit.: 3).

Data analysis Analytical framework The present analysis is concerned with reference as a textual phenomenon. An approach for the analysis of reference in a text linguistic framework has been described by Klein and von Stutterheim (1991) via the notion of referential movement. Figure 1 is a simplified variant of their presentation and may serve as an illustration of the text functions to be described. REFERENTIAL MOVEMENT position speci²ed before?

no

yes

INTRODUCTION

CONTINUATION relation to previous speci²cation maintenance

Figure 1.

shift





Ingvild Nistov

The tasks of the L2 learner that will be studied under the mentioned headings may roughly be described as follows: a. The L2 adolescent learner has to learn how to mark linguistically the introduction of new information in the text (Referent introduction). b. Furthermore the L2 learner has to learn how to express the distinction between new and given information and thus how reference is continued (Reference continuation). c. Thirdly, in continuing reference, the learner is faced with the more subtle task of choosing an appropriate linguistic expression to fit the referent’s place on a scale of givenness and importance (Givón 1983, 1995). This chapter will address the two latter tasks (b and c). There is no room in this presentation to go into ‘‘the incredible complexity’’, as Givón (1995: 341) characterises it, of what is involved in these tasks, or to go into the different theoretical views on how to approach this complexity. Explanatory approaches to reference in narratives may according to Fox (1987) be divided into those which see narratives as a continuous, linear string of clauses and those relying on global text hierarchical structure of narratives. The approach in this analysis can be briefly described as one mainly based on recency or distance (cf. Givón 1983) (as also applied by e.g. Bamberg 1987; Broeder 1991; Aarssen 1996; Lindberg 1995), where accessibility is measured in terms of maintenance and shift of reference to story characters. These notions are here used to describe discourse functions which are defined in relation to the flow of information in the piece of text analysed. Even if this kind of analysis may be said to be of the ‘‘look back type’’ and is linear more than hierarchical, the global perspective is also taken care of when we consider the thematic status or plot centrality of the protagonists and differentiate between major and minor characters. The analysis here will focus on reference to major characters. A problem is how to identify the ‘‘major’’ and ‘‘minor’’ characters of a story. A survey of the characteristic features valid for a major story character given in McGann and Schwartz (1988: 216) could be summarised as pointing out four main criteria: high degree of agency, high degree of animacy, high frequency of appearance and first appearance. The story participants identified as major characters in accordance with these criteria in the stories of this study are indicated in the Appendix together with the pictures used for elicitation.7


In his account of referential management in German L1 narratives of children and adults, Bamberg (1987) classifies the referential behaviour he finds into various referential strategies. The prototypical referential pattern in the adult narratives was found to be the anaphoric strategy, in which full nominal forms are the most predominant for reference shift and pronominal forms for reference maintenance. The anaphoric strategy, which is also considered as the adult-like referential behaviour in Lindberg (1995), may be regarded as the expected decontextualised solution by the adolescent L2 learners in continuing reference.

Coding procedures In the various studies with analyses of maintenance and shift of reference, one crucial problem appears: the lack of a unified format for segmentation and coding of the data. The ‘‘look back’’ perspective of this kind of analysis has already been mentioned, but there seems to be a lack of consensus on which item to look back to as well as on how far back in the discourse to look. In the present analysis I will largely follow the coding procedures as presented by Broeder (1991), who gives the following definitions of maintenance and shift of reference: • •

Shift of reference: those instances where another character is referred to than the one in the preceding utterance. Maintenance of reference: those instances where the same character is referred to as in the preceding utterance (op.cit.:136).

Even if these definitions may look quite straightforward at first sight, we are still to some extent left with the two problems mentioned above: I. What counts as an instance? (i.e., which item to look for) II. What exactly constitutes the preceding utterance? (i.e., how far to look back). With respect to the which item problem, I will follow Bamberg (1987) and Broeder (1991) in including all references to the story characters irrespective of syntactic function. Other studies restrict their analysis to referential expressions in subject position (cf. e.g. Aarssen 1996; Lindberg 1995).8 When it comes to how far to look back we arrive at the problem of defining the preceding utterance. In the present study, the preceding utterance is seen as





Ingvild Nistov

identical with the preceding clause.9 The definition of a clause follows largely Berman and Slobin (1994: 657): ‘‘By clause is meant a unit that contains a unified predicate. By unified, we mean a predicate that expresses a single situation.’’ This clause definition of a clause is illustrated in (1). (1) AYFER The Pear Story III Han er falt på jora # Noen gutter kommer og hjelper han (MAI)10 He has fallen on the ground # Some boys come and help him

The two finite verbs in (1) kommer og hjelper ‘come and help’ may be said to express one action. That is, according to the definition of a clause as a unified predicate, this sequence is treated as a single clause, which has consequences for the operationalisation of maintenance or shift of reference. Analysing the sequence as one clause makes the second use of the pronoun han ‘he’ a case of maintaining reference, since the character referred to is also mentioned in the preceding clause, whereas a segmentation into two clauses would give a case of reference shift here. Moreover, ‘‘preceding’’ does not necessarily imply the immediately preceding utterance (=clause) if this clause does not contain reference to a character that contributes to the thematic progress in the story. This is in accordance with Broeder (1991) and Bamberg (1986, 1987). Example (2) illustrates this coding procedure. (2) BIRHAN The Pear Story III datt han ned # og alle pærene ble borte # så satt han (MAI) og så på beinet sin fell he down # and all the pears disappeared # then he was looking at his leg

As can be seen from example (2), it is however, not so straightforward to decide which statements should be considered as contributing to the thematic progress. I have followed Bamberg (1986) in including only those statements containing a potential actor in the story, and not those with reference to other objects (like pærene ‘the pears’).

Results — research question I: Reference continuation in the source language and the target language A short sketch of the inventory of referential devices in the source and target languages was presented above. However, as already mentioned, in order to specify how the ‘‘rules’’ for referential management are realised in the specific narrative discourse contexts, an analysis of data from native speakers of


Norwegian (NSN) and the data from Turkish speakers in Turkey (TST) is in order. The results of the analysis of the TST narratives and the NSN narratives are given in Tables 3 and 4. Table 3. Reference to major characters in the TST narratives (in number of occurrences and percentages)11 Ref. Device

MAI

SHI

Full NP

84 (32.9%)

61 (87,1%)

Pronoun Ø

11 (4.3%) 160 (62.7%)

7 (10%) 2 (2.9%)

Pro-form

171 (67.1%)

Total

255

9 (12.9%) 70

Ø=zero-marking, pro-form pronouns and zero-markings, MAI=maintenance of reference, SHI=shift of reference. These abbreviations are also applied in the following tables.

Table 4. Reference to major characters in the NSN narratives (in number of occurrences and percentages) Ref. Device

MAI

SHI

Full NP

23 (9.5%)

53 (74.6%)

Pronoun Ø

163 (67.6%) 55 (22.8%)

18 (25.4%)

Pro-form

218 (90.5%)

18 (25.4%)

Total

241

71

As can be seen in Tables 3 and 4, the anaphoric strategy with a predominant use of pro-forms in the maintenance function and full NPs in the shift function is confirmed for both languages. The results for the Norwegian reference data (NSN) are largely in line with the anaphoric pattern presented for other Germanic adult narrative data; for Swedish by Lindberg (1995: 70), for German by Bamberg (1987: 57) and for Dutch by Broeder (1991: 128). As might be expected with respect to the type of pro-form (zero or pronoun) used in the maintenance function, there is a clear tendency in the TST data for a predominance of zero-marking, whereas the NSN data reveal a larger proportion of pronouns. This indicates that in a third person narrative, the zero in Turkish will often correspond to the anaphoric function of pronouns in Norwegian. It should be noted that all but one of the pronominal instances of reference continuation in the TST data occurred in non-subject position. Moreover, in the shift function the TST data show less use of pronouns than the NSN data. This result is also according to expectations, since the elicitation





Ingvild Nistov

material (cf. Table 2) contains quite a few two-gender referential contexts, where the use of pronouns in Norwegian — in contrast to Turkish — does not cause ambiguity.

Results — research question II: Reference continuation in the L2 learners’ narratives Having looked at patterns for reference continuation in the source and target languages of the learners, we now turn to the results of the analysis of research question II concerning the L2 learners’ referential management. The focus is on how the L2 learners handle decontextualised referential solutions for reference continuation in terms of the anaphoric strategy, and moreover, on how the L2 learners whose L1 is Turkish approach this strategy as described above for the target language Norwegian. The linguistic challenge of the L2 learners seems in particular to be related to the differences between the source language and the target language in the distributional pattern of zero-marking having to do with the stronger syntactic constraints in Norwegian, and to the differences in the referential capacity of third person pronouns when it comes to expressing gender distinction. Given that the L2 learners might apply ‘‘rules’’ or conventions from the source language, one could predict that the L2 learners will use more zero-markings for reference maintenance than the NSN group. Furthermore, due to the frequency of two-gender referential contexts in the narratives under scrutiny, one could also predict that the L2 learners will use more full NPs for reference shift than the NSN group. If we assume that the L2 learners gradually adopt the target language conventions, a development may be expected where such possibly transferred L1 conventions in their interlanguage will be replaced by patterns more similar to those of the target language.

Ayfer In Table 5 the analysis of reference to major characters in Ayfer’s narratives is presented. Ayfer uses only pro-forms, mostly pronouns, for maintaining reference, and so far this points to the anaphoric strategy. It is striking that in Ayfer’s stories there is not a single occurrence of a full NP in maintaining reference to major (human) characters in any of the narratives. However, also for the shift function, we find an overall higher proportion of pronouns than of full NPs in reference to major characters at all stages, ranging from 63.6 per


Table 5. Maintenance and shift of reference to major characters in Ayfer´s narratives (in number of occurrences and percentages) Data point

Function

Full NP

Pronoun

Ø

Pro-form

Total

I

MAI SHI

15 (39.5)

3 (33.3)

23(60.5) 6 (66.7)

38 (100) 6 (66.7)

38 9

MAI SHI

26 (68.4) 10 (71.4)

12 (31.6)

4 (28.6)

38 (100) 10 (71.4)

38 14

MAI SHI

22 (88) 5 (83.3)

3 (12)

1 (16.7)

25 (100) 5 (83.3)

25 6

MAI SHI

32 (82) 9 (75)

7 (18)

3 (25)

39 (100) 9 (75)

39 12

MAI SHI

27 (79.4) 7 (63.6)

7 (20.6)

4 (36.4)

34 (100) 7 (63.6)

34 11

II III IV V

cent to 83.3 per cent, whereas the average proportion in the NSN data was found to be as low as 25.4 per cent. Ayfer’s preference for pronouns is exemplified in the following text extract from the first elicitation, which shows maintenance and shift of reference to the two major characters in The Pear Story, the pear picker and the bike boy (see Appendix): (3) AYFER The Pear Story I En gutt kommer12 og han (MAI) har sykkel han (MAI) ta kurv og Ø (MAI) gå men de andre han (SHI) vet ikke han (MAI) bare plukke pærer Han (SHI) ta kurv men han (MAI) treffe stain og Ø (MAI) fale på gulve

A boy comes and he has bike he take basket and Ø go but the other he does not know he only pick pears he take basket but he hit stone and Ø fall on the floor13

The same pattern can be observed in the parallel text sequence from the elicitation session one year later: (4) AYFER The Pear Story IV Den gutten (SHI) som kjører sykkel han (MAI) stjålet en kurv og Ø (MAI) gå sin vei Men han (SHI) vet ikke han (MAI) bare plukket pære

That boy who drives bicycle he stolen a basket and Ø go away But he does not know he only picked pear





Ingvild Nistov

These examples reveal that the preference for pronouns is stable throughout the elicitation period. The anaphoric strategy with pro-forms most commonly reserved for maintaining reference and full NPs for reference shift is not found in Ayfer’s stories in any of the data points. As can be observed in (3) and (4), Ayfer’s referential choice with the excessive use of pronouns, leads to ambiguity in this same-gender context of The Pear Story and cannot be said to be represent an adequate decontextualised solution. The analysis of Ayfer’s narratives gives no clear indication of transfer of L1 conventions in terms of frequency. The prediction that the L2 learners will use more full NPs and less pronouns in the shift function than what was found in the NSN group is not supported, since Ayfer in fact overall uses pronouns more often than the NSN group for reference shift. Furthermore, in contrast to the predominance of zero-marking in the maintenance function — as shown by the TST data is the case for Turkish — Ayfer on the whole makes more use of pronouns. In the last three data points the proportions of zero forms in Ayfer’s narratives are in fact lower than the mean proportion shown for the NSN data.The results in the two first data points from Ayfer may, however indicate L1 influence since she here does exceed the proportion of zero forms for reference maintenance found in the NSN data. One possible explanation for Ayfer’s excessive use of pronouns is that she may not have fully acquired the target language convention of having zero marked subjects in coordinated finite verb constructions, and that she instead uses pronouns as a substitute for the obligatory agreement marking of the verb which she knows from her L1. Ayfer does have cases of zero-marking with coordinate structures already in the first data point, as can be seen in example (5) (marked with →): (5) AYFER The Beach Story 1 og han legge seg på gulve → og Ø (MAI) lese på boka [. . .] Hun er gjemme på tre → og Ø (MAI) kast vann til guttten

and he lie down on the floor and Ø read in the book [. . .] She is hide on tree and Ø throw water to the boy

However, the adequate use of zero-marking in such contexts increases in the last data points as illustrated by the following two examples: (6) AYFER The Bird Story II Hun ser noen kjenner henne → Hun snakke med dem

She sees somebody knows her She talk with them


(7) AYFER The Bird Story IV Hun så på gammel vennene sine → og Ø begynt å snakke med dem

She looked at her old friends and Ø begun to talk with them

These two extracts in (6) and (7) show how an identical story event from The Bird Story is recounted in data points II and IV, and illustrates how Ayfer’s cohesive strategy changes from a repetitive use of pronouns to use of zeromarking in coordinate constructions. If we look more closely at the use of zero forms in Ayfer’s stories however, we also find occurrences of zero-markings in early as well as later stages that do not accord with target language constraints, as shown in example (8) from the second data point: (8) AYFER The Beach Story II Han er litt redd først Men etter på stå *Ø (MAI) opp skrike og rope

He is a bit afraid first But afterwards Ø get up shout and scream

The use of zero form here marked as *Ø is at variance with syntactic constraints in Norwegian.14 Zero-marking in the ‘‘equivalent’’ context in Turkish would, however, be highly adequate, and L1 conventions could be a factor which has influenced Ayfer’s choice of referential expression here. Questions of indirect L1 influence should also be looked into in the attempt to explain Ayfer’s preference for pronouns. In the extract from Ayfer’s Pear Story presented in (3) and repeated below in greater length as (9), the occurrence of pronominal reference shift (marked out with →) could complicate the picture we have got of Ayfer’s strategies so far. (9) AYFER The Pear Story I En gutt kommer og han (MAI) har sykkel han (MAI) ta kurv og Ø (MAI) gå → men de andre han (SHI) vet ikke han (MAI) bare plukke pærer Han (SHI) ta kurv men han (MAI) treffe stain og Ø (MAI) fale på gulve de adre gutter den er kameraten hans hjelpe han(MAI) og Ø går

A boy comes and he has bike he take basket and Ø go but the other he does not know he only pick pears he take basket but he hit stone and Ø fall on the floor the other boys that are his friend help him and Ø go





Ingvild Nistov

→ De andre mann (SHI) plukke pærer og Ø (MAI) stige og Ø (MAI) ser

The other man pick pears and Ø ladder and Ø looks

This sequence de andre han ‘the other he’ could be interpreted as juxtaposed de andre — han [the other — he], the pronoun being right-dislocated, or as one expression, de andre han [the other he], which is the analysis chosen here. One factor supporting this solution is found in a construction further on in the text: De andre mann ‘the other man’, which could be seen as a parallel structure. In that case the sequence de andre han could indicate that Ayfer interprets third person pronouns rather like lexical entities.15 The present analysis is focused on reference to human story characters. It is worth noticing, however, that if we add the bird in The Bird Story to the list of protagonists, we do not find this preference for pronouns in Ayfer’s production. Ayfer maintains and shifts reference to this story participant by full NPs, as can be seen in example (10). (10) AYFER The Bird Story IV Han åpnet fuglebur og fuglen (SHI) blir borte Og hun så fuglen (MAI) Hun skjønner ikke hvordan fuglen (SHI) kom ut Fuglen (MAI) står opp i tre

He opened bird cage and the bird vanishes And she saw the bird She does not understand how the bird got out The bird stands up in tree

This pattern of referring to the bird by full NPs can be observed in all data points. Seen against the background of Ayfer’s consistent preference for pronouns in referring to human characters, this finding is startling, and indicates that the variable ±human might also be an influencing factor on the choice of referential expressions (cf. Nistov 2000).

Birhan The analysis of Birhan’s narratives is presented in Table 6. With regard to reference maintenance to major characters, Table 6 shows that in Birhan’s narratives the findings of pro-forms in all data points ranging between 82.1 per cent and 100 per cent come close to the mean proportion for pro-forms as found in the NSN data, i.e. 90.5 per cent. In the reference shift function the proportions of pronouns are somewhat higher in Birhan’s narratives than what was displayed in the NSN material, especially in the last data point,


Table 6. Maintenance and shift of reference to major characters in Birhan’s narratives (in number of occurrences and percentages) Data point

Function

I

MAI SHI

II

Full NP

Pronoun

Ø

Pro-form

Total

15 (46.9)

7 (63.6)

17 (53.1) 4 (36.4)

32 (100) 4 (36.4)

32 11

MAI SHI

5 (17.9) 8 (57.1)

19 (67.9) 5 (35.7)

4 (14.3) 1 (7.1)

23 (82.1) 5 (42.9)

28 14

III

MAI SHI

3 (9.1) 8 (66.7)

26 (78.8) 4 (33.3)

4 (12.1)

30 (90.9) 4 (33.3)

33 12

IV

MAI SHI

2 (5.3) 10 (71.4)

28 (73.7) 4 (28.6)

8 (21.1)

36 (94.7) 4 (28.6)

38 14

V

MAI SHI

20 (64.5) 6 (46.2)

11 (35.5)

7 (53.8)

31 (100) 6 (46.2)

31 13

which shows 46.2 per cent pronouns. Even so, we can conclude that Birhan’s solutions are much closer to the anaphoric strategy as displayed in the NSN data, than Ayfer´s. As regards the prediction that the convention of zero-markings for the maintenance function in the learners’ L1 may be transferred to the L2, the results from Birhan in data point I are interesting; there Birhan has 46.9 per cent zero forms which is a much higher proportion than was found in the NSN data (22.8 per cent zero forms). Thus, the quantitative analysis here indicates a possible influence from the L1 convention. After the first data point there is a decrease in zero-markings to below the average proportions found for the NSN group, until the last data point, when an increase again can be observed. It should also be noted that eight of the 15 zero form occurrences found in Birhan’s production in the first data point are non-acceptable according to target language syntactic constraints. This is illustrated in examples (11) and (12). (11) BIRHAN The Pear Story I Da han komme neders tømme *Ø (MAI) pærene i en stor kurv

When he came down Ø emptied the pears into a big basket

(12) BIRHAN The Pear Story I og han så pære så stopper *Ø (MAI) med en gang

and he saw pear then Ø stops at once

It can be seen that Birhan, like Ayfer, uses zero-marking in contexts which are





Ingvild Nistov

at variance with syntactic constraints in Norwegian, but which would be adequate as zero anaphora in Turkish. Example (11) shows zero-marking with subordination, whereas example (12) illustrates zero-marking in coordination with conjunctive adverb and is parallel to (8) from Ayfer’s production. Contrary to the tentative prediction concerning L1 influence and reference shift, the proportion of pronouns in the shift function in Birhan’s narratives throughout the elicitation period is on the whole higher than what was found in the NSN data. Not all of Birhan’s referential solutions are adequate as seen in the following Pear Story extract from the first elicitation, where Birhan’s referential choice does not discriminate between the bike boy and the paddleball boy: (13) BIRHAN The Pear Story I men en gutt en av dem finner hatten hans og så stoppern (MAI) å Ø (MAI) plystre → så gutten (SHI) ser bak → gutten (SHI) sa er det din hatt → han (SHI) sa ja ‘‘det er min hatt’’ Og da gutten (SHI) får tre pære av den gutten (MAI) og så gikk dem

but one boy one of them finds his hat and then he stops and Ø whistle then the boy looks back the boy said is it your hat he said yes ‘‘it is my hat’’ And then the boy gets three pear from that boy and then they went

The violation of the anaphoric strategy by using a pronoun to shift reference in this same-gender context renders ambiguity. Furthermore, we see that both characters are shifted reference to by the common noun gutten ‘the boy’. This example illustrates that in this type of referential context, in which the interacting story characters are of both the same gender and of the same age group, i.e. ‘‘same-type participants’’ (Brown and Yule 1983), the anaphoric strategy of applying full NPs for reference shift does not work, since the common noun used is also ambiguous. Since the proportion of pronominal shifts is particularly high in the last data point (46.2 per cent), it may be worth looking more closely at Birhan’s use of pronouns here. Example (14) contains an instance of a pronoun used for the shift function.


(14) BIRHAN The Pear Story V men det var tre gutter som var der og de hjalp han (SHI) for samlet alle pærene

but there were three boys who were there and they helped him to collect all the pears

This occurrence is a shift from a plural to a singular referent. A pronoun is here an adequate choice of referential expression in the target language. Accordingly we have a context where the anaphoric strategy of applying proforms for maintenance and full NPs for shift is not required. A look at other occurrences of pronouns for reference shift in the data from the last elicitation shows the following: (15) BIRHAN The Shop Story V så stakk han med en gang → etter hun (SHI) ble ferdig med handlingen *Ø kom til kassa sa kassadamen det er 100 kr → Men hun (SHI) fant ikke lommeboka si (16) BIRHAN The Beach Story V Gutten reiste seg opp → så pekkte han på henne (SHI)

then he ran away at once after she had finished shopping Ø came to the cash register said the woman cashier it is 100kr But she did not find her wallet

The boy got up then he pointed at her

The first occurrence of pronoun shift in The Shop Story in (15) and the one occurrence from The Beach Story in (16) must be seen as related to a twogender participant context, and pronouns are here adequate expressions for reference shift in the target language. When it comes to the second instance of reference shift in (15), the choice of pronoun is less successful. As the lexical NP in the preceding clause is referring to a woman (kassadamen ‘the woman cashier’), the use of the feminine third person pronoun for the shift function here creates a context of potential ambiguity (cf. Givón 1983). However, compared to the same-gender context in example (13) commented on above, the semantic components of the predicate in this case reduce this risk of ambiguity (see also comments on example (21) from Cafer).





Ingvild Nistov

There seems to be no clear developmental tendency in Birhan’s choice of referential expressions throughout the elicitation period.

Cafer In Table 7 the analysis of Cafer’s narratives is presented. Table 7 shows that Cafer most commonly expresses the maintenance function by pro-forms, which is according to the pattern described for the anaphoric strategy. In data points III and IV the proportions of pro-forms for maintenance are very close to the described target language pattern. However, especially in the first and last data points, there is a remarkably higher proportion of full NPs in the maintenance function than what was found in the NSN data (see comment on (24)). As regards shift of reference, Cafer, like Ayfer and Birhan, applies more pronouns than full NPs except for the first data point, and in the last data point there is an even distribution of pronouns and full NPs. The proportions of pronouns are in all data points higher than what was found for the anaphoric pattern in the NSN narratives. In data point IV the results for reference shift came out as high as 68.4 per cent pronouns. Hence Cafer, although keeping to the anaphoric strategy, shows a deviance from the ‘‘prototypical’’ pattern that was established for the NSN data, both by applying fewer pro-forms and more full NPs for the maintenance function and by using more pronouns and fewer full NPs for the shift function. Table 7. Maintenance and shift of reference to major characters in Cafer’s narratives (in number of occurrences and percentages) Data point

Function

Pronoun

Ø

Pro-form

Total

I

MAI SHI

8 (21.1)a 6 (60)a

23 (60.5) 4 (40)

7 (18.4)

30 (78.9) 4 (40)

38 10

II

MAI SHI

6 (13.3) 8 (47)b

25 (55.5) 9 (53)

14 (31.1)

39 (86.7) 9 (53)

45 17

III

MAI SHI

4 (7.8) 7 (46.7)

34 (66.7) 8 (53.3)

13 (25.5)

47 (93.2) 8 (53.3)

51 15

IV

MAI SHI

7 (8.3) 6 (31.6)

48 (57.1) 13 (68.4)

29 (34.5)

77 (91.7) 13 (68.4)

84 19

V

MAI SHI

14 (22.6) 9 (50)

31 (50) 9 (50)

17 (27.4)

48 (77.4) 9 (50)

62 18

a

Full NP

Of the MAI instances in elicitation I, three were proper nouns; of the SHI instances there was one proper noun. b In data point II, one (1) instance of the full NPs was a proper noun.


The first prediction concerning L1 influence in terms of high proportions of zero-marking is not confirmed by the quantitative analysis. Cafer’s use of zero forms for maintaining reference is on the whole quite close to the percentages found for the NSN group. In data point I, the number of zero forms is — in sharp contrast to what was found for Birhan — remarkably low with only 18.4 per cent zero-marking. Only in the last two data points are the proportions somewhat higher than for the average percentage found for NSN. Still Cafer, like the other two informants, has a few occurrences of nontarget-like zero-markings (*Ø) for reference maintenance. One example is given in (17). (17) CAFER The Pear Story III Etter en stund gikk de guttene gjennom han # og *Ø (MAI) bare så på den

After a while those boys went through him # and Ø only looked at them

The referential movement in example (17) is from a non-subject to a subject referent in the following clause. Zero-marking here is inappropriate in Norwegian. In accordance with the syntactic rules for coordination in Norwegian, one would have to interpret the zero as referring to de guttene ‘those boys’, whereas we know from the film that the reference must be to the pear picker. Hence Cafer´s referential choice here in fact leads the reader to a wrong conclusion. In Turkish, however, a zero form would not lead to any ambiguity in this context since the verb would be marked for singular or plural. The L1 convention might thus be a factor influencing Cafer’s inadequate choice in this case. The second prediction concerning reference shift and L1 transfer is not supported in the quantitative analysis either, since Cafer, as already mentioned, does not use more full NPs but on the contrary more pronouns for reference shift than the NSN group. In data point IV the proportion of pronouns for reference shift is especially high (13 out of 19 occurrences). It might be worth looking more closely at them to try to understand why we get this high proportion of pronouns for shifting reference, and whether they cause ambiguities in the text. It turns out that five of the pronominal shifts occur in a two-gender context, which in Norwegian renders no risk of ambiguity (cf. example (16) from Birhan). This can be seen in the following examples:





Ingvild Nistov

(18) CAFER The Beach Story IV Gutten så søstra si og Ø sa til henne at hun må komme frem og han (SHI) sa at hun (SHI) må ikke gjøre det igjen

The boy saw his sister and Ø said to her that she should come out and he said that she should not do it again

(19) CAFER The Bird Story IV Han klatret ned og Ø gav den til henne (SHI)

He climbed down and Ø gave it to her

Moreover, some pronominal shifts in data point IV can be ascribed to referential contexts involving a singular and a plural referent, as illustrated in example (20): (20) CAFER The Pear Story IV Dem plukket alle pærene ned i bakken og Ø puttet dem inni kurven igjen Han (SHI) satt kurven opp på sykkelen igjen

They picked all the pears and Ø put them inside the basket again He put the basket on the bike again

The use of a pronoun in this type of referential context is, as previously mentioned, an adequate solution in the target language. There is, however, one occurrence of pronominal reference shift in Cafer’s narratives from data point IV which is less successful. This is in a same-gender context in The Shop Story, where the major character the lady has reached the cash register, and there is interaction between the woman cashier and the lady, as shown in example (21): (21) CAFER The Shop Story IV Hun som sitter på kassen slådd på prisene og Ø sa hvor mye det kostet Hun (SHI) skulle ta ut lommeboka si i lomma for å betale

She who sits on the cash register punched the prices and Ø said how much it cost She was going to take out her wallet in the pocket in order to pay

Pronominal reference shift in this same-gender context causes potential


ambiguity. However, the semantic context is of help to the reader who can infer from experience and ‘‘world knowledge’’ that it is not the cashier who is looking for the wallet. This is also a factor to be taken into account when evaluating this instance as adequate or not. As we saw in example (15), Birhan also had problems with continuing reference in this scene of The Shop Story. Another same-gender context — but with ‘‘same-type participants’’ — was illustrated in example (13) showing an inadequate solution from Birhan. In example (22) it can be observed how Cafer has handled this referential context in data point III. (22) CAFER The Pear Story III En av dem så hatten hans og Ø (MAI) tok den og Ø (MAI) plystret til han (SHI) Han (MAI) stod og så bak seg Han (SHI) leverte hatten til han (MAI) og fikk han (MAI) tre pærer

One of them saw his hat and Ø took it and Ø whistled at him He was looking back He delivered the hat to him and he got three pears

The use of pronouns for maintenance and shift of reference to both of the male characters here makes the rendering of this scene highly ambiguous. One may be inclined to ascribe the L2 learners’ problems in expressing the kind of referential movement involved in this episode to a lack of linguistic competence in their L2. However, in the NSN data we could also observe similar problems, in fact it was difficult to find one good example which could serve as a ‘‘normative’’ illustration of full referential control among the NSN informants in this same-gender context. This lack of control is illustrated in (23). (23) NSN OLA The Pear Story Gutten (SHI) sykler videre og de tre andre guttene går den andre veien En av de tre guttene (MAI) finner hatten til sykkelisten Han (MAI) plustrer på ham (MAI) og Ø (MAI) løper til han (MAI) med hatten Da får han (MAI) tre pærer en til dem vær

The boy cycles further and the three other boys go the other way One of the other boys finds the hat of the cyclist He whistles at him and Ø runs towards him with the hat Then he gets three pears one for each of them





Ingvild Nistov

In the text written by Ola we can see that there is ambiguity as to the identity of the two referents encoded by pronouns in the last three clauses, even if pronouns are here used to maintain reference. Cafer’s use of full NPs for the maintenance function was mentioned above. This usage is most outstanding in the last data point, as is illustrated in (24), an extract from The Pear Story: (24) CAFER The Pear Story V Det var en gang en mann som plukket pærer Mannen (MAI) klatret opp og Ø (MAI) begynte å plukke pærer Mannen (MAI) hadde på seg en hatt Mannen (MAI) puttet pærene i en forkle og Ø (MAI) klatret ned og Ø (MAI) tok pærene inn i kurvene Etter at han (MAI) puttet pærene inn i kurvene klatret mannen (MAI) opp på treet igjen

Once upon a time there was a man who picked pears The man climbed up and Ø began to pick pears The man wore a hat The man put the pears in an apron and Ø climbed down and Ø took the pears into the baskets After he had put the pears into the baskets the man climbed up the tree again

This repetitive use of full NPs (common nouns) for reference maintenance does not follow target language conventions. It might be asked whether the seemingly overexplicitness by use of full NPs reflects a less developed perception of the factor of referential distance in relation to accessibility (cf. Givón 1983). However, a lack of perception of the discourse principles that need to be adhered to seems to be a very implausible explanation at this stage, since Cafer has demonstrated in earlier stages that he has developed this capacity to express decontextualised referential solutions. A pattern like this rather reminds us of ‘‘Turkish narrative style’’,16 where there is a tendency to use neutral nouns like ‘man’ (Turkish adam) or ‘child’ (Turkish çocuk) as anaphoric devices in a ‘‘pronoun-like way’’, a style used by Cafer in this oral Turkish narrative from the same point of elicitation:


(25) CAFER The Beach Story (oral Turkish) V çocuk iyice ıslanmıs¸ tı çocuk (MAI) kalktı ve Ø (MAI) etrafında baktı kim yaptıdiye çocuk (MAI) çok kızmıs¸tı

the child got rather wet the child got up and Ø looked around saying who did it the child got very angry

If we go back to the results of the TST narratives shown in Table 3, a finding not yet commented on should be noted. It can be observed that there is also a strikingly high proportion of full NPs (32.9 per cent) in the maintenance function in the Turkish narratives, whereas the native Norwegian data showed only 9.5 per cent. This referential strategy is demonstrated by the text from Funda in the TST group: (26) TST FUNDA The Pear Story Çocuk bisikletiyle yola çıktı Biraz ilerledikten sonra kars¸ı dan bir kızın bisikletle geldig˘ini gördü Çocuk (MAI) kıza bakıyordu kız çocug˘un (MAI) yanından geçti ve çocuk (MAI) hala kızın arkasından bakıyordu Çocuk (MAI) baktı ama sonra bisikleti bir tas¸a çarparak düs¸tü

The child went out on road with his bike After had moved forward saw that a girl with bike came along the child looked at the girl the girl passed by the child and the child was still looking at the girl ‘s back the child looked but then his bike hit a stone and fell

This L1 convention may be one factor which has influenced Cafer’s choice of referential expressions, as shown in (24). It may be that he for a moment switched to the option for maintaining reference in the way that is common in his L1, where the use of unmarked lexical NPs like mannen ‘the man’ may correspond to pronoun use in Norwegian. This referential choice, which is at variance with the described anaphoric strategy, could not be predicted on structural grounds or in relation to discourse constraints, but might have to do with an L1 option of rhetorical style transferred into the L2.





Ingvild Nistov

Like for the two other learners, there seems to be no significant developmental pattern from the first to the last elicitation with respect to reference continuation to major characters in Cafer’s narratives.

Conclusions and discussion In this chapter we have examined how three L2 learners Ayfer, Birhan and Cafer express continuation of reference to story participants as part of their referential management in producing Norwegian L2 narratives. The L2 learners’ referential solutions have been analysed in the light of the source and target language patterns established for the specific discourse contexts through the analysis of narratives from Turkish and native Norwegian speakers. The referential solutions of the three L2 learners were largely found to be decontextualised. That is, violations of the anaphoric strategy in many cases turned out to be adequate solutions in the specific referential context. Moreover, it was suggested that some inadequate solutions could be ascribed to possible influence from L1 conventions, and they were thus interpreted as due to a linguistic deficit on the part of the learners, rather than due to lack of socio-communicative or cognitive skills. The inadequate solutions in some same-gender contexts might be an exception in this respect. It should be noted, however, that these referential contexts also proved difficult for the native Norwegian group. This suggests that the difficulties the L2 learners showed in handling same-gender referential contexts are due to a general heavy cognitive processing load and not that they lag behind their native peers with respect to decontextualised language use. The fact that only one source language is represented in the present study naturally sets a limitation with respect to the conclusions which can be drawn with regard to L1 transfer. One tentative prediction was that the L2 learners would use more zero-markings than the native Norwegian narrators. The use of zero forms for maintenance of reference was much higher in Ayfer’s and especially Birhan’s first data points than was found for the NSN group. This is the only finding that could suggest L1 influence in terms of frequency. Still, the analysis revealed that all three learners used zero-markings in contexts which are not acceptable according to syntactic constraints in the target language, but which would be acceptable in ‘‘equivalent’’ structures in their L1. A second prediction was that the L2 learners would use fewer pronouns than the NSN group in the shift function since the L1 pattern does not allow


for this solution. The overall quantitative analysis of all stories does not support this prediction as pronouns are quite extensively used for reference shift by the three L2 learners; the proportion of pronouns is even higher than what was found in the native Norwegian data. One finding of the analysis, which could not be predicted on structural grounds, was Cafer’s repetitive use of full NPs in the maintenance function, for which L1 influence in terms of transfer of rhetorical style was suggested as an explanation. Moreover, with respect to Ayfer’s extensive use of pronouns, it was indicated that L1 influence may play a role in a more indirect way. There might be reason to suggest that the process of pronominalisation is not necessarily the same for an L2 learner of Norwegian with Turkish as L1, as for a native speaker of Norwegian. That is, when Ayfer applies the word han ‘he’ and Ola from the NSN group applies the same word, this is not necessarily an expression of the same cognitive representation (cf. Karmiloff-Smith 1981). With respect to development and possibly L1 influenced referential solutions, the analysis indicates individual variation and also suggests that transfer at different levels may occur at different stages in the acquisition process. The prediction that L1 influence would be strongest in the beginning of the elicitation period can be said to be confirmed for the use of zeromarkings by Ayfer and Birhan, whereas the results for Cafer rather indicate that L1 influence in terms of rhetorical style may have taken place later. One the whole, no clear developmental tendencies for referential management could be detected for the three L2 learners during the 16 months of elicitation. Lack of development was also reported by Broeder (1991) and Hendriks (1998) in their adult L2 data. Broeder concludes, however, that the learners throughout make use of a global protagonist strategy, by reserving the use of pronouns in the shift function to reference to major characters. The present analysis does not give any indication that this is true for the informants in this study. Ayfer, whose choice is most at variance with the anaphoric strategy, also uses pronouns for shift of reference to minor characters in the stories (cf. Nistov 2000). The results show no correlation with the findings reported in Klein and Perdue (1992), for ‘‘the basic variety’’ with respect to the development of the third person pronoun system in adult L2. One of the generalisations made about anaphoric reference by Klein and Perdue for early adult learner language (the basic variety) is that ‘‘Definitely referring NPs are generally used before overt pronouns’’ (op.cit.: 318). Referential behaviour (extensive use of





Ingvild Nistov

full NPs) ascribed to the basic variety stage for anaphoric reference was for one learner (Cafer) in the present study found in stages when the learner was clearly ‘‘post-basic variety’’ in terms of inflectional patterns. Moreover, one learner (Ayfer), who in the first data points could be said to be close to an interlanguage system in accordance with the basic variety inflectionwise, extensively produced pronominal referential solutions. One factor of variation which has been focused in this chapter, and which in my view lacks sufficient attention in earlier dealings with referential management is that of story type, in terms of whether the protagonists are acting as single characters or in groups (producing singular or plural referents), whether the interacting protagonists are of the same gender or not (same-gender vs. two-gender contexts), and also whether they are animals or humans. This context dependency is reinforced if seen in a crosslinguistic perspective, as was here illustrated for Turkish and Norwegian. What from the quantitative analysis may seem like ‘‘violations’’ of the anaphoric strategy may in fact turn out to be adequate choices in some contexts, for example when a pronoun is used for the shift function in a two-gender context in Norwegian. Moreover, the adherence to the anaphoric strategy in terms of using full NPs for the shift function turned out not always to produce adequate decontextualised solutions, as illustrated in ‘‘same-type contexts’’ where protagonists of the same gender and age group were referred to by identical common nouns. Furthermore, it has been suggested that neither discourse constraints nor syntactic constraints, or rhetorical style may always fully account for referential choice. The results for Ayfer suggested that a different explanatory factor related to the ontological status of the story characters (cf. Fraurud 1996) in terms of ±humanness may be at issue. The factor of cognitive ontology seems to deserve more attention and should be accounted for when results of studies based on narratives which do or do not have animal/human story characters are compared. This indicates how essential the factor of story type may be when accounting for the referential patterns found. Accordingly large scale acquisition studies based upon the same story may be a fruitful way of getting a grip on these problems. On the other hand, however, one could wonder whether results obtained from analyses of one and the same story have any bearing beyond the mapping of the evolution of referential management in one specific story context. The short discussion of coding procedures included in this chapter seeks to illustrate the fact that maintenance and shift of reference should not be


regarded as given categories. What we have here is rather the analyst’s reconstruction of processes that can only grasp part of the complexity involved in the L2 learners’ choice and acquisition of referential devices. An attempt is also made to demonstrate that the analysts who apply the linear and sequential ‘‘look-back’’ approach which is implied in the description of maintenance and shift of reference in fact are not always concerned with the same kind of referential movement since there appears to be a great deal of variation as to how far back and at what items to look. This kind of variation in coding procedures, which is often disregarded in the literature, should be borne in mind when the results from studies of referential management are compared. The findings for the three L2 learners point to a high degree of individual variation. It should be noted that individual variation was also found in the narratives produced by the informants in the reference groups. The fact that in learning how to express reference continuation in the target language, the L2 learner seems to a large extent to be faced with optionality and preferences in language use may, as Berman (1999) points out, present him/her with problems other than those posed by the acquisition of ‘‘core grammatical elements’’ (op.cit:198ff). What we are concerned with seems rather to be the acquisition of what Berman calls ‘‘rhetorical expressiveness’’. However, as Berman also points out, the different levels of nativeness, like core grammatical elements and rhetorical expressiveness are often interlinked. This has here been shown to be the case with referential management, since the L2 learner’s choice between nouns, pronouns and zero forms seems to be constrained by universal as well as language specific factors of syntactic, semantic, discourse and stylistic character.

Notes . I am grateful to Victoria Rosén and Kirsten Meyer Bjerkan for valuable comments on this presentation, to Patrick Chaffey for his indispensable corrections of my English and to Sinan Çorbacıoglu ˇ for help with document processing. The presentation was made possible with support from The Norwegian Research Council. . E.g. for French (Karmiloff-Smith 1981, 1985), German (Bamberg 1987; Hendriks 1998), English (Wigglesworth 1992), Japanese (Clancy 1992), Chinese (Hendriks 1998) and also — as particularly interesting for our purposes — for Turkish (Verhoeven 1988, 1993; Schaufeli 1991; Küntay 1992; and Aarssen 1996). . It should be noted that we are here referring to third person narrative discourse. In





Ingvild Nistov

conversational discourse, the constraints for subject deletion will be different, where Norwegian colloquial style allows for omission of first person pronouns. . In 1975 an Immigration Act was passed prohibiting new labour immigration for foreign citizens. However, family reunion regulations allowed family members of the labour immigrants to continue to immigrate to Norway up till the present. . According to the population statistics in 1994 (the year of the last session of data collection) there were 7,766 inhabitants in Norway of Turkish origin, and of these 2,000 were between 7 and 19 years old. By 1 January 1999 the number of persons with Turkish origin had increased to 9,859. This may seem like a very small group, but in fact it is one of the larger immigrant groups in Norway, which has a total population of 4.5 million. A person of Turkish origin is defined as a person having both parents born in Turkey. The numbers thus exclude persons born in Norway with one foreign born parent (Statistisk Sentralbyrå — Særtrykk fra ukens statistikk, 42, 1994). . The pear film is a six minute long colour and sound film produced under the direction of Wallace Chafe and is distributed by Palmer Film and Video services, Belmont, CA. The cartoon series used to elicit The Beach Story and The Shop Story were borrowed from J. Lalleman (1986) Dutch Language Proficiency of Turkish Children in the Netherlands. Foris Publications, and was used in this study with the permission of the author. The Bird Story pictures were drawn specifically for the present study by Eva Burianová, at the time a student at The Charles University, Prague. The cartoons and a summary of the film as presented by Chafe (1980) are given in the Appendix. . In the major part of the analysis presented here only reference to human story characters is included; reference to animal participants is left out in the analysis of frequency distribution. . Further issues concerning the which item problem are accounted for in Nistov (2000), including the coding of dislocations, pro-nominal copies, genitives, possessive pronouns and singular vs. plural nominal forms, all of which are treated in diverse ways in the literature. . It is with regard to the how far to look back problem that the lack of a unified format for coding becomes particularly evident in the literature. In some studies maintenance and shift of reference are coded in relation to other types of unit than those defined here as a clause. The degree of embedding, which is a dimension discussed by Marslen-Wilson, Levy and Tyler (1982: 346) as an influencing factor, is not considered in the present study. The most intricate question with respect to coding procedures seems to be related to relative clauses, which are treated very differently in the various studies. In some studies relative clauses are left out of the analysis, as is argued for by Bamberg (1987) and McGann and Schwartz (1988), who claim that ‘‘Relative clauses serve a descriptive or identifying function rather than bearing directly on ‘‘thematic progression’’ (Bamberg’s phrase) (McGann and Schwartz 1988: 222–3). The present analysis does not follow this procedure strictly. Quirk and Greenbaum (1977: 383) point to the fact that relative clauses, in particular the nonrestrictive ones, may have more than a descriptive function, and are sometimes semantically indistinguishable from co-ordination or adverbial subordination. The relative clauses do not only contain descriptions of states but also actions which may contribute to the


thematic flow in the narrative. In the present analysis the following coding pattern is followed: When the relative clause contains reference to another character or entity than the one referred to by the relative pronoun itself, it will be regarded as the preceding utterance to look back to. Reference expressed by the relative pronoun itself is not counted as an instance of reference continuation. . # is used to mark clause boundaries, MAI=maintenance of reference, SHI=shift of reference. . In the analysis of the reference data used for this presentation, narratives in which the story characters were given proper names have been excluded. . In the renderings of the stories references to the story characters are coded as follows. The Pear Story: THE BIKE BOY in bold and THE PEAR PICKER in bold italics, the minor characters in italics. The Beach Story: THE BOY in bold, THE GIRL in bold italics. The Shop Story: THE LADY in bold, THE BOY in bold italics, and THE CASHIER in italics. The Bird Story: THE GIRL in bold, THE HELPER in bold italics and THE NAUGHTY CHILD in italics. . The stories are rendered in Norwegian in accordance with the L2 learners’ orthography. The English glosses attempt to give a close rendering of meaning, and do not reflect Norwegian word order or non-target-like word order solutions of the L2 learners. They do however, aim to display problems in morphology with respect to marking of tense, number and definiteness. . The asterisk is used to indicate zero-marking which is ungrammatical in the target language. The zero is indicated by Ø in the target language required position of the ‘‘missing’’ constituent. In the English glosses the Ø is placed in accordance with English word order. . The fact that the L2 learners’ use of pronouns may be influenced in an indirect way by their L1 is elaborated on in Nistov (1995) and Nistov (2000). . This is especially valid for oral colloquial Turkish narrative style.





Ingvild Nistov

Appendix • Cartoons and film summary • Major and minor characters as identified in the analysis. The Beach Story THE BOY, THE GIRL

The Bird Story THE GIRL

The Shop Story THE LADY

The Pear Story The following is a summary of The pear film as presented by Chafe (1980). The film begins with a man picking pears on a ladder in a tree. He descends the ladder, kneels, and dumps the pears from the pocket of an apron he is wearing into one of the three baskets below the tree, He removes a bandana from around his neck and wipes off one of the pears. Then he returns to the ladder and climbs back into the tree. Toward the end of this sequence we hear the sound of a goat, and when the picker is back in the tree a man approaches with a goat on a leash. As they pass by the baskets of pears, the goat strains toward them, but is pulled past by the man and the two of them disappear in the distance. We see another closeup of the picker at his work, and then we see a boy approaching on a bicycle. He coasts in toward the baskets, stops, gets off his bike, looks up at the picker, puts down his bike, walks toward the baskets, again looking at the picker, picks up a pear, puts his back down, looks once more at the picker, and lifts up a basket full of pears. He puts the basket down near his bike, lifts up the bike and straddles it, picks up the basket and places it on the rack in front of his handlegears, and rides off. We again see the man continuing to pick pears. The boy is now riding down the road, and we see a pear fall from the basket on his bike. Then we see a girl on a bicycle approaching from the other direction. As they pass, the boy turns to look at the girl, his hat flies off, and the front wheel of his bike hits a rock. The bike falls over, the basket falls off, and the pears spill out onto the ground. The boy extricates himself from under the bike, and brushes off his leg. In the meantime we hear what turns out to be the sound of a paddleball, and then we see three boys standing there, looking at the bike boy on the ground. The three pick up the scattered pears and put them back into the basket. The bike boy sets his bike upright, and two of the other boys lift the basket of pears back onto it, The bike boy begins walking his bike in the direction he was going, while the three other boys begin walking off in the other direction. As they walk by the bike boy’s hat on the road, the boy with the paddleball sees it, picks it up, turns around, and we hear a loud whistle as he signals to the bike boy, The bike boy stops, takes three pears out of the basket, and holds them out as the other boy approaches with the hat. They exchange the pears and the hat, and the bike boy keeps going while the boy with the paddleball runs back to his two companions, to each of whom he hands a pear. They continue on, eating their pears. The scene now changes back to the tree, where we see the picker again descending the ladder, He looks at the two baskets, where earlier there were three, points at them, backs up against the ladder, shakes his head, and tips up his hat, The three boys are now seen approaching, eating their pears. The picker watches them pass by, and they walk off into the distance. (Chafe 1980: xiii–xv) Major characters: THE BIKE BOY, THE PEAR PICKER


References Aarssen, J. (1996). Relating events in two languages. Acquisition of cohesive devices by Turkish–Dutch bilingual children at school age. Studies in multilingualism 2. Tilburg: Tilburg University Press. Bamberg, M. (1986). A functional approach to the acquisition of anaphoric relationships. Linguistics, 24–1, 227–84. Bamberg, M. (1987). The acquisition of narratives. New York: Mouton de Gruyter. Bartlett, E. J. (1984). Anaphoric reference in written narratives of good and poor elementary school writers. Journal of Verbal Behaviour, 23, 540–52. Berman, R. A. (1999). Bilingual proficiency/Proficient bilingualism: Insights from narrative texts. In: G. Extra and L. Verhoeven (eds.), Bilingualism and migration. Berlin: Mouton de Gruyter., 187–208. Berman, R. A. and D. I. Slobin (1994). Relating events in narratives: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Biber, D. (1988). Variation across speech and writing. Cambridge: Cambridge University Press. Broeder, P. (1991). Talking about people — A multiple case study on adult language acquisition. European Studies on Multilingualism 1. Amsterdam: Swetz and Zeitlinger. Brown, G. and G. Yule (1983). Teaching the spoken language. Cambridge: Cambridge University Press. Chafe, W. (ed.) (1980). The pear stories. Norwood: Ablex Publishing Corporation. Clancy, M. P. (1992). Referential strategies in the narratives of Japanese children. Discourse Processes, 15, 441–67. Cummins, J. (1983). Language proficiency and academic achievement. In: J. Oller (ed.), Issues in language testing. Rowley: Mass: Newbury House, 108–29. Erguvanlı-Taylan, E. (1986). Pronominal versus zero representation of anaphora in Turkish. In: D. I. Slobin and K. Zimmer (eds.), Studies in Turkish linguistics. Amsterdam: John Benjamins, 209–31. Fox, B. (1987). Anaphora in written popular narratives. In: T. Tomlin (ed.), Coherence and Grounding in discourse. Amsterdam: John Benjamins, 157–74. Fraurud, K. (1996). Cognitive ontology and NP form. In: T. Fretheim and J. K. Gundel (eds.), Reference and referent accessibility. Amsterdam: John Benjamins., 65–87. Givón, T. (ed.) (1983).Topic continuity in discourse. A quantitative cross-linguistic study. Amsterdam: John Benjamins. Givón, T. (1995). Functionalism and grammar. Amsterdam: John Benjamins. Hatch, E. (1984). Theoretical review of discourse and interlanguage. In: A. Davies, C. Criper and A. P. R. Howatt (eds.), Interlanguage. Edinburgh: Edinburgh University Press, 190–203. Hendriks, H. (1998). Reference to person and space in narrative discourse: A comparison of adult second language and child first language acquisition. Studi Italiani di Linguistica Teorica e Applicata, anno XXVII, 1998, numero 1, 67–86. Hickmann, M. (1982).The development of narrative skills. Pragmatic and metapragmatic aspects of discourse cohesion. Doctoral thesis, University of Chicago.





Ingvild Nistov

Hickmann, M. (1987). The pragmatics of reference in child language. In: M. Hickmann (ed.), Social and functional approaches to language and thought. Orlando: Academic Press, 165–84. Johanson, L. (1992). Strukturelle Faktoren in türkischen Sprachkontakten. Sitztungsberichte der Wissenschaftlichen Gesellschaft and der Johann-Wolfgang-Goethe-Universität Frankfurt am Main: Bd 29, Nr 5. Stuttgart: Steiner. 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. London: Academic Press, 121–97. Karmiloff-Smith, A. (1985). Language and cognitive processes from a developmental perspective. Language and Cognitive Processes, 1–1, 61–85. Kerslake, C. (1987). Noun phrase deletion and pronominalization in Turkish. In: H. E. Boeschoten and L. Verhoeven (eds.), Proceedings of the third conference on Turkish linguistics. Tilburg: Tilburg University Press., 91–104. Klein, W. and C. Perdue (1992). Utterance structure, developing grammars again. Studies in Bilingualism 5, Amsterdam: John Benjamins . Klein, W. and C. von Stutterheim (1991). Text structure and referential movement. Sprache und Pragmatik, Arbeitsberichte 22, Lund. Küntay, A. (1992). The development of referential continuity in elicited Turkish narratives. Paper presented at the 6th International Conference on Turkish Linguistics, 12–14 August 1992, Eskis¸ehir, Turkey. Lindberg, I. (1995). Referential Choice in Swedish Narrative Discourse. Scandinavian Working papers on Bilingualism, 10. Also presented in Lindberg, I. (1995). Second language discourse in and out of the classrooms: studies of learner discourse in the acquisition of Swedish as a second language in educational contexts. Doctoral dissertation, Centre for research on bilingualism, Stockholm University. Mac Donald, K. (1997). Spørsmål om grammatikk. Når norsk er andrespråk. Oslo: Cappelen Marslen-Wilson, W., E. Levy and L. Tyler (1982). Producing interpretable discourse: the establishment and maintenance of reference. In: R. Jarvella and W. Klein (eds.), Speech, place and action. Chichester: Wiley, 339–78. McGann, W. and A. Schwartz (1988). Main character in children’s narratives. Linguistics, 26–2, 215–33. Nistov, I. (1995). ‘‘Nomen, pronomen eller Ø?’’ Referensielle uttrykk i narrativar skrivne av tyrkiske innvandrarelevar som lærer norsk som andrespråk. In: M. Kalin and S. Latomaa (eds.), Nordens språk som andraspråk 3, Tredje forskarsymposiet i Jyväskylä 24–5. March 1995, 169–80. Jyväskylä: Language Centre for Finnish Universities, University of Jyväskylä. Nistov, I. (1999). Zero anaphora in Turkish learners’ Norwegian L2 narratives. In: B. Brendemoen, E. Lanza and E. Ryen (eds.), Language encounters across time and space. Oslo: Novus Press, 149–78. Nistov, I. (2000). Referential Choice in L2 Narratives. A study of Turkish adolescent immigrants learning Norwegian. Doctoral dissertation, Department of Linguistics, University of Oslo.


Ochs, E. (1979). Planned and unplanned discourse. In: T. Givón (ed.) Syntax and semantics Vol 12: Discourse and syntax. London: Academic Press, 51–80. Pfaff, C. (1987). Functional approaches to interlanguage. In: C. W. Pfaff (ed.), First and Second Language Acquisition Processes. Cambridge, MA: Newbury House, 81–102. Quirk, R. and S. Greenbaum (1977). A university grammar of English. Longman. Schaufeli, A. (1991).Turkish in an immigrant setting. A comparative study of the first language of monolingual and bilingual Turkish children. Doctoral dissertation. University of Amsterdam. Strömqvist, S. and D. Day (1993). On the development of narrative structure in child L1 and adult L2 acquisition. Applied Psycholinguistics 14, 135–58. Strömqvist, S. (1996). Discourse flow and linguistic information structuring: explorations in speech and writing. Gothenburg Papers in Theoretical Linguistics, 78. Verhoeven, L. T. (1988). Acquisition of discourse cohesion in Turkish. In: S. Koç (ed.), Studies on Turkish Linguistics. Proceedings of the fourth international conference on Turkish linguistics. Middle East Technical University, Ankara, 437–52. Verhoeven, L. T. (1993). Acquisition of narrative skills in a bilingual context. In: B. Ketteman and W. Wieden (eds.), Current issues in European second language acquisition research. Tübingen: Günter Narr Verlag., 307–23. Wigglesworth, G. (1992). Investigating children’s cognitive and linguistic development through narrative. Doctoral dissertation. Department of Linguistics, La Trobe University, Australia.



Chapter 4

Age-related and L2-related features in bilingual narrative development in Sweden Åke Viberg

Studies of bilingual development tend to be concerned with the ideal case of small children acquiring two languages simultaneously and with more or less complete balance between the two languages (balanced bilingual acquisition). It is, however, common that one of the languages is weaker (unbalanced bilingual development) or introduced after one language has already been established, which is usually referred to as second language (L2) development, but might as well be called late bilingualism, especially if the effect of L2 on L1 is also taken into consideration. In all these cases as well as in (monolingual) first language acquisition, age is a critical factor. Age for natural reasons has always been a central factor in studies of first language acquisition. Most studies have concentrated on the early years, which is natural in view of the fact that children acquire the core of their first language very early in life. In spite of that, language development in many important respects continues throughout the school years and is not completely terminated even in adulthood, see, for example, Gleason (ed.) (1993) and Romaine (1984). The study to be presented below will focus on certain aspects of language structure and language use that continue to develop after five years of age, when the basic structure of the L1 has already been acquired. In particular, discourse structure is subject to continuous reorganization up to quite late in life (Piéraut-le Bonniec and Dolitsky (eds.) 1991; Hickman 1995) and the lexicon tends to expand dramatically during the school years (Anglin 1993; Miller and Gildea 1987). In recent years, the development of narrative structure has attracted great interest with a number of major studies comparing speakers of various ages (Berman and Slobin 1994; McCabe and Peterson 1991). In second language acquisition studies, which are concerned with the later acquired and weaker language of late bilinguals, age has attracted interest primarily in relation to the critical-age hypothesis (see the by now classic



Åke Viberg

collection by Krashen, Scarcella and Long 1982, and for a more recent overview, see Long 1990;1993). According to Long, there may exist separate critical (or sensitive) ages for different aspects of language proficiency: Starting after the first five or six years of life, it is impossible to achieve native-like competence in phonology; starting later than the early teens, the same is true of morphology, syntax and semantics. Preliminary results suggest that the same will eventually be found to hold for collocation, discourse and pragmatics once more of the relevant research is undertaken.’’ (Long 1988: 32)

Obviously, aspects of language such as discourse organization that are not completely established in L1 until rather late cannot be constrained by an early sensitive period. (According to some researchers, such as Flege 1991, even certain phonetic parameters such as VOT are not fixed until the upper teens.) Linguistic features that are not acquired until late in L1 acquisition are, however, theoretically interesting from another point of view. There is a complex interaction between age-related and L2-related features in the development of the second language in unbalanced or late bilinguals. First, age is important for the status of the L1 of second language learners. Young children who acquire a second language do not have a completely developed L1 like adult learners have. The preconditions for transfer will thus vary with age. Age is also important for what should count as the target for L2 acquisition. It is not reasonable to expect that young L2 learners will acquire features of L2 that are not acquired until a later age by L1 learners of the same language. In addition to that, age will influence the input, which is characterized by age-related patterns of language use. Closely related to the input is the primary social domain in which the L2 is acquired. It typically varies for learners who arrive at different ages in the country where the L2 is spoken, especially for learners who have restricted contact with native speakers of the L2 and are exposed to the L2 primarily in institutional settings which are age specific, such as nursery school, school or the workplace. At the same time, there are several characteristics of L2 language development that are not primarily related to the age of the learner. Some of these depend on the stage of acquisition in general and may have parallels in early L1 acquisition. Certain types of simplification belong here. For the purposes of this chapter, features of L2 development which are not agerelated will be counted as L2-related features in a broad sense, leaving open the question as to whether they have parallels in early L1 acquisition or not. As a first step, these features will be identified by comparing L2 learners to native controls of the same age. All features that distinguish the L2 learners

Age- and L2-related features in bilingual narratives in Sweden

from the controls will be regarded as L2-related features in a broad and unqualified sense. On closer examination, some of these features may turn out to have parallels in early L1 acquisition and represent early developmental features in a general sense. Others will turn out to be L2-related features in a narrow sense, i.e. features that lack any parallel in L1 development. There are certain fundamental differences between L1 and L2 acquisition. The error rate (in relation to adult, native norms) tends to be much lower in first than in second language acquisition. Both in terms of product and process there are qualitative differences. (See, for example, Andersson 1992 on Swedish gender, Håkansson 1992 on word order and Viberg 1993a on lexical semantics.) According to Schlyter (1993), the weaker language in unbalanced simultaneous bilingual acquisition resembles to a great extent an L2. This implies that age of acquisition is less important than variation along the continuum rich ↔ restricted input for such features. The explanation for the existence of L2-related features in a narrow sense is far from clear. Transfer is one obvious factor, but the fact that the context of acquisition is almost universally more restricted in L2 compared to L1 acquisition has often been overlooked. In addition to the theoretical interest, the relationship between age-related and L2-related features is a problem of great practical concern in designing L2 teaching for the increasing number of preschool and school-age minority or immigrant children in Western Europe (Eldering and Kloprogge 1989) and North America (Genesee 1994). Such teaching must build on age-appropriate models of the L2 for acquisition and take into consideration variations in the acquisition process depending on age. It will not be possible in this chapter to deal with more than some aspects of the questions raised in the introduction. In the section below, a brief description will be given of a data-base containing oral narratives elicited from bilingual and monolingual children of varying ages from five to sixteen years. The subsequent sections will focus on various aspects of narrative development with short illustrative examples from the data-base. The first section is devoted to narrative structure, i.e. the overall organization of the narrative and the relationship between its major components such as orientation, plot and resolution. The following sections will be concerned with various expressive devices mapping forms to functions (and vice versa) within linguistic subsystems such as clause combining, certain aspects of the lexicon, and, finally, tense and temporal relations.





Åke Viberg

The data-base The chapter is built on data from three projects concerned with bilingual children at various ages ranging from preschool (5-year-olds) to early and later school age (with 16 years as the upper limit). In all three cases, Swedish clearly was the weaker language of the bilinguals and will be referred to as the L2. The design varies somewhat from project to project as can be observed in Table 1 but is similar in certain crucial respects. The primary data were recordings of oral production carried out individually with each child at several points in time. In addition to conversation guided by a researcher, one central component consisted of the task of narrating a number of silent video clips or picture stories, some of which recurred in a cyclical fashion both within and across the various projects. All recordings have been transcribed and stored on computer, which has made it possible to annotate the transcriptions with word class labels. In the preschool project, 15 five-year-old children with Swedish as L2 were recorded in Swedish during five recording sessions relatively evenly distributed over one year. 10 monolingual Swedish children serving as a control group were recorded once in a recording session structured in a way that was similar to the first session with the L2 children. This project has not yet been reported on in full.

Table 1. Three corpora including oral narratives The preschool project (5-year-olds) 15 five-year old Swedish L2 children recorded five times during one year 10 five-year old Swedish L1 children recorded in parallel at time 1 The early school-age project (end of preschool + grades 1–4) 30 Swedish L2 children recorded once a year during four years beginning at age 6 10 Swedish L1 children recorded once a year during four years beginning at age 6 The later school-age project (grades 4–9) (Bilinguals at School: BAS-project) 20 bilingual children recorded in Swedish and Finnish in grade 4 and grade 6 10 Swedish L1 children recorded in Swedish in grade 4 and grade 6 10 Finnish L1 children recorded in Finnish in grade 4 and grade 6 20 bilingual children recorded in Swedish and Finnish in grade 7 and grade 9 10 Swedish L1 children recorded in Swedish in grade 7 and grade 9 10 Finnish L1 children recorded in Finnish in grade 7 and grade 9


In the early school-age project, recordings were carried out during the early school years, first at the end of preschool, when the children were 6 years old, and then at the end of grades 1, 2 and 4. The general design of this project together with the findings in a study concerned with lexical development is presented in Viberg (1999). In total, 30 Swedish L2 children1 and 11 Swedish L1 children were recorded at all four recording sessions. The Bilinguals at School (BAS)-project (Viberg 1994), which for the purposes of this chapter will also be referred to as the later school-age project, was concerned with the bilingual development of children in home language classes, where teaching is relatively evenly divided between the L1 and Swedish L2. In the main study, two groups of Finnish children were followed. Each group consisting of 20 bilingual Finnish/Swedish children were recorded in both languages at two points of time. The first group was recorded in grades 4 and 6 and the second in grades 7 and 9. At each point in time, parallel recordings were made in Finnish and Swedish by researchers with the respective language as L1 on diffferent occasions separated by approximately one week. For each bilingual group, there were two monolingual control groups with children of the same age from Stockholm and Helsinki, respectively.

Narrative structure In this section, various aspects of the overall organization of a narrative will be briefly discussed using the narration of one particular video-clip as an example. It will be referred to as The sand-pit story and is represented in all the three projects. In its simplest form, which is predominant at preschool and early school-age even with L1 children, the story is in the main narrated concretely event by event without any clear marking of the overall significance of the sequence of events as in Example 1 from an L1 child in grade 4. According to Labov (1972), the following major components can appear in a narrative: 1. 2. 3. 4. 5. 6.

Abstract (optional) Orientation Complicating action Evaluation Resolution Coda





Åke Viberg

Example 1. Narration of the video-clip ‘‘The Sand-pit’’. Swedish L1 child K04; Grade 4. (Int. kan du berätta den också can you tell that one, too?) (1) mm de va ehh / en mamma som va ute me sin / lille son eller dotter ja vet inte va de va mm there was eh- a mummy who was out with her little son or daughter I don’t know what it was (2) / ja sen kom en / man å / satte sej där / bredvid å så fråga om han fick sitta ner (NÄÄ) well, then a man came and sat down there beside (her) and then asked if he could sit down (NO) (3) å de fick / han / hon flytta sej då / and he could / she moved then (4) men då blev ungen / arg / på han då / but then the kid got angry with him (5) å så / ville mannen hälsa på ungen and then the man wanted to greet the kid (6) men då blev / den arg å sparka / han på smalbenet / but then it was angry and kicked him on his shinbone (7) å så / tänkte han bjuda på en karamell / ja and then he was going to treat him a candy / yes (8) men då bet han va heter de / han / på fingret / but then he bit – what do you call it – his finger (9) å så / gick en som säljde ballonger förbi and then someone who was selling balloons went by (10) å så / köpte han alla då / and then he bought all of them (11) å så / gav han dom till henne / eller han då / .hhh and then he gave them to her -or him (12) å så / flög hon and then she flew (13) å s-/ så / pussade han / henne and th- then he kissed her In Example 1, (1) represents the Orientation introducing one of the main characters, the child, and its mother. In addition to that, the orientation in many versions includes mention of the ‘‘scene’’ (a park and the sand-pit (AmE sand-box), where the child is playing). The Complicating action, which marks


the beginning of the real plot, is the entry of the man on the scene in (2) and, as is stressed in more mature versions, his starting to flirt with the child’s mother. Subsequently, the child gets angry (or jealous), which is countered by the man´s attempts to pacify the child. Each attempt by the man is answered by an aggressive act by the child. The Resolution to the problem (seen from the man’s perspective) is the arrival of the balloon vendor and the child’s disappearence, which makes it possible for the man to reach his goal and kiss the child’s mother. Needless to say, the point of the story is rather absurd. It is taken from a series of short video-clips showing the world in reverse. Example 1 presents a minimal solution and narrates as already mentioned the events as shown on the video without much deeper interpretation of goals or motives of the characters. Other versions contain optional elements of the narrative structure. One possibility is to give an Abstract, a short summary, at the beginning of the story as in the following example from a student in grade 9: Well, you might say something like: there is a guy who wants the chick but the kid gets in between (K80. Grade 9. Swedish: njaa man kan väl säja ungefär som / att de e en kille som / vill ha / tjejen (MM) men ungen kommer emellan). After the abstract, this student narrates some of the individual events. The abstract also introduces a more abstract interpretation of what is shown on the video. Some of the older students consistently use abstractions to narrate the whole story in a condensed form as in Example 2. Hardly any of the concrete events is explicitly mentioned in this version. Example 2. Narration of the video-clip ‘‘The Sand-pit’’. L1 child K70: Grade 9. (1) mm // ja vet inte om de va pappan de kanske de inte va mm I don’t know if it was his father, maybe it wasn’t (Int. de va de nog inte (NÄE) men de / jaa probably, it wasn’t (NO) but it / yes) (2) de / kom en liten gubbe / å ville / ville ha mamman a little old man came and wanted the mummy (3) men de ville inte ungen / (NÄE) but the child didn’t want that (NO) (4) så han gjorde alla möjliga knep (.HM) so he played a lot of tricks on him (5) för att / han inte skulle få va me mamman / (.HM) in order that he shouldn’t be able to be with his mummy





Åke Viberg

(6) men sen så / lura gubben upp honom me ballonger i luften / but then the old man fooled him with balloons (into going) up in the air (7) så fick han sitta brevi mamman then he could sit beside the mummy Abstractions appear as an option primarily in versions related by older children and adults. In a few of the adult versions such as Example 3, a vivid account of the individual events is fit into an abstract frame that makes the overall purpose and direction of the actions clear The narrative structure in this version forms a complex hierarchy of events. In particular, the man’s motives are described explicitly in (2), (4) and (9), which adds a deeper significance to the concrete events. Example 3. Narration of the video-clip ‘‘The Sand-pit’’. L1 Adult. (1) ja de e en kvinna som sitter på en parkbänk / och hennes barn / leker i sandlådan me hink å spade well, there is a woman who is sitting on a park bench and her child is playing in the sand-pit with a bucket and a spade (AmE shovel) (2) plötslit kommer en man förbi / som bestämmer sej för att // göra sej intressant för kvinnan suddenly, a man comes by who decides to make himself interesting to the woman (3) så han går fram å börjar prata me henne // hon flyttar / sej ifrån honom / å han sätter sej brevi henne // so he walks up (to her) and begins to talk to her. She moves away from him and he sits down beside her. (4) för att ääh ytterligare // göra sej ääh bekant me kvinnan så försöker han // bara bli vän me barnet som leker i sandlådan in order to get more acquainted with the woman, he just tries to make friends with the child who is playing in the sand-pit. (5) och ääh // han gör de genom å gå fram å försöka krama honom får då en / spark i magen and erh- he does that by going up to him and trying to hug him then gets a kick in the stomach


(6) ääh han gör de genom att / bjuda på / en karamell å då biter barnet honom i fingret erh he does that by offering a candy and then the child bites him on the finger (7) å ti slut så / blir han så lessen att han börjar gråta / den / äldre mannen and, finally, he gets so sad that he starts crying, this older man (8) plötslit kommer de i alla fall förbi en / ballongförsäljare suddenly, a balloon vendor comes by (9) å då ser han sin chans här and then he sees his chance here (10) å han går bort till ballongförsäljaren köper ett .hhh en stor knippe ballonger å ger ti barnet and he goes up to the balloon vendor buys a erh big bunch of balloons and gives to the child (11) varvid barnet lyfter / och försvinner upp i / atmosfären when the child lifts and disappears up into the atmosphere (12) å berättelsen slutar me att mannen ger kvinnan en / kyss på kinden and the story winds up with the man giving the woman a kiss on her cheek The sand-pit story was studied more closely within the project concerned with Finnish–Swedish bilingual school-children (Juvonen, Lindberg and Viberg 1989; Lindberg, Juvonen and Viberg 1991). Finnish and Swedish versions from 20 bilingual children in grade 4 were analysed and compared with versions from 20 monolingual Finnish controls and 10 Swedish monolingual controls of the same age. In addition to the material from the children, adult versions were recorded with adult L1 controls in both Swedish and Finnish. In order to facilitate a systematic comparison between groups, the contents of the story were represented in the form of 35 simple propositions, which were classified according to their function in the narrative structure. For each group, a number of core propositions were identified, propositions which in some form were represented by at least 50 per cent of the speakers in the group. It turned out that the two adult groups used more abstract core propositions than any of the groups of children. Certain propositions belonging to the Orientation and the Resolution were always



Åke Viberg

realized by the adults, which indicates a strong sense of awareness of the need on the part of the listener to get a clear picture of the background and the point of the story. Even if these propositions were often realized even in the children’s versions, this did not apply with the same consistency, suggesting that the children still did not have a firm grasp of all aspects of narrative structure. The bilingual children had a tendency to give more detailed and concrete versions in both languages than the monolingual children, who tended to provide more condensed versions. The bilingual children also tended to give similarly structured versions in both languages with the exception of seven children, who gave reduced versions in Swedish, obviously due to the fact that their Swedish was in general weak. The number of propositions in the Swedish and the Finnish versions of the bilingual children are compared in Figure 1 The students, which are identified by numbers on the horizontal axis, have been ordered so that the seven students with reduced Swedish versions are shown to the right. As you can see, the differences between the two versions are small for the rest of the students with respect to the number of propositions. The overlap is also great with respect to the individual propositions that were used as is demonstrated in Figure 2, which shows the number of propositions appearing in both the Swedish and the Finnish versions in comparison to the number of propositions which appeared only in one of the versions. 25 Number of propositions



Finnish

20 15 10

Swedish

5 0 6

1

2 10 16 15 22 8 25 17 11 5 18 9 21 13 24 23 12 14 Students

Figure 1. The number of propositions in the Swedish and Finnish versions of the re-narrations of the bilingual children.


30 Finnish Both Swedish

Number of propositions

25 20 15 10 5 0 6

1

2 10 16 15 22 8 25 17 11 5 18 9 21 13 24 23 12 14 Students

Figure 2. Overlap between Swedish and Finnish versions in the renarrations of the bilingual children

The similarity between the two versions of the bilingual children indicates that narrative structure belongs to a general conceptual level which is relatively independent of individual languages. In this particular case, the interpretation of the video-clip which is completely non-verbal must obviously proceed at a nonlinguistic level before it is verbalized. However, verbalization requires that a certain linguistic threshhold has been passed for the basic components of narrative structure to be realized in a specific language. Students 23, 12 and 14 have extremely short Swedish versions and do not seem to have passed this threshold in Swedish.

Connectors One important aspect of the linguistic realization of a narrative is clause combining (Haiman and Thompson 1988) and the mastery of the formal devices serving to link or combine clauses and their various functions. In the learner corpora, striking differences can be observed with respect to the use of certain connectors (Viberg 1993b). Connector is a cover term for a number of clause-initial markers such as coordinating conjunctions, subordinators and





Åke Viberg

complementizers. In Swedish, there is also a small set of short adverbs, which very frequently (at least in speech) introduce a main clause — I am referring to sequential markers corresponding to (and) then in English. In Swedish, there are three frequently used equivalents of English then in this function as a sequential marker: då (‘just then’; simultaneous), sen (‘then’; sequential) and så (sequential ‘so’). When these adverbs appear initially in a clause, they are analyzed as a separate (functional) category: connective adverb. In addition, two of the connective adverbs can also be used as topic placeholders. This function refers to the frequent use in spoken Swedish of så (or, occasionally, då) after certain preposed adverbial phrases: Igår kom Peter. Igår, så kom Peter

Yesterday, Peter came Yesterday, Peter came

The connective adverbs reach a particularly high frequency in the speech of children (both native and non-native speakers), which is illustrated in Example 4, where all sentential connectors appear in bold italics. Example 4. Narration of the video-clip ‘‘The car accident’’, L1 child KB12: Grade 2 (Int. så ska du berätta den också well, can you tell that one too?) (1) de handlade om en gubbe it was about an old man (2) som körde bil who drove a car (3) å sen så blev de nån fel på bilen and then (so) there was some problem with the car (4) å så börja den svänga and then it began swerving å kunde inte riktit styra and could not really steer (5) å sen så körde dom in i ett träd and then (so) they ran into a tree å ramla ur bilen and fell out of the car


(6) å då kom en ambulans me en bår and then an ambulance came with a stretcher å la bilen på båren and put the car on the stretcher å bar in den i ambulansen and carried it into the ambulance (7) sen så satte dom plåster på trädet then (so) they put band-aids on the tree (8) å sen så åkte dom iväg and then (so) they went away (9) å han låg kvar and he was left The strong reliance on the connective adverbs sen/så/då ‘then’ is an age-related feature. In the study of the sand-pit story referred to above, it was found that around 50 per cent of the connectors in the Swedish versions told by children consisted of connective adverbs compared to around 25 per cent in the adult versions. The proportion was very similar for both Swedish L1 and L2 children, when the connective adverbs as a global type were compared to other types of connectors. There was, however, an interesting difference with respect to the choice of connective adverb as will be demonstrated shortly. In the Finnish versions, the proportion of the connective adverbs sitten and niin was also markedly higher in the narrations by the children. The differentiation between the three connective adverbs in Swedish is a language-specific characteristic. English has one general correspondence (then), while some other languages such as Finnish and Spanish have two. Swedish då clearly marks simultaneity, while sen and så both mark sequence with a rather subtle contrast that is difficult to pin down. It appears that så indicates that the events are more closely related. In addition, så is highly polysemous and appears with several other functions. The meaning of sen on the other hand is rather transparent and invariable. This is probably the major reason why L2 learners have a very strong tendency to favour sen at the expense of så, which is the most frequent marker in the speech of native children. The frequent use of sen is actually one of the most characteristic L2-related features of the narrations. It disappears only gradually and very slowly over time. This is demonstrated very clearly in the material from the early school-age project, where children were followed longitudinally for four years. Table 2 shows the use of connective adverbs and topic place-holders in the complete set of recordings.



 Åke Viberg

From the table it is possible to find out how many of the children in a certain group use a certain element The total number of children in the group is symbolized by capital N, which is given for each recording time since there was some dropping off over time. The number of children who use a certain element is symbolized as n. The table shows that sen, så and då are consistently used by all the L1 children at all four times, which is in stark contrast to the L2 children, who in the first recordings tended to master only sen. In preschool 20 of the 23 L2 children used sen, while så was used only by 11 children and då by 13. The frequency of occurence of sen at the group level (symbolized f) is also remarkably high in the L2 group. The percentage of sen in relation to the total number of connective adverbs (% in the table) was as high as 78 per cent for the L2 children in this recording, which should be compared to the 37 per cent of the L1 children. The difference also turned out to be significant when tested with a Mann-Whitney test (p170,000 263 3,543


With only a few exceptions (such as kan ‘can’ and måste ‘must’), the present tense ends in r. There is one completely productive class of Swedish verbs, the ar-verbs, which end in ar in the present tense. These verbs have a stem that ends in a, to which the following endings are added: -r in the Present tense, Ø in the Imperative and Infinitive, -de in the Past and -t in the Supine. A specific characteristic of spoken Swedish is that the endings in the Past and the Supine of the ar-verbs may be omitted and this is also observed in the data from the Swedish L1 children. The short verbs consist of a small number of usually monosyllabic verb stems ending in a stressed, long vowel. They have endings which are basically the same as those of the ar-verbs, except that the vowel is shortened and the consonant lengthened (marked by dubbling of the consonant in writing), when the Past and Supine endings are added. In addition, some of the most frequent short verbs have an irregular past form (e.g. såg ‘saw’ and gick ‘went’). The er-verbs end in -er in the present tense. The weak er-verbs are regular and form the Supine with a t like the ar-verbs and the Past by adding -de or -te (phonologically conditioned by the voicing of the preceding segment). The strong er-verbs mark the Past by a vowel shift in the stem according to a number of semi-regular patterns without adding any suffix. The Supine is marked by the suffix -it, usually in combination with a vowel shift in the stem. At the top of the table, regularity and productivity are indicated as continua. Only when the forms of a verb are completely unrelated is there pure irregularity as in är ’is’ and var ‘was’. Strong verbs, for example, have related but not completely predictable forms. Productivity, i.e. an open pattern with which new forms can be generated (in borrowings or nonsense words), presupposes regularity, but only the regular pattern of ar-verbs can be used to accomodate new words (e.g. faxar-faxa- faxa-faxade-faxat ‘to fax’). Regularity and productivity in the input to the learner often give rise to overgeneralizations in the output. The frequency distributions of the various form classes also constitute an important factor. Ideally, we should have reliable data on the input frequency for the individual learners. The best that can be offered, however, is a rough estimation based on the frequency distributions in written Swedish based on the frequency dictionary published by Allén (1971), which can be compared to the distribution in the oral data from the Swedish L1 children. In Table 7, the data from the first recording in the Preschool project has been used as a source. It is important to keep apart lexical (or type) frequency and textual (or token) frequency. In Swedish, as in most European languages, a small number





Åke Viberg

of very frequent verbs are heavily represented in texts (Viberg 1993a). These verbs, however, generally belong to closed and unproductive conjugation classes. A look at the 20 most frequent verbs is illustrative. According to the Swedish frequency dictionary (Allén 1971), which is based on approximately 1 million running words collected from Swedish newspapers, 18 of the 20 most frequent verbs are irregular and together cover approximately 45 per cent of the total number of tokens. Among the 20 most frequent verbs in the first recording with the L1 children in the Preschool project, only two (tro, åka) were regular. The other 18 of the 20 most frequent verbs covered close to 67 per cent of the running verbs and since the majority of these verbs tend to reach a high percentage of occurrence in spoken conversation in general, this means that the mastery of the correctly inflected forms of these verbs by a learner would boost over-all accuracy in this area considerably. On the other hand, the textual frequency of the productive and regular ar-verbs reached only a little more than 10 per cent, whereas their lexical frequency reached almost 50 per cent, which is very high with regard to the fact that the restricted size of the corpus must have affected the lexical frequency in a negative way. In the frequency dictionary, the lexical frequency of the regular and productive ar-verbs is as high as 67 per cent, whereas their textual frequency is only 25 per cent. In spite of the indirect nature of the evidence about learner input, it can safely be concluded that if the learner attends to the textual frequency of individual verbs, the lexicalized and irregular forms will predominate, whereas sensitivity to lexical frequency will lead to a predominance of regular forms. Roughly, acquisition of the verb paradigms can be hypothesized to proceed along the following path. To begin with, the learner will identify and extract individual verb forms from the input. At this stage, irregular verbs should be favoured and there will be no basis for forming regular patterns. Once a greater number of verbs have been extracted and internalized, however, the learner will reach a state where the number of regular verbs will predominate in the internalized lexicon. At this stage, patterns can be identified and become productive in the learner’s internalized system. Often such patterns are strong enough to lead to the production of overregularized forms, which are the clearest sign in the output of the existence of a productive pattern. The next section will be devoted to a case study, which lends support to this learning scenario.

Case study 2: Yavuz The Turkish child with the fictive name Yavuz was one of the informants in the preschool project. In this project five-year-old L2 learners of Swedish were


recorded five times during a year which made it possible to follow the development more closely than in the other projects where the period of time between the recordings was much longer. In the narrations from the first recording, practically all of the verbs directly tied to the plot-line were marked with the Supine, the form of the Swedish main verb in the perfect construction. However, no form of ‘have’ appeared except in two cases in the whole transcription from time 1. Example 10 shows the typical usage of verb forms at this time. Example 10. Verb forms in Yavuz’ narration of the Sand-pit story at Time 1. (1) han kommit honoms he come (SUPINE) hims (2) han pusshe kiss(3) å han säj nånting and he say something (4) ja vill älska dej säjer I want love you says (5) å sen han kastat nånting honom and then he thrown something him (6) å sen honom eh han kommit and then him eh he come (SUP.) (7) å sen pau auh sägit and then (SOUNDS) said (SUP.) (8) han kastat honom sand he thrown him sand (9) å sen / ballong kommit and then / balloon come (SUP.) (10) å sen han köpit alla ballonger and then he bought (SUP.) all balloons (11) å sen köpit dom (HM) and then bought (SUP.) them (12) å sen han flyget and then he flown



 Åke Viberg

(Int.: ja vem va re som flög i väg yes, who flew away?) (13) en liten pojken a little boy

The Supine in Swedish has several different allomorphs. At time 1, Yavuz has a strong tendency to overgeneralize -it, the ending of the strong verbs, which is the most salient Supine marker. It is used correctly with kommit ‘come (Supine)’ but also overgeneralized to sägit (correct: sagt) ‘said (Supine)’, köpit (correct: köpt) ‘bought (Supine)’ and flyget ‘flown’ (with the vowel i reduced to e; correct: flugit). Strong verbs in Swedish also tend to change the stem vowel in the Supine. This is a feature Yavuz tends to avoid during all the recording sessions. The only example of a verb belonging to the most productive conjugation (the ar-verbs) in Example 10 is kastat ‘thrown’, which has the correct Supine allomorph -t (basic form: kasta ‘(to) throw’). The tendency to use the Supine as a generalized past tense is very striking in the first recordings. As can be observed in Table 8, there is a continuous development across time. The Past forms increase in frequency, while the Supine forms decrease. Table 8. Distribution of forms in Yavuz’ speech across the five recording sessions Time Base-form a Present Past

Supine Imperative Total number of verb tokens in recording

1 2 3 4 5

25.9% 23.2% 8.8% 19.5% 10.5%

a

16.3% 25.2% 27.6% 21.7% 23.5%

51% 44% 56.2% 47.6% 50.1%

5.4% 7.4% 7.1% 10.3% 15.7%

1.2% 0% 0.4% 1.0% 0%

239 202 283 410 387

Infinitive + all cases where an identical form can be used in spoken Swedish

Since the regular ar-verbs can appear in the same form as the Infinitive when they have a Past or Supine function, such forms are included with the Infinitives in the column marked Base-form. At Time 1, the Supine is five times as frequent as the Past. Subsequently there is a constant decrease in the Supine except at Time 4 and a constant increase in the Past to the extent that it clearly dominates over the Supine at Time 5. (The high proportion of the Supine at Time 4 depends to a great extent on the forms kommit and tagit which account for more than half the total number of Supines in this recording.)


The developmental trend is clearly illustrated by a comparison of Example 10 above with the renarration of the Sand-pit story at Time 5 as shown in Example 11. In this version, the majority of the verbs appear in the Past form except kommit ‘come (Supine) and ätit ‘eaten’. Example 11. ‘‘Yavuz’’. Verb forms in the narration of the Sand-pit story at Time 5. (1) han pussade he kissed (2) han kommit där först he come (SUP.) there first (3) och han slog sin ben (MHM) and he hit (Past) his leg (4) han kastade upp s- mm på spaden he threw up on the spade Int.: mh just de de gjorde han / va gjorde han mer rå / han skulle ge han godis va correct, he did, what else did he do, he would give him candy, right (5) ah och sen han ätit sin tummen så aaaah aaoooh and then he eaten his thumb so (SOUNDS OF PAIN) (Int.: han bet honom ja (AH) sen så kom den hära he bit him, yes, then came this) (6) ballongen? balloon? (Int.: just de va gjorde han då exactly, what did he do then) (7) han köpte he bought (Past) (8) och han lämnade barnen and he left (Past) the child (9) han flygde hu- upp he flyed (Over-reg. Past) up



 Åke Viberg

(10) och han pussade en flicka and he kissed a girl

An inspection of the complete set of transcriptions from all the recording sessions shows that individual verbs follow specific lines of development; some verbs appear exclusively in either the Supine or the Past form throughout all the recording sessions, while others change pattern at some stage as demonstrated in Table 9. The verbs komma ‘come’ and ta ‘take’, which are among the most frequently used verbs, appear almost exclusively in the Supine forms kommit ‘come (Supine)’ and tagit ‘taken’. The reason the Past forms are avoided is probably that they are short without any suffix marking their function. (The relevant forms are kom ‘came’ and tog ‘took’.) There is, however, one very frequent verb which with only a few exceptions is used only in the past form, namely göra [jœ:ra] ‘make, do’, which has the Past form gjorde [jT:e]. Even if the stem of this verb is irregular, its Past form is marked by the regular suffix -de, which is more salient than the Supine form ending in -t (gjort [jT(:) ]. The verb säga ‘say’ changes pattern. In the first recording, there are four occurrences of the overgeneralized strong Supine -it (sägit replacing the ordinary Supine sagt) and one occurrence of the reduced Past form sa, which is more frequent in spoken Swedish than the ordinary written Past form sade. At the following two times, there are a few occurrences of sägit (Time 2) and of sa alternating with sade (Time 3). At Times 4 and 5, the form sade with its clear Past ending -de is clearly established as the dominant alternative (with the exception of 1 sa at Time 4). It dominates even relative to the Present form säger, which in the early recordings is also used where the Past could be expected. The choice of past markers seems to follow one primary constraint, namely that past tense should be marked by a suffix. In the first recording, the supine ending -it of the strong verbs predominates in Yavuz’ speech, because of the high textual frequency of such verbs. This suffix is even overgeneralized to some verbs which do not have this supine ending in native Swedish. The regular ar-verbs which have a clear past ending -de tend to replace the Supine with the Past in later recordings and in the last recording the past ending clearly dominates in the ar-verbs (19 Past versus 1 Supine). The fate of the weak er-verbs is particularly illuminating; the vowelless Supine ending -t tends to be replaced by the formally more salient ending -it, which is characteristic of the strong er-verbs in native Swedish. In the last recording, Supine forms such as stängit ‘closed’ (Supine: stängt) and köpit ‘bought’ (Supine: köpt) can

Age- and L2-related features in bilingual narratives in Sweden 

Table 9. Past and Supine forms of verbs during all five recording times in Yavuz’ speech Verb

Tense

Strong and irregular verbs (selection) vara ‘be’ Past Supine göra ‘make’ Past Supine säga ‘say’ Past Supine ta ‘take’ Past Supine komma ‘come’ Past Supine gå ‘go’ Past Supine se ‘see’ Past Supine flyga ‘fly’ Past Supine knyta Past Supine Total strong and irregular verbs

Regular ar-verbs (selection) titta ‘look’ kasta ‘throw’ Total ar-verbs

Regular er-verbs (selection) köpa ‘buy’ Total er-verbs

a

Past Supine Types Past Supine Past Supine Past Supine Types

Recording session 1 2 3 4 5

8 1 4 15

1

4

5 2

8 13 12 1 3 10 9

3 1 4

14 10 2 1 4 1 1

3 2 1

va varit gjorde gjort sade sägit tog tagit kom kommit gick gått såg sett flygde flygit

4

knytit

7 17 10 2 6 31 10 3 1 2 1

1

1

1

2 1

7

Forms Used

3

Correcta va(r)

sa(de) sagt

flög flugit knöt knutit

9 11 16 40 43 40 28 23 58 34 11 11 14 13 17

3

1 4

6

2

2 1 16 16 6 6

Past Supine

4

Past Supine Types

1 6 3

3 3 4

1

1 2 2

3 1 5

tittade tittat kastade kastat

3 5 19 2 20 1 4 7 9

1 1 2

1

1 1

1 2 3

6 6b 6

köpte köpit;Time 5: köpt

Correct refers to the form and not to the choice between Past and Supine. The correct form is given only if it differs from the used form. b Four of these are stängit (incorrect strong supine; Correct: stängt)

 Åke Viberg

still be found. Thus, the choice of forms to a great extent tends to be sensitive to the salience of the forms in the input and to the frequency of occurrence. Yavuz’ development may be interpreted approximately as follows. At an early stage, the high textual frequency of the strong verbs seems to be most important and leads to the overgeneralization of their supine ending -it, but with time the high lexical frequency of the ar-verbs grows in importance and in the last recording there are several cases where their past ending -de has been overgeneralized to strong verbs marking the past with a vowel change in the stem, in examples such as springde ‘ran’ (correct: sprang), flygde ‘flew’ (correct: flög), blidde ‘became’ (correct: blev). The vowel change in strong verbs was avoided even in the first recording, when the Supine marker -it was used to mark Past tense, in examples such as sittit (Supine of sitta ‘sit’; correct: suttit) och drickit (Supine of dricka ‘drink; correct: druckit). The strong tendency to avoid vowel changes in the stem may have been reinforced by the learner’s L1 Turkish, in which only transparent suffixes are used. The changing pattern in marking past time reference seems to be governed by problems in identifying the correct forms. It does not seem possible to find any semantic contrast between Supine and Past in the early recordings. Both when he favours the Supine -it of strong verbs and later when he favours the regular Past form -de, Yavuz follows the principle of marking past time reference with a formally salient suffix.

Conclusions and discussion The major results are summed up in Table 10. Narrative structure, the overall organization of the basic components of a narrative, belongs to a general conceptual level which is available even in the processing of the L2 once a certain linguistic threshold has been passed. The basic components can be observed even in the narrations of the youngest children in this study but there is a continuous development throughout the age-span 5–16 years, which is primarily focused in this chapter, and well into adulthood. The hierarchical relations between events and the motives behind the actions become more and more articulated and the narrator develops a clear evaluative stance with increasing age. Before the linguistic threshold has been passed in L2, there is a general tendency to focus on concrete events. The basic linguistic expressive devices related to the linguistic subsystems dealt with in the present chapter are as a rule acquired already before 5 years


of age by the L1 children. What remains to be acquired with increasing age are primarily more sophisticated uses or discourse functions of elements such as connectors or tense markers. As regards the lexicon, exponents of the core concepts in basic semantic fields are already acquired at this early age but finer distinctions and stylistically marked choices are subsequently added in increasing numbers. The most tangible L2-related features are various neutralizations of basic but L2-specific functional or semantic contrasts. Among the connective adverbs, the functionally most salient marker sen was favoured at the expense of the other options. As pointed out in Viberg (1993a), the neutralization of language-specific lexical contrasts represented by the Swedish placement verbs sätta/ställa/lägga ‘put’ is an L2-related feature in the narrow sense and does not have a parallel in early Swedish L1, where these verbs emerge more or less simultaneously. Within the domain of temporality, expressive devices developed more varied discourse functions with increasing age. The optional, periphrastic markers of the progressive in Swedish were used more frequently to elaborate the background, while the switch to narrative present provided a means to express greater involvement on the part of the speaker in the unfolding of events in the foreground. Even the youngest L1 controls who were five years of age had already acquired the adult inflectional forms of most verbs, although occasional overregularized forms can be found even in the material from older L1 children. In the material from the L2 children, it was possible to observe the extraction of the basic tense forms from the input and how they were mapped to various functions. The complexity of the formal system appeared to lead to a neutralization between the Past and Perfect (or Supine) in the data from two children, who had a strong tendency to use the Swedish Perfect form as a generalized past tense marker, probably due to the greater formal salience of the Perfect markers in Swedish. Lacking similar data from early Swedish L1, it is impossible to decide with certainty whether this represents an L2-specific feature in the strong sense. Probably, overgeneralizations of a similar type are at least not as extensive in early L1. Age-related and L2-related features have been observed also with respect to other aspects of narrative development in studies based on the same materials as this chapter; Lindberg (1995) describes the development of referential strategies in narrations by the Finnish-Swedish bilingual students and Franzén (1996) describes the prosodic structure in some narrations by two L2 learners and two Swedish controls in the early school-age project.



 Åke Viberg

Table 10. Age-related and L2-related features in narrative development Narrative structure (Story grammar) Age-related development: Elaboration of goal-oriented hierarchical structure More abstract compression of events Emergence of Evaluative stance L2-related development: Focus on concrete events until a general linguistic threshold has been passed Connectors Age-related development: Decreasing proportion of connective adverbs L2-related development: Over-use of one connective adverb: sen Lexicon Age-related development: Increased variation and idiomaticity More mental concepts expressed More adjectives (?and adverbs) L2-related development: Neutralization of L2-specific contrasts (example: sätta/ställa/lägga → lägga) Tense marking Age-related development: Introduction of Narrative present L2-related development: Perfect and/or Supine as generalized past Referential strategies Prosodic structure

The present study indicates that important age-related differences exist at all levels of language structure in particular with respect to the functions with which various elements are used. The second general result was that at least some of the L2-related features turned out to be distinct from early L1, which means that L2 acquisition in certain important respects represents a different process.

Notes . In Viberg (1999), data from only 18 L2 children were presented. In this chapter additional data have been added in some cases from a second group of 12 L2 children, who were recorded simultaneously (cf. in Viberg 1999: fn. 2).

Age- and L2-related features in bilingual narratives in Sweden 

References Allén, S. (1971). Frequency dictionary of present-day Swedish. Stockholm: Almqvist and Wiksell. Andersson, A.-B. (1992). Second language learners’ acquisition of grammatical gender in Swedish. Ph.D. Diss. Gothenburg Monographs in Linguistics 10. Department of Linguistics, Gothenburg University. Anglin, J. (1993). Vocabulary development: a morphological analysis. Monographs for the Society for research in Child Development. No. 238. Univ. of Chicago Press. Berman, R. and D. Slobin (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Eldering, L. and J. Kloprogge (eds.) (1989). Different cultures, same school. Ethnic minority children in Europe. Amsterdam: Swets and Zeitlinger. Flege, J. E. (1991). Perception and production: the relevance of phonetic input to L2 phonological learning. In: T. Huebner and C. A. Ferguson (eds.), Crosscurrents in second language acquisition and linguistic theories. Amsterdam: John Benjamins, 249–289. Franzén, V. (1996). En studie över utvecklingen av några diskursparametrar hos barn med svenska som L2 och L1. (Term paper.) Department of Linguistics, Lund University. Genesee, F. (ed.) (1994). Educating second language children. The whole child, the whole curriculum, the whole community. Cambridge: Cambridge University Press. Gleason, J. (ed.) (1993). The development of language. 3rd edn. New York: Macmillan Publishing Co. Haiman, J. and S. Thompson (eds.) (1988). Clause combining in grammar and discourse. Amsterdam: John Benjamins. Hickman, M. (1995). Discourse organization and the development of reference to person, space, and time. In: P. Fletcher. and B. MacWhinney (eds.), The handbook of child language. Oxford: Blackwell, 194–218. Håkansson, G. (1992). Variation och rigiditet i ordföljdsmönster. In: M. Axelsson and Å. Viberg (eds.), Nordens språk som andraspråk. Centre for Research on Bilingualism, Stockholm University, 314–324. Juvonen, P., I. Lindberg, and Å. Viberg (1989). Narrative skills in two languages. Scandinavian Working Papers on Bilingualism 8. Centre for Research on Bilingualism, Stockholm University. Krashen, S., R. Scarcella and M. Long (eds.) (1982). Child-adult-differences in second language acquisition. Rowley/Mass.: Newbury House. Labov, W. (1972). Language in the Inner City. Philadelphia: University of Pennsylvania Press. Lindberg, I. (1995). Referential choice in Swedish narrative discourse. Referential strategies in the narratives of Finnish-Swedish bilingual students. Scandinavian Working Papers on Bilingualism 10, 1–104. Lindberg, I., P. Juvonen and Å. Viberg (1991). Att berätta på två språk. In: U. Nettelbladt and G. Håkansson (eds.), Samtal och språkundervisning. Tema Kommunikation, Linköping University.

 Åke Viberg

Long, M. (1988). Maturational constraints on language development. University of Hawaii Working Papers in ESL. Vol. 7:1. Long, M. H. (1990). Maturational constraints on language acquisition. Studies in Second Language Acquisition 12:3, 251–85. Long, M. H. (1993). Second language acquisition as a function of age: research findings and methodological issues. In: K. Hyltenstam and Å. Viberg (eds.), Progression and regression in language. Cambridge: Cambridge University Press, 196–221. McCabe, A. and C. Peterson (eds.) (1991). Developing narrative structure. Hillsdale, NJ: Lawrence Erlbaum. Miller, G. A. and P. M. Gildea (1987). How children learn words. Scientific American, Sept. 1987. Reprinted in: W. Wang (ed.) (1991). The emergence of language. Development and evolution. Readings from Scientific American. New York: W H Freeman and Co. Piéraut-le Bonniec, G. and M. Dolitsky (eds.) (1991). Language bases … discourse bases. Amsterdam: John Benjamins. Platzack, C. (1979). The semantic interpretation of aspect and Aktionsarten: A study of internal time reference in Swedish. Dordrecht: Foris. Romaine, S. (1984). The language of children and adolescents. Oxford: Blackwell. Schlyter, S. (1993). The weaker language in bilingual Swedish-French children. In: K. Hyltenstam and Å.Viberg (eds.), Progression and regression in language. Cambridge: Cambridge University Press, 289–308. Viberg, Å. (1993a). Crosslinguistic perspectives on lexical organization and lexical progression. In: K. Hyltenstam and Å. Viberg (eds.), Progression and regression in language. Cambridge University Press, 340–385. Viberg, Å. (1993b). The acquisition and development of Swedish as a first and as a second language: the case of clause combining and sentential connectors. In: B. Kettemann and W. Wieden (eds.), Current issues in European second language acquisition research. Tübingen: Gunter Narr Verlag, 293–306. Viberg, Å. (1994). Bilingual development of school-age students in Sweden. In: G. Extra and L. Verhoeven (eds.), The cross-linguistic study of bilingual development. Amsterdam: North-Holland, 181–198. Viberg, Å. (1996). The study of lexical patterns in L2 oral production. In: K. Sajavaara and C. Fairweather (eds.), Approaches to second language acquisition. (The proceedings from EUROSLA 2 in Jyväskylä in June 1992.) Jyväskylä Cross-Language Studies 17. Jyväskylä University, Finland, 87–107. Viberg, Å. (1998a). Lexical development and the lexical profile of the target language. In: D. Albrechtsen, B. Hendriksen, I. Mees and E. Poulsen (eds.), Perspectives on foreign and second language pedagogy. Odense: Odense University Press, 199–134. Viberg, Å. (1998b). Crosslinguistic perspectives on lexical acquisition: the case of languagespecific semantic differentiation. In: K. Haastrup and Å. Viberg (eds.), Perspectives on lexical acquisition in a second language. [Travaux de l’Institut de Linguistique de Lund 38]. Lund: Lund University Press, 175–208. Viberg, Å. (1999). Lexical development in a second language. In: G. Extra and L. Verhoeven 1. (eds.), Bilingualism and migration. Berlin: Mouton de Gruyter, 165–185.

Chapter 5

Sociocultural aspects of bilingual narrative development in Sweden Sally Boyd and Kerstin Nauclér

Mercer Mayer’s picture storybook Frog, where are you? (1980, hereafter The Frog Story) is probably the stimulus material which has been used most often and most successfully of all such material in the study of language and cognitive development (see e.g. Bamberg 1987; Berman and Slobin 1994). But what sort of activity is it, or rather, what different sort of activities can be carried out by looking at the pictures and telling the story of the boy, the dog and the runaway frog? In most of the studies using the Frog story, the aim is for the story to be told by a single individual, who has looked at the pictures one or more times in advance of telling the story to an experimenter. The experimenter sees the pictures while the story is being told, and it is clear to both that she or he already knows the story (Berman and Slobin 1994: 20–25). Berman and Slobin discuss some of the problems of collecting comparable Frog story data among young children so that what is elicited is a narrative rather than a series of picture descriptions, especially when it is clear to the child that the experimenter has access to the pictures and already knows the story. The use of a third person, who cannot see the pictures and who has not heard the story before, is suggested (Berman and Slobin 1994: 24–25) as one remedy, to make retelling the story less dependent on gesture and shared knowledge. Another major problem discussed by Berman and Slobin is that of the experimenter giving too much help in retelling the story, particularly when it is used with children. A hierarchy of acceptable prompts for using the Frog story with children is suggested, all of them minimal (e.g. ‘‘uh-huh’’, ‘‘go on’’). The problem they see with scaffolding or co-construction is primarily that less neutral prompts may influence the child’s ‘‘choice of verb tense, aspectual marking or perspective’’. The aim is to collect controlled, comparable ‘‘texts’’, with minimal input from an experimenter. When versions of the Frog story have been used for cross-linguistic comparison, they have mainly been used to

 Sally Boyd and Kerstin Nauclér

study the development of form and function in different languages (Berman and Slobin 1994) and their relation to cognitive development. The same procedure, or variations of it, have been used with adults and children of varying ages in a variety of languages and cultures. In the project Language socialization in immigrant families and its relation to language learning in the Swedish pre-school (1992–1995) we used the Frog story in a different way, to look at patterns of language socialization among Swedish and Turkish families and in Swedish preschools and the concomitant development of communicative skills among Swedish monolingual and TurkishSwedish bilingual preschool children in Sweden. For us, some of the ‘‘problems’’ discussed above have been at the center of our interest. The Frog story was one of four activities we asked adults both at home and at the preschool to carry out with the children who were the focus of our study. But in contrast to Berman and Slobin, but as in Bamberg’s earlier study (1987), we instructed the fieldworkers to ask the adults, after giving them a chance to look through the book, to ‘‘read’’ the story to the child in whatever way they felt seemed most natural. Thus, the task was first for an adult to tell the story while looking at the pictures with a 5–6-year-old child. Later, having had the story ‘‘told’’ to them two times by different adults, the children were asked to tell the story to the fieldworker while looking at the book. Our aim was not primarily to compare form and function in the stories as told by the adults and children in Swedish and Turkish, but rather to see how the somewhat novel activity of telling a story based on a picture series would be carried out by different adults and children in different settings. With other research on language socialization as a point of departure, we expected to find interesting and important differences in the way this and other activities were handled by adults and children of different ethnic background (Turkish and Swedish) in different settings (home and preschool). We hypothesized that differences in language socialization in certain activities may have important consequences for the development of reading skills later in school, and therefore for school success in general. Activities involving books and stories have been shown in other studies (see below) to be of particular importance for literacy development. Since our study was cross-cultural and partially cross-linguistic, it was natural to use a book without text, so that the material was the same for both groups of adults and children. The Frog story was also a book which was new both to the adults and the children in the study. Using the Frog story enabled us also to compare our results with those of others, although different methods have been used here, as compared to those in most other studies. It also

Sociocultural aspects of bilingual narrative development in Sweden

turned out that this activity was handled quite differently by adults together with Turkish children and adults together with Swedish children. This chapter will present the results of our study regarding how the Turkish and Swedish parents and the Swedish pre-school teachers carried out this task with different monolingual and bilingual children, and offers some tentative explanations for the differences found. We also discuss possible implications these different patterns of language socialization at home and in the preschool with Swedish and Turkish children may have for communication development in the context of the Western classroom. But first it may be necessary to look at what different sorts of narratives there are, and how the activity of telling the Frog story may fit into a more general framework of narrative and of discourse development in general.

The Frog story as a narrative According to Peterson and McCabe (1991), most researchers within the field of narrative research seem to agree that a necessary component of narrative, as opposed to other types of discourse, is a recounting of at least two events, removed from the recounting in both time and space (Peterson and McCabe 1991: 30, Norrby 1998). For the purposes of the present discussion, we will accept this as a definition of narrative. Within this broad category, distinctions can be made among a number of different types of narratives, based on the relationship between the person or persons doing the recounting and the person or persons whom the narrative is about (personal vs. vicarious), and whether or not the narrative is intended to recount events which actually happened or not (experience vs. fantasy). The Frog story would seem to be a story fitting into the ‘‘fantasy’’ category, although it is a relatively realistic fantasy story. All three types of narratives so far distinguished can be recounted as spoken or written narrative. Telling the Frog story is not clearly belonging either to the realm of spoken or of written narrative alone, as the pictures are found in a book, with (as Berman and Slobin 1994: 21, point out) a written language format: left to right and front to back. At the same time, there is no text (except for the title), so the story is normally created anew by one or more ‘‘tellers’’ each time through the medium of spoken language. We would like to suggest that we make a further distinction within spoken narrative of all types between what we will call performed narratives and cooperative narratives, based on the degree of participation or collaboration



 Sally Boyd and Kerstin Nauclér

between (primary) teller and listener in the activity — in our case between adult and child, respectively. Our distinction is similar to the one Norrby (1998: 68–69) makes between narratives in everyday conversation presented in dialogue vs. monologue form, and the distinction discussed below between Anward’s dialogue vs. monologue discourse in general. Since the adults and children we studied were not instructed to carry out the Frog story alone for a silent fieldworker, all the narratives collected involve a certain amount of dialogue or co-operation between child and adult. Anward (1983: 76–89) puts the genre of narrative (or story) in a larger context of discourse and in a developmental perspective. He claims that children acquire a repertoire of participant roles in the following hierarchy of discourse activity types, which increase in their degree of speaker autonomy: Directed dialogue → free dialogue → directed monologue → free monologue Directed dialogue is, according to Anward, the type of discourse we find in many adult–child conversations, from baby talk to classroom interaction, where the adult exercises maximum control over the choice of linguistic activity carried out, and of turntaking, but the child participates on the adult’s terms. Even clearer cases would be interrogation of a witness or a suspect. We call this sort of discourse, when carried out in the framework of narrative ‘‘asymmetric co-operative’’. What Anward calls free dialogue is typical of conversations among equals, where both activity and turntaking are decided collectively. We call this sort of discourse ‘‘symmetric co-operative’’. What Anward calls directed monologue is typical of meetings and unstructured interviews, where one person decides which activity should be carried out, and usually also controls turntaking, but attempts to minimize her/his own participation in the activity. Free monologue is found in speeches, radio programs, and in most writing. We call both types of monologue discourse when carried out as a narrative activity ‘‘performance’’. Parallel with this one, another hierarchy of discourse ‘‘genres’’ develop, according to Anward (1983: 70–75), which distinguish different relationships between participants and topics, or the degree of decontextualization of the discourse: ‘‘intervention/comment’’ → story → presentation The first genre is discourse about the here and now (e.g. including what Norrby 1998: 54, calls ‘‘momentaneous descriptions’’). Stories relate an event


removed from the here and now, but from one participant’s point of view; they utilize both textual and participants’ pre-knowledge as contextual background. Presentations normally have a neutral perspective and background is provided primarily if not only by the textual context. This hierarchy is thus one of what Snow (1991) calls ‘‘non-immediateness’’, or others call degree of decontextualization (e.g. Cummins 1996). Telling a story can thus be seen as an important activity in the context of language and communication development in the Western classroom, because it is often by means of this activity that children develop the skill of presenting a series of events both in speech and in writing (Hultman 1989). These skills can be seen as an inroad to more decontextualized or non-immediate writing and speech in school. During the school years, Anward claims, children develop primarily the skills of participant in directed dialogue and monologue (where the teacher is ‘‘director’’), as well as experience with the genres of story, and finally presentation. One way in which we can recognize the different types of narrative discourse in adult–child interaction is to see what sort of questions are asked by the adult. If the questions are of the ‘‘display’’ sort, where the adult already knows the answer to the question asked, we have strong evidence that the dialogue is ‘‘asymmetric’’. If the questions are referential, chances are that the dialogue is more ‘‘symmetric’’. In performance narrative, the most usual type of question would probably be feedback elicitation or rhetorical questions. Below is an overview of the different kinds of narrative discourse, with corresponding participant roles for the child and adult questions characteristic of each type. Narrative discourse type

Child’s participant role

Dominant type of adult question

Listener Feedback

Rhetorical, Feedback elicitation

Listener and speaker Listener and speaker

Display Referential

Performed (monologue)

Co-operative (dialogue) Asymmetric Symmetric

Being a fictional picture series for children, the Frog story will necessarily be a fantasy narrative carried out in a primarily spoken form, but with certain elements of written language. Since we asked the adults to carry out the activity with the children, but gave no more specific instructions, what varies is the degree to which the activity is carried out as 1) a performed narrative by



 Sally Boyd and Kerstin Nauclér

the adult alone, 2) as a co-operative asymmetric narrative directed by the adult or 3) as a co-operative symmetric narrative between adult and child. The narrative activities children were exposed to as listeners, and encouraged to participate in as speakers can be expected to reflect different ways of regarding knowledge, learning and displaying knowledge within different groups in society. We are not the first ones (see below) to predict that these differences might have important consequences when children enter school and are expected to acquire and display knowledge in appropriate ways. As the amount of participation of the child and the degree of direction exercised by the adult in the interaction varies, we find that the opportunities and ways for the child to display knowledge in various ways also vary.

Discourse at home and at school Previous research gave us reason to expect important differences in how certain communicative activities are handled in different settings (home vs. school) and by different cultural groups (Swedish vs. Turkish). These differences are often seen to have important implications for children’s success in school, a problem which has been noted repeatedly for non-mainstream children in general (Heath 1983, 1986), and Turkish children in Sweden in particular (Eyrumlu 1991; Taube 1994; Taube and Fredriksson 1995). The majority of researchers we consulted would seem to assume that, as Nilholm (1991) writes, ‘‘children’s appropriation of linguistic and conceptual repertoires [is] intrinsically linked to the communicative challenges that they meet in their interaction with adults’’. Discourse patterns used at home can be conducive to learning, as Tizard and Hughes (1984) have demonstrated, and there can be a considerable overlap, even in low-income homes, between discourse patterns in the home and at school, as Wells (1985) has shown, but the prevalence of certain discourse patterns in some homes would appear to give children from those homes a head start in acquiring the corresponding discourse patterns in school, according to Snow (1991). We call these ‘‘corresponding patterns’’ because we do not want to consider any speech activity as practiced in the home as ‘‘the same activity’’ as a corresponding activity in the school. Too many aspects of the speech situation necessarily differ for us to consider e.g. the telling a narrative of personal experience in school as the same activity as telling a narrative of personal experience at home. Even when parents engage in teaching activities, like Wells’


example of Rosie’s mother, who teaches her about the clock at home (Wells 1981: 210ff), we would argue that the activity is different in important ways from a situation where a teacher teaches about the clock in school. The number of participants in the interaction, their relationships with each other, the setting, the purpose of the activity, how the child’s behavior is judged etc. all differ in important ways between most if not all discourse activities in the home and in school. Still, skills learned at home clearly seem to have an effect on how one carries out activities in school. Heath (1983, 1986) tied the way parents in Roadville, a working class Appalachian piedmont community, approached text and story reading for children to the way they approached their most sacred text, the Bible. Bedtime stories were regularly read to children, but with little or no deviation from the text as printed, and with little or no relation made to the everyday life of the child. As Sjögren (1996) points out, in many societies, the Koran or Bible is learned by rote and by heart, and the scholar is expected to be able to quote at length from it. Quotation is considered an adequate support for an argument. We can also see similarities between the traditional teaching of the catechism, a preset series of questions and answers, learned by rote and the use of display questions + answer + evaluation in classroom interaction in Western European and North American classrooms, although teaching by means of display questions has an even more ancient history. The teacher or priest asks questions for which the pupil gives answers already known to the asker. All three of these forms of learning inspired by different religious traditions emphasize learning by rote the form rather than understanding the content of a text. While modern Western pedagogy emphasizes understanding content rather than repeating a text which has been rote-learned, classroom discourse form still reflects the tradition of the catechism.

Literate and non-literate cultures? One explanation offered for different discourse practices among mainstream and non-mainstream families has been that mainstream families are to a greater extent literate than non-mainstream families. A number of anthropologists as well as other scholars, from Levi-Strauss (1966) through Goody (1977) and Ong (1982) have suggested that cultures change fundamentally when written language is introduced and skills in reading and writing are acquired. Subcultures within Western industrialized societies are sometimes characterized as



 Sally Boyd and Kerstin Nauclér

having ‘restricted literacy’, as opposed to mainstream culture which has ‘full literacy’. Further claims have been made about this dichotomy having important consequences for the thought patterns of persons belonging to either group. Only those being brought up in ‘‘fully literate’’ social groups are claimed to have the capacity for logical and analytical thinking. According to Ong (1982), for example, the development of writing and print enable members of literate society to break down texts into smaller parts and to analyze the parts and their relationship to other parts. Ong claims that science as we know it would be impossible without writing and print. We would never be able to keep track of our own arguments or results, let alone those of others, if we had to rely on our memories. Since written language is an aid to talking about things removed from the here-and-now (non-immediate, in Snow’s terms), Ong’s line of argument is clearly relevant to the claim that children from nonmainstream families have difficulties with non-immediate discourse. Gee (1994: 168), while accepting that literacy and world view are closely connected, argues instead that there are different types of literacy in different societies, and that literacy has different social and mental effects in different social and cultural contexts. He writes: the discourse practices associated with our schools represent the world view of mainstream and powerful institutions in our society; these discourse practices and their concomitant world view are necessary for social and economic success in our society. But they are also tied to the failure of non-mainstream children in our schools and are rapidly destroying alternative practices and world views in less technologically advanced cultures throughout the world.’’ (Gee 1994: 169)

Whether we accept Ong’s evolutionistic view of the relationship between literacy and thought, or Gee’s more relativistic, we must in any case realize that the school expects children to learn the discourse practices of mainstream culture, including non-immediate uses of language in general and written language in particular. We think it is quite clear from the research referred to above that this involves a greater learning task for some children than others. One way the gap between preliteracy and literacy is often bridged in Western schools is by means of work first with oral, later with written narrative.

The activities studied In the study reported on here, the groups under investigation were children and adults in eight Turkish families living in Sweden, and seven working class

Sociocultural aspects of bilingual narrative development in Sweden 

Swedish families with children 5–6 years old attending Swedish pre-schools. The Turkish children are all born in Sweden. In fact, some have one parent born in Sweden of Turkish parents, making them partly second generation, party third generation Turks in Sweden. The preschools for both the Turkish and Swedish children were in typical working class, ethnically heterogeneous neighborhoods, where a large proportion of the children had languages other than Swedish as their first language. The mothers in both groups had at most 2–3 years of high school education, many had only compulsory school. For each child, recordings were made of adult–child interaction in the four activities mentioned above in the home (for the Turkish children, these activities were carried out with one exception in Turkish) as well as ‘‘the same’’ activity in the pre-school (for all the children in Swedish).1 Finally, each child was given a battery of language tasks to do individually with a fieldworker. These tasks were done, where possible, both in Swedish and Turkish by the bilingual children, and in Swedish by the monolingual children. The four activities recorded in the home and in the pre-school were Adult–child interaction: 1. A mealtime 2. Playing with a toy provided by us. 3. Looking at a family or pre-school photo album. 4. ‘‘Reading’’ the Frog story (i.e. an elicited narrative) Child alone: 5. Language tasks The ‘‘battery of language tasks’’ included a task where the child should, with minimal help of the field worker, retell the Frog story by her/himself. Interaction in the first four activities was intended to be centered around an adult (either the mother or the pre-school teacher, respectively) and the focus child. The role of the field worker was intended to be as minor as possible. Transcriptions of the entire elicited narrative activity, as well as of the second five minutes of the interaction in each of the other activities in each setting, together with transcriptions of the ‘‘language tasks’’ were made according to the CHILDES format (MacWhinney 1991). This chapter will focus on comparisons between cultural groups and settings as well as between children and adults for the elicited narrative activity, the Frog story. In this chapter we will look at only one aspect of the interaction in the elicited narrative activity, and compare the way the activity is handled by Turkish mothers and Swedish

 Sally Boyd and Kerstin Nauclér

mothers together with their children, and between Swedish pre-school teachers’ handling of this activity with Swedish and Turkish children. Different ways to tell a story Above, we presented a taxonomy of different ways a narrative activity might be handled, based on the Frog story as stimulus material, depending on the participant role of the child, and the degree of direction exercised by the adult, as indicated by the type of questions asked by the adult. We can get a rough idea of the difference in the role of the child in this activity, compared to the activities of Play and Meal in the table below, where we see the total number of turns by the Swedish and the Turkish target children in each activity. (These numbers include turns which consist only of feedback.) Table 1. Number of target children’s turns in different activities and settings Activity

Home

Preschool

Turkish

Swedish

Turkish

Swedish

MEAL (5 min.) PLAY (5 min.)

N=8 219 360

N=7 215 269

N=8 153 211

N=7 115 223

STORY

243

510

225

503

While the number of turns in the other activities is roughly comparable between the Turkish and Swedish children in both settings, there is a major difference in the total number of child turns in the Story activity in both settings. Note particularly that the there are only seven Swedish children and eight Turkish children, so the difference between the two groups is even larger than it first appears. This difference in the total number of child turns in this activity led us to look more closely at how this activity was carried out by adults and children of different groups and in different settings. The ways we outlined above to carry out the Frog story activity will be illustrated with excerpts from four Frog story activities carried out by adults and children. First we will have a look at an example from a Turkish mother and her 5-year-old daughter, Feliz. (Translation from Turkish.) (1) Mother: While they looked at the frog it got dark outside. It’s time to go to bed.

Sociocultural aspects of bilingual narrative development in Sweden 

Feliz: mm Mother: Ali and his dog get up on the bed and go to sleep. When Ali and his dog have fallen asleep the frog climbs carefully out of the jar. Look, because frogs can’t live in a jar. Frogs live by lakes. They live where there is water. Is it OK? Feliz: mm Mother: That’s why he carefully climbs out. The frog runs away from the house. Morning comes and the sun comes up. It gets nice and light. Ali wakes up. First he looks in the jar to see what the frog is doing. Then he sees that his frog is gone. It has left the jar. The dog looks too. They look together. Ali is surprised when he can’t see him and wonders where it’s gone. Look! they start to look for the frog right away. They look in the house. They lift up the furniture. They look in the boot in case it went in there. Did you see? Feliz: mm

In this example, which is an unusually clear example representing many of the Turkish stories at home, the mother is doing the work of telling the story to her child. She is describing the characters, their actions and motives, the causes and effects of various actions and events, while the child’s participation in the conversation is more or less restricted to feedback, elicited by the mother to check that the child is following the story and answering a limited number of questions asked mainly in the beginning and at the end of the story. In other words, it is the sort of discourse we would classify as ‘‘performance’’ in the taxonomy presented above. In the next example (2) a Swedish mother and her daughter, Mona, are ‘reading’ the same story. (Translation from Swedish. See pictures 2–4 in Figure 1.)

 Sally Boyd and Kerstin Nauclér

(2) Mother: Mona: Mother: Mona: Mother:

And now the boy and the dog have gone to bed but - and the boy is lying with his legs on the pillow. do you think so? and there you see the frog getting away. yes he sneaks away.

Picture 1.

Picture 2.

Picture 3.

Picture 4.


Mona:

and there you see in the morning the boy. sees and the dog that the frog has escaped. Mother: yes. Mona: and then Mother: - and they look – and look the boy looks in the boots and the dog looks in uh the jar. Mona: and look there at the dog. yeah except you see that there isn’t any frog there. Mother: no it has gotten stuck on his head now Mona: yes hehehe it looks funny.

This story is more of a collaborative activity, where Mona and her mother construct the story together. Both of them play the roles of speaker and listener and throughout the story both mother and child take turns in describing the different characters, their actions and motives, the various events happening to them, etc. In this respect, this story is representative of the stories told with Swedish children, both at home and in the preschool. In contrast to most of the other Swedish adults, however, Mona’s mother is not directing or dominating the activity to the same extent. She puts very few questions to the child, and when she does, they are mainly referential questions, i.e. requests for exclusive information that the questioner does not know the answer to beforehand. In other words, this sort of narrative would be classified as ‘‘symmetric co-operative’’ in the taxonomy above. In terms of adult dominance in the interaction, the next example (3) is more representative (albeit unusually clearly) for the stories told with Swedish children. This is an example from the preschool, and it is the same girl, Mona and her preschool teacher who are ‘reading’ the story together. (Translation from Swedish. See picture 2 in Figure 1) (3) Teacher: And the next picture, what do you think happens there? Mona: There’s the boy. The boy is sleeping and the frog escapes. Teacher: Yes, that’s right. And what time is it, day or night? Mona: It’s night. Teacher: And how can you see that Mona? Mona: You can see that because the boy is asleep. Teacher: You can see that, but you can be really sure that he’s not sleeping in the middle of the



 Sally Boyd and Kerstin Nauclér

day because eh if you look really closely there’s something that shows that it’s night. You see that? Mona: The frog. Teacher: What do you see through the window? Mona: A cow or something. Teacher: If you look really carefully, we’ll take your hair away, if you look, what do you see there? Mona: The moon.

This example — extreme as it may seem — illustrates the strategies commonly practiced in interaction with most of the Swedish children, not only by their teachers, but also to a certain extent by the Swedish mothers. The children are expected to participate actively in the construction of the story. However, with few exceptions, their contribution is severely restricted by the adult, who dominates the interaction, primarily through an abundant use of display questions, illustrated in (3), above. The function of these requests for information is twofold: one is to make the children display their knowledge of the world, knowledge which is already known to the questioner, the other is for the adult to check that the child possesses and/or is focusing on specific information at a certain point of time, information which is determined by the adult. In (3) we can also note the teacher’s evaluation of Mona’s answer ‘‘Yes, that’s right’’. What may pass as a correct answer is primarily related to what is in the teacher’s mind at the moment, not to what may seem reasonable in the actual context, i.e. that you may conclude it is night ‘because the boy is asleep’, which is a very reasonable answer to the teacher’s question. The interaction illustrated here between preschool teacher and child is more or less equivalent to the canonical Western classroom interaction, which to a large extent is built on sequences of question-answer-evaluation. In other words, this way of handling the Frog story activity is a clear example of ‘‘directed dialogue’’ as defined in the taxonomy above. Finally, we will have a look at an example of the interaction between a Swedish preschool teacher and the same Turkish child we saw in example (1). In (4) the teacher and Feliz are ‘reading’ the story in Swedish. (Translation from Swedish. See pictures 1–4 in Figure 1) (4) Teacher: Once upon a time there was a boy who was five years old. He lived in a nice house in the woods. And the frog lived in a glass jar.

Sociocultural aspects of bilingual narrative development in Sweden 

Now it was evening and the moon shone outside the window. They were soon gonna go to bed. And the boy lay down in his bed. But the frog was in his glass jar. And when the night came and the boy and the dog fell asleep, the frog sneaked out. He used his long legs and jumped out of the jar and was gone. Then when it was morning then the boy and the dog woke up and thought: Now we’re gonna play with our friend the frog. The jar was completely empty. Where could the frog be? They looked under the bed They looked. The boy looked in his boots and the dog he looked really carefully in the jar one more time. But no. There was no frog.

In contrast to their interaction with Swedish children, many Swedish preschool teachers handle this activity when they do it with the Turkish children as in the example above, i.e. as a performance very much like the majority of the Turkish mothers do. In fact, the Turkish children’s role in the conversation is even more passive in preschool than at home. There are only a couple of exceptions to this general picture. Unlike the Swedish children, the Turkish children are generally not invited to contribute much to the construction of the story and they are asked few questions. As in all human interaction, and specifically in learner data, there is of course a lot of variation in the stories we have collected. All the Frog stories have elements of co-operation, performance, symmetry and asymmetry. The types of discourse practices we have sketched should therefore be looked upon as generalizations across rather heterogeneous material. In spite of this variation, however, there is a significant difference between how this activity is carried out with Turkish and with Swedish children (Nauclér, ms.). In terms of the number of children’s turns in this activity, there is a statistically significant difference between the Turkish children and the Swedish children, and this is true for both settings, home and preschool.

 Sally Boyd and Kerstin Nauclér

All the Swedish children in our study were requested to construct the story together with the adult. At home, there were a couple of cases of a more symmetric co-construction of the story, like the one illustrated in (2). In most cases, however, the co-operation is an asymmetric one, both at home and in the preschool, and it is clear that all the Swedish children we studied get socialized into the norms for interaction in the Western classroom, whereas this is generally not the case for most of the Turkish children in this study. Judging from the results of our study, most of the Turkish children get used to listening to a story performed by an adult, both at home and in the preschool. There are some exceptions to this general picture where the Turkish children are requested to participate not only as listeners but also as speakers. A few of the Turkish children participated more than the others in the construction of the story at home, and also in preschool two of these children played a more active role in this activity. For a couple of Turkish children, who had very limited skills in Swedish, the activity in the preschool turned out to be more of a labeling game with very few narrative components. The performance strategy was however also practiced together with other Turkish children with limited skills in their L2. From our analysis it seems rather safe to conclude that the Turkish and the Swedish children are socialized into somewhat different norms of participation in this type of story activity. This type of literacy event is indeed handled rather differently together with Turkish and Swedish children, both at home and in the Swedish medium preschool. It should be emphasized, however, that despite their passivity in the Frog activity when carried out by the adults, the Turkish children manage quite well when asked to carry out the activity themselves. Their own Frog stories in Turkish are on average at least as long as those of the Swedish children, and even in Swedish their Frog stories were well constructed (excepting of course the two Turkish children who felt unable to carry out this activity in Swedish at all). They showed at least as great an ability to take a global perspective on the story, and to change perspective from that of the boy to other characters, as compared to the Swedish children (Veres and Boyd 1996). Their lack of active participation in the Frog activity when together with adults did not seem to imply that they were not learning the accepted way (both in Swedish and in Turkish (see Walker 1990–1993) of telling a story. Rather, they were not trained in the accepted ways of displaying knowledge, or invited to share their previous knowledge and experiences with the preschool teacher in that activity.

Sociocultural aspects of bilingual narrative development in Sweden 

Conclusions and discussion In this chapter we set out to investigate the ways in which Turkish and Swedish children and adults handle a story activity at home and in the Swedish mainstream pre-school. The results of the study indicate that the activity is handled differently by adults together with Turkish and Swedish children in both settings. In this section, we will discuss why these differences may arise and what the implications of them might be. We will start by discussing the differences we have found in the home setting.

Different activities at home The Turkish children have a very limited number of turns in this activity. Their contribution to the conversation is mostly that of a passive listener, and the activity is more of a performance on the part of the narrator, the adult. In contrast, the Swedish children have a great number of turns in the conversations, and the activity is a co-operative one. Both adult and child contribute in constructing the story. It is interesting to compare our results with those obtained by Heath (1986). In her study, she investigated children’s literacy socialization in families with different socio-economic background in the same Carolina piedmont community. She found that young children in middle class mainstream families were socialized in roughly the same way as regards books, literacy and reading which our study indicates that Swedish working class children are socialized. The norms for handling these activities in Turkish families, on the other hand, have certain parallels with those described by Heath for working class white families. There is thus no reason to believe that the difference we find between Turkish mothers on the one hand, and Swedish mothers on the other, is simply a question of cultural difference per se, i.e. Turkish vs. Swedish vs. American. Certain Turkish mothers actually carry out this activity in much the same way as Swedish mothers do. Neither can we say that differences in orientation to literacy simply are a question of socio-economic background, i.e. working class vs. middle-class, which has also very convincingly been shown by Wells (1985), since many of the Swedish working class mothers behave in ways similar to those of American middle class mothers. In our interviews with the children’s parents, Turkish and Swedish parents expressed different orientations to literacy in their families, and somewhat

 Sally Boyd and Kerstin Nauclér

different views on their own role for children’s language learning. Swedish mothers attached more importance to reading and other literacy events for their children’s language development than Turkish mothers did. Turkish mothers stressed more the importance of interaction in social activities as beneficial for the development of language. The Swedish mothers also reported that they read books, magazines and newspapers themselves to a greater extent than did the Turkish mothers. Some, but probably not all of this difference may be attributable to the lower availability of reading materials in Turkish. The differences in orientation toward and in practices with regard to literacy in the two groups may contribute to the different interpretations of the task made by Turkish and Swedish mothers (Sjögren 1996). There also seems to be a difference between the two groups of mothers in their views on how children may learn from the activity we asked them to carry out, i.e. the way children are expected to take meaning from texts, to use Heath’s phrase. Swedish mothers request their children to display their knowledge about different characters, events, actions and states in the story, more or less in the same way as the pre-school teachers do when interacting with Swedish children. This type of interaction is very similar to the typical Western classroom conversation. These Swedish working class mothers with 9–12 years experience of the interaction in the Swedish classroom often act like teachers in an activity which they judge important for their child’s language development (at least when they are asked by researchers to carry out the task). The Turkish mothers, on the other hand, who had little or no experience from a Western classroom, do not act like a typical Swedish (Western) teacher. They interpret the task as one in which they should perform a narrative and thus expect very little contribution on the part of the child. Two of the three Turkish mothers who had grown up in Sweden, on the other hand, handled the Story activity as a co-operative activity. Although the Turkish children seem to learn the norms for constructing a narrative, judging from their ability to tell the Frog on their own, not very surprisingly, they may not learn the Swedish sociocultural norms to display their knowledge in the Story activity together with the pre-school teacher.

Different activities in the pre-school This leads us to the differences found in Turkish and Swedish children’s interaction with their teachers in the Swedish mainstream pre-school. When interacting with Swedish children, the teachers handled the activity more or

Sociocultural aspects of bilingual narrative development in Sweden 

less like the Swedish mothers, using the same co-operative strategy, which resulted in a great number of turns on the part of the Swedish children in the pre-school setting as well. When interacting with Turkish children, on the other hand, the Swedish pre-school teachers handled the activity quite differently. With few exceptions, the teachers told the story to the children in the way the Turkish mothers did. The activity was in most cases a performance on the part of the teacher, which resulted in a very limited number of turns on the part of the child. There may be several reasons for this difference in the teachers’ interaction strategies towards Swedish and Turkish children. One is the fact that some of the Turkish children had a very limited proficiency in Swedish, their L2. However, the teachers used a performance strategy even with children who managed quite well in the task of telling the story in Swedish on their own. Furthermore, when pre-school teachers carry out this activity with Swedish monolingual children who had limited verbal proficiency, they never choose the performance strategy. In contrast to the Turkish children, these children get a lot of requests for information, often simple labeling questions, but still, they are prompted to display their knowledge in the interaction with the teachers in a typical classroom fashion. Thus, it cannot be the Turkish children’s lack of narrative skills or L2 skills in general that lead to different interaction strategies on the part of the preschool teachers. It is rather that the Turkish children are accustomed to another way of handling the storytelling task. But is this simply a matter of a mismatch between home norms and school norms? No, because the Swedish preschool teachers handle the same narrative task differently with Swedish children and with Turkish children. With Turkish children they handled the task much as the majority of the Turkish mothers did; with Swedish children they handled it much as Swedish mothers did. This difference in the way the Swedish pre-school teachers handle the activity with different children may be a result of their expectations of Turkish children, mainly based on the children’s limited proficiency in lexical and structural aspects of their L2. Because they are unsure about the children’s level of understanding of the story as presented in Swedish, they limit the child’s role in the activity to that of giving minimal feedback. As a reflection of these expectations, whatever their source, teachers do not invite the Turkish children into the conversation to the same extent as they do with Swedish children, and they do not request the Turkish children to display or share their knowledge of the world with the teacher.

 Sally Boyd and Kerstin Nauclér

There are at least two possible interpretations of the teachers’ interaction with Turkish children in this activity. One is that the they accommodate to the children’s sociocultural norms in this activity, and hence the performance strategy might be seen as an expression of a principle of accommodation. It might be the case that teachers respond to the children’s quiet listening strategy by carrying out a performance narrative with them. Looking at the interaction in pre-school in a wider societal perspective, however, the interaction between teachers and children can be seen as a reflection of the power relations between majority and minority members in the surrounding society. In this perspective, the interactions between the preschool teachers and the Turkish children reflect the unequal power relations between the Turkish minority and the Swedish majority in Swedish society (Cummins 1996). Together with majority children, teachers use a co-operative strategy, capitalizing on the children’s knowledge and prior experience, which in many cases is similar to or at least predictable by the preschool teacher. This is often not the case in interactions with minority children. The minority children’s knowledge and prior experiences are not as well known to the teacher, and are not equally valued in the pre-school context. This tendency can be illustrated by the following short excerpt from near the end of the Frog story as told to Feliz by her preschool teacher. As Feliz has already heard the story once, as told by her mother, Feliz wants to contribute to the telling of the story in the preschool. However, the teacher clearly wants to retain control of the activity. (5) Feliz: Now I know! Teacher: You know that. But what is happening here? Now the dog caught sight of something. Shhhh said the boy Feliz: Not the dog. Teacher: There they went and both of them looked behind the log.

The performance strategy used by the teachers may thus be seen as a reflection of their limited expectations on Turkish children in general and on their performance in literacy activities in particular. Interactions always entail a process of negotiating identities; who has a valuable contribution to give and who has not; who has something to tell, and who has not. The outcome of this process in the pre-school context may not always promote a positive image of self for minority children.

Sociocultural aspects of bilingual narrative development in Sweden 

Implications Let us finally return to one of the research questions addressed by our study, namely why it is that many minority children experience a lack of success in the mainstream school. In previous research, it has been claimed that part of the problem is that some minority children do not learn the ‘appropriate’ norms for Western classroom interaction. The Swedish children in this study appear to learn these norms both at home and in the pre-school. Most of the Turkish children did not, at least to judge from the activity we asked the mothers and teachers to carry out, neither at home, nor in the Swedish medium pre-school. We are not claiming here that differences in sociocultural norms for interaction influence Turkish children’s learning and knowledge directly, but that, judging from our results, many Turkish children get only limited opportunities to learn to display their knowledge in ‘appropriate’ ways. This may in turn have consequences for their interaction with mainstream teachers in pre-school and school, where many minority children’s knowledge is made invisible and where their experiences are not requested and appreciated. The fact that the Swedish pre-school does not provide the Turkish and Swedish children with the same opportunities in this respect is a cause for concern. The more so, as Swedish medium pre-schools (rather than minority language medium preschools) have been advocated as a solution to the problems many minority children experience at school. Limited skills in the second language are only part of the reason for problems in acquiring literacy in school, however, and perhaps not the primary part either. Neither is the source of the problem so much a mismatch between the Turkish children’s cultural practices at home and those at the preschool, but rather the fact that in this activity and perhaps in others as well, they receive different treatment by Swedish preschool teachers than what Swedish children receive. In other words, the source of the problem is at least as much centered in the context of the receiving society as in the context of the child’s family (cf. Rojas 1995). Lack of experience with typical classroom interaction, and with ‘appropriate’ norms of displaying knowledge will continue to have consequences for minority children’s success in schools, as long as mainstream schools — and pre-schools — continue to assume there is only one way to learn and one series of steps to achieve it. In a recent follow-up project, we have carried out a test of reading comprehension among the children studied in the project described above, who are now in the fourth grade in school, and we have interviewed teachers and

 Sally Boyd and Kerstin Nauclér

parents about the children’s language and reading development. The results of this project will help to determine the extent to which it is justified to draw conclusions about language skills in general, and reading comprehension in school in particular, based on exposure to different discourse practices at home and at school.

Notes . We would like to thank the field workers and transcribers who assisted with the field work and transcriptions of the Turkish material: Meral Ünwer and Gunel Meydan.

References Anward, J. (1983). Språkbruk och språkutveckling i skolan. Lund: Liber. Bamberg, M. (1987). The acquisition of narratives: Learning to use language. Berlin: Mouton. Berman, R. and D. Slobin (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Cummins, J. (1996). Negotiating identities: Education for empowerment in a diverse society. Ontario Institute for Studies in Education. California Association for Bilingual Education. Eyrumlu, R. (1991). Turkar möter Sverige. Stockholm: Carlssons förlag. Gee. J. P. (1994). Orality and literacy: From the savage mind to ways with words. In: J. Maybin (ed.), Language and literacy in social practice. Clevedon, UK: Multilingual Matters. Goody, J. (1977). The domestication of the savage mind. Cambridge: Cambridge University Press. Heath, S. B. (1983). Ways with words. Language, life and work in communities and classrooms. Cambridge: Cambridge University Press. Heath, S. B. (1986). What no bedtimes stories means: narrative skills at home and school. In: B. Schieffelin and E. Ochs (eds.), Language socialization across cultures. Studies in the Social and Cultural Foundations of Language. Cambridge: Cambridge University Press. Hultman, T. (1989). Skrivutveckling i ett pragmatiskt perspektiv. In: C. Sandqvist and U. Teleman (eds.), Språkutveckling under skoltiden. Lund: Studentlitteratur, 117–142. Levi-Strauss, C. (1966). The savage mind. Chicago: University of Chicago Press. MacWhinney, B. (1991). The CHILDES project: Tools for analysing talk. Hillsdale, NJ: Lawrence Erlbaum. Mayer, M. (1980). Frog, Where are you? New York: Dial Press Nauclér K. (ms.). Differences in norms for interactions with Turkish and Swedish children. Göteborg: Dept. of Linguistics, Göteborg University.


Nilholm, C. (1991). Communicative challenges. A comparative study of mother-child interaction. Linköping: Linköping studies in arts and sciences, 64. Norrby, C. (1998). Vardagligt berättande. Form, funktion och förekomst. Nordistica Gothoburgensia 21. Göteborg: Acta Universitatis Gothoburgensis. Ong, W. J. (1982). From orality to literacy. The technologizing of the word. London: Methuen Routledge. Peterson C. and A. McCabe (1991). Developing narrative structure. Hillsdale, NJ: Lawrence Erlbaum. Rojas, M. (1995). Sveriges oälskade barn. Att vara svensk men ändå inte. Stockholm: Brombergs. Sjögren, A. (1996). Brist på respekt, men för vem? In: Å Daun and B. Klein (eds.), Alla vi svenskar. Stockholm: Nordiska museet, 61–81. Snow, C. (1991). The theoretical basis of the home school study of language and literacy development. The social prerequisites of literacy development: Home and school experiences of pre-school-aged children from low-income families. Home-School Study of Language and Literacy Development. Harvard Graduate School of Education, Cambridge, MA. Taube, K. (1994). Läsförmågan hos 5 325 nioåringar i Stockholm. Stockholms skolor utvärderar, 1994, 11. Taube, K. and U. Fredriksson. (1995). Hur läser invandrarelever i Sverige? Skolverkets rapport nr 79. Stockholm: Skolverket. Tizard, B. and M. Hughes (1984). Young children learning. Boston, MA: Harvard University Press. Veres, U. and S. Boyd (1996). Perspective-taking in Frog Stories by Swedish monolingual and Turkish bilingual children and adults. Paper presented at VIIth International Congress for the Study of Child Language. Istanbul. 14–19 July 1996. Walker, B. (1990–1993). The art of the Turkish tale. Lubbock, TX: Texas Tech University Press. Wells, G. (1985). Language development in the preschool years. Cambridge: Cambridge University Press.



Chapter 6

The development of co-constructed narratives by Turkish children in Germany Carol W. Pfaff

Research in sequential bilingual language acquisition (SBLA) of children of minority populations has burgeoned in recent years, particularly in North American studies of Hispanics and European studies of children from Mediterranean countries. Such studies address not only important theoretical issues of language development in contact situations, but also numerous pressing practical issues related to the schooling of such bilingual children. This chapter is devoted to aspects of language development in individuals and to the possible mechanisms of language change in the migrant community which may be seen in progress in the interactions between children and adults as they co-construct narratives and conversations. The development of SBLA children’s grammatical proficiency in the second language has been the focus of numerous studies and attention has also been paid to bilingual children’s development of communicative competence in L2. Increasingly, this work on child L2 acquisition has been paralleled by work on the development of grammatical and communicative competence in the first language, with studies of children’s proficiency, use and subsequent shift away from use of the ethnic minority language and the (interim) development of mixed varieties in the minority communities. However, unlike the work on monolingual first language acquisition (MFLA) and simultaneous acquisition of two first languages (BFLA), which have also considered the role of input and interactional strategies of adult (and child) interlocutors, as summarized in Snow (1995), de Houwer (1995) and Lanza (1997), much less attention has been paid to input and interaction in SBLA. The role of the interlocutors in the development of varieties in SBLA is an area of investigation which deserves more attention and this is the topic addressed in the present chapter. Adult interlocutors’ speech is important to the development of the child’s abilities in several ways. First, the contributions of the interlocutors provide

 Carol W. Pfaff

linguistic input which, if comprehended linguistically and pragmatically, can be assimilated by the child, and contribute to the development of proficiency and to broaden the child’s repertoire of interactional strategies in the first or second language. Second, the interlocutor facilitates (or occasionally hinders) the communication in the immediate ongoing interaction. Third, the speech of the interlocutor can steer the course of the conversational interaction, strongly influencing the child’s production toward or away from particular grammatical and lexical structures as well as conversational foci and strategies, which provide the material available for analysis and interpretation. This chapter explores aspects of adult interlocutors responses to two bilingual children from the longitudinal KITA study of Turkish/German SBLA,1 drawing on transcripts which document the speech of the adult interlocutors and other child participants as well as the target-child utterances, recorded in co-constructed narratives and conversations. In this chapter, we look at the strategies that adult interlocutors, primarily the project staff interviewers and also occasionally older siblings and parents, use in conversations with two target children as they progress from day care (nursery school and preschool) and begin primary school. The setting and methodology relevant for the interactional aspects addressed in the present chapter are presented in the next section.

The KITA study Setting and methodology Tape recordings of child-adult interactions in Turkish and German were initially made at the Kita, separately for each language, usually on different days, usually by a pair of adult researchers, Turkish/German bilinguals for the Turkish recordings, Germans with some knowledge of Turkish, for the German recordings. Sometimes other children and caretakers were present as well as the ‘‘interviewers’’. Recordings at the home of the child were carried out by the same Turkish or German interviewers who had made recordings at the Kita. On these occasions parents or siblings of the target child were sometimes present and occasionally participated in the interactions. Their presence allowed us to record the target child’s more usual ‘bilingual mode’ as well as his or her more ‘monolingual mode’, which was elicited in the more directed conversation with the researchers.

Co-constructed narratives by Turkish children in Germany

A variety of elicitation techniques were employed: playing with toys, looking at books or pictures, free conversation (personal narratives about home and family activities, vacations, etc.) and elicited narratives of fairy tales and traditional stories. All of these activities involved verbal interaction between the children and their interlocutors (interviewers, other Kita children, family members in recordings made at the child’s home), and these were more or less ‘‘conversational’’ as the child’s proficiency and communicative competence developed.

Co-constructed narratives and conversation in the KITA study interactions Our goal was to elicit speech that would be comparable across languages for different children of varying proficiency. Further we wanted to investigate the children’s abilities to communicate non-shared as well as shared information. Thus the elicitation of conversation about events and experiences beyond the range of the ‘‘here and now’’ was integrated as much as possible into all of the elicitation techniques. Child digressions into past experiences (personal narratives, fantasy adventures, description of films or videos they had seen at home, etc.) were frequently invited and pursued when they occurred spontaneously. Figure 1 represents our conception of how the narrative events in our study are embedded in the conversational interactions with the researchers, which are themselves embedded in the ongoing interactions and events in the Kita or at home, which in turn are part of the children’s (and adults’) wider experience.

Outside world (including participants’ knowledge and ideas about present, past and future states and events) Setting and on-going interaction(s) among children and adults present

Conversation between target child and interlocutor(s) Narrative

Figure 1. Place of narrative in conversational interactions in the KITA study



 Carol W. Pfaff

As indicated in Figure 1, we conceive of all of the frames of reference as providing potential topics the child-adult interaction. Even when we used picture books or elicited performances of traditional stories, we were not attempting to elicit monologic narratives, but, rather were interested in the development of conversation, using the books and other stimulus materials as a starting point for discussion of personal experiences and extended general commentary. Thus, in contrast to the ‘‘Frog Stories’’ (Bamberg 1987; Berman and Slobin 1994) or similar elicitation techniques, our interactions show much more involvement of the interviewers in the co-production of the narrative and discussion of encyclopedic knowledge of the world. In this respect they are more like sessions with parents or caretakers. This shows up particularly in the amount of scaffolding provided by the interviewer, particularly in the children’s weaker language, with a concomitant relatively high frequency of content questions posed by the interviewer, which often resulted in elliptical answers on the part of the children.

Adult contributions to co-constructed narrative and conversation Several countervailing forces shape the form and content of the ‘‘interviews’’ in the KITA study. Both child and adult participants are interested in accomplishing the (implicit) goals of the interactions: getting the stories told, exchanging information and opinions including background about the interlocutors’ life and language use outside the interview context. In addition, as researchers, we wanted to elicit as much comparable speech in the monolingual modes as possible from children whose proficiency was often rather low. This influenced us to keep our interventions and scaffolding to a minimum and to refrain from modeling structures that we wanted to investigate. On the other hand, we wanted to facilitate extended discourse and, as common for Western middle class adult caretakers and teachers, we were interested in providing useful linguistic input and feedback to the children as possible without turning the interaction into a language lesson with many explicit negative and positive responses to the children’s utterance structures. Our scaffolding thus involved several global and local strategies, including considerable use of questions, varying in form as required by the child’s proficiency, to introduce or change the frame of reference or to further the narrative or conversational flow. Our policy with respect to language mixing or grammatical ‘‘errors’’ was not to interrupt the flow of conversation to insist

Co-constructed narratives by Turkish children in Germany 

on ‘‘appropriate language’’ vocabulary or forms. Rather we attempted to incorporate translation equivalents and standard grammatical forms in recastings and expansions. How this worked in practice is discussed below, in terms of the following three strategies employed: (1) providing explicit metalinguistic comment and models of linguistic form; (2) providing implicit models of forms in recasting, correction and expansion of content; and (3) no reaction, simply accepting the child’s utterance as legitimate in the ongoing interaction as part of the flow of the narrative or conversation. Before turning to our examination of the interviewers use of these scaffolding strategies, we briefly introduce the two target children, Ilknur and Serkan, both second generation children of migrants from Turkey who live in the neighborhood of the Kita but whose personal biographies have resulted in different patterns of language dominance and language use.

The target children, Ilknur and Serkan Ilknur (‘‘T16’’ in previous chapters), whose language development we followed from 3;10–7;02, is the youngest girl in a large extended family which migrated to Berlin from the Black Sea area of Turkey. Some of her immediate family members still live in Turkey or have returned. Turkish is the language generally used at home both with adults and frequently also among the children. Ilknur remained Turkish-dominant throughout the period of investigation and her Turkish is very similar to that of monolingual children in Turkey (Pfaff and Savas¸ 1988). Her direct regular contact with German began when she entered the Kita, but it remained her weaker and less-preferred language during the period of investigation. Though her proficiency and communicative competence in her second language developed very significantly, her grammar remained noticeably nonstandard. Serkan (‘‘T05’’ in previous chapters), whose language development we followed from 4;03–7;02 , is the middle of three sons who, prior to and at the outset of our study, spent some time away from his family in a children’s home with mainly German children and caretakers. As a result, German had already become his dominant language by 4;03. When he subsequently returned to live at home, a German-speaking friend of the family was frequently present and Serkan commonly used German with him, with his brothers and also frequently replied in German to his mother’s Turkish or mixed language utterances. Although both the neighborhood and the Kita are

 Carol W. Pfaff

predominantly Turkish-speaking environments, Serkan remained Germandominant. According to his self-reports and our observations he was less and less inclined to use Turkish and had rather low proficiency in it, particularly in complex syntax and vocabulary (Pfaff and Savas¸ 1988; Pfaff 1993). Despite stuttering at times his German is very fluent, though nonstandard in many respects. The following section summarizes our observations on the relationship of Ilknur’s and Serkan’s proficiency in Turkish and German and the communicative strategies they employ to accomplish their conversational goals.

Language dominance and conversational strategies In Pfaff (1993) and Pfaff (1994), I sketched out a framework for linking child proficiency with child’s conversational strategies, introducing the notions ‘‘global and local strategies’’ in linking the child’s proficiency to his/her conversational strategies. Global strategies relate to the children’s interaction with the interlocutors and local strategies relate to the narrative perspective and rhetorical, grammatical and lexical devices employed by the children as they manage to satisfy the requirements of their communicative ‘‘tasks’’ as successfully as possible given their linguistic proficiency. For instance, in a coconstructed narrative about a picture book, children can adequately participate by producing descriptions of the individual pictures or by producing a connected narrative, or by offering us their opinions and suggestions about how the characters in the story should have acted. Table 1 gives a schematic overview of our findings on the global strategies employed by these two children in each language. Some of these characteristics are illustrated in the following parallel examples2 from the narrative elicited with the picture book Lady and the Tramp, introducing the first episode, in which a man gives his wife a puppy for Christmas. In the examples (1)–(4) below, all of which are adequate and appropriate in the context of the co-constructed narratives, we can observe how the children employ production strategies which allow them to make the most of their (limited) competence as seen in their use of lexical mixing and of nonstandard forms of various types and their employment of various global strategies in minimal or extended utterances (see Pfaff 1994 for extended commentary on these and other parallel utterances from the same narrative).

Co-constructed narratives by Turkish children in Germany 

Table 1. Global and local converstaional strategies in Turkish and German for Turkish-dominant and German-dominant children Turkish-dominiant girl, Ilknur

German-dominant boy, Serkan

In Turkish

In German

In Turkish

In German

Rarely Frequently (to personal narrative or conversation) Little

Very rarely Frequently (to here and now activity)

Rarely Frequently (to here and now activity)

Much

Much

Sometimes Frequently (to personal narrative or conversation) Little

Rarely

Frequently (to sisters)

Rarely Frequently (to interviewer or mother)

Local strategies Deictic reference Specifies detail Questions interlocutor

Some Frequently Rarely

Frequent Rarely Rarely

Quotes direct speech

Sometimes

Frequently (formulaic)

Very Frequent Rarely Sometimes (for vocabulary) Rarely

Global strategies Initiates topics Digresses to a different activity or topic Relies on scaffolding Delegates turn to another

(1) Ilknur (5;06) Turkish burda Weihnachtsmannbaum@s [*] yapmıs¸lar (2) Ilknur (5;07) German: Tannenbaum [Interviewer mentions gift of puppy] ‘‘Herzlichen Glückwunsch x Weihnachtsmann’’ (3) Serkan (6;00) Turkish bu Weihnachten@s diye (4) Serkan (6;00) German guck mal, der Mann hat ihn [*] als [*] Weihnachten ein kleines [*] Hund gekauft.

Some Frequently Sometimes (for information) Sometimes

here (they) made a Santa Claus tree@s Christmas tree [Interviewer mentions gift of puppy] ‘‘Best wishes x Santa Claus’’

this [= tree] (is) for Christmas@s look, the man bought him [=wife] a little dog for Christmas

 Carol W. Pfaff

In the following sections I offer examples of some ways interlocutors responded to these and other types of language mixing and nonstandard usage of the two children and discuss how such interventions (or lack of interventions), with explicit or implicit modeling of lexicon and grammatical structures in Turkish and German interactions in the monolingual mode, may shape the notions the children form of the lexicon and grammar of the ‘‘target’’ languages.

Interlocutors’ reactions to language mixing Discussions of language mixing in language contact situations, independent of the stance taken with respect to the issues of borrowing vs. codeswitching or the nature of grammatical constraints, has generally recognized different patterns of usage depending on the grammatical category of the mixed item and the length of the mixed sequence. Here my attention is restricted to insertion of single nouns from one language into ongoing interaction in the other language, excluding longer alternations and switches of other syntactic categories: verbs, adverbs and non-constituent switching. In these examples the ‘‘grammaticality’’ of the mixed utterances is not at issue; all of the examples here are either integrated into the matrix language morphologically or are bare forms which occurred in syntactic position where such forms are grammatical. As reported in Pfaff (1991) and Pfaff (1998), mixing into the two languages is asymmetrical in this Turkish/German SBLA setting, as is generally found in such minority language communities. Mixing of Turkish into German sentences declines rapidly for all children; while mixing of German into Turkish increases rapidly, spreading from insertion of single lexical items to ragged switching to complete alternation to German by children with extensive contact to German peers, and increasing in frequency, but still mostly limited to insertion by children with less contact to German peers. Our two target children here fit these patterns.

Responses to German lexical items in Turkish Serkan is a child who, as noted above, has almost completely shifted to German, though he still understands and is capable of carrying on conversations in Turkish. His Turkish, however, is full of lexical items from German and longer switched stretches. Here are a few examples of how adult interlocu-


tors, our interviewers and his mother, react to his usage of single nouns from German in Turkish. In example (5), Serkan and Filiz, the interviewer are talking about what kind of food he likes to eat at home. Here we see that the interviewer accepts and repeats his use of the German term pommes frites ‘French fries’ — itself a loanword from French also adopted in Turkey. Serkan brings up the question of translation equivalents here and Filiz has some difficulty coming up with a Turkish term patates kızartması ‘fried potatoes’. She uses the opportunity to shift the conversation to the topic of his language use at home. (5) Serkan (5;00) Turkish: conversation about food Serkan: ama pommes frites@s da [*] Filiz: pommes frites@s deg˘il mi yapıyo? Serkan: mhm ama pommes frites@s Türkçe ne demek? Filiz: ne demek? aa canım benim. Türkçe ne demek? Türkçe ne demek? nasıl deriz? patates kızartması. Filiz: peki Serkan sen evde Türkçe mi konus¸uyorsun Almanca mı?

but pommes frites@s too She makes pommes frites@s too? mhm but what are pommes frites@s called in Turkish? What are (they) called? oh sweetie What is (it) in T.? what is (it) in T.? how do we say (it)? fried potatoes Ok Serkan, what do you speak at home Turkish or German?

In (6) from the co-constructed narrative based on the picture book Lady and the Tramp, the interviewer first introduces reference to the mouse in Turkish and gets involved in a digression about whether it is dogs or cats that usually chase mice. Serkan, who probably did not previously know the word fare ‘rat’, informs Tülay of the translation equivalent in German, Ratte ‘rat’. She agrees but uses the Turkish terms fare and sıçan, ‘rat’ and ‘mouse’, to which he responds by stressing the German, as if he has taken on the role of teacher here.3 (6) Serkan (6;00) Turkish: picture book, Lady & Tramp, Lady sees rat in the garden Tülay:

hemen aklına o geliyor fareyi it comes to your mind görünce. immediately when you see the mouse.



 Carol W. Pfaff

Tülay: sonra? Serkan: o zamn@c [*] böyle yapıyo yicek diye. Tülay: köpek yer mi fareyi dersin?

then? then (she) does like that because (she) will eat (it) do think (the) dog eats the mouse? Serkan: ııh no Tülay: kimler fareleri yerler, who [pl] eats the mice, catches yakalarlar? (them)? hangi hayvanlar farelerin which animals run after mice, pes¸inden (and) catch kos¸ar yakalar biliyor musun? (them), do you know? kediler, deg˘il mi? cats, right? kedilerle fareler hiç cats and mice don’t understand anlas¸amazlar. each other at all Serkan: bu Ratte@s this (is a) rat@s Ratte@s rat@s Tülay: mhm fare, sıçan. mhm, mouse, rat Serkan: Deutsch@s Ratte@s German@s, ‘‘rat’’@s

On other occasions, the use of German items in Turkish is simply accepted and even taken up by the adult interlocutors, as in (7) where Serkan and his mother are looking at a book about playing doctor: (7) Serkan (6;08) Turkish: book about playing doctor Mother: bu ne yapmıs¸? what has this (person) done? Serkan: o da Krankenschwester@s olmus¸ and she became (a) nurse@s Mother: ja@s Krankenschwester@s olmus¸. yes@s, (she) became (a) nurse@s

In (8) Serkan is talking with Tülay about another book and points and refers to two families, using the German word Familie ‘family’ rather than the Turkish word aile. She accepts this term, but questions more closely about the kind of family to which Serkan responds that it is a German family. We can only speculate about whether Tülay would have reacted differently had the family portrayed been obviously Turkish. (8) Serkan (6;10) Turkish: picture book Serkan: Tülay: Serkan: Tülay:

orda bi tane Familie@s mhm orda da bi tane Familie@s. mhm

there (is) one family@s mhm and there (is) one family@s mhm

Co-constructed narratives by Turkish children in Germany 

kim bunlar ama? nasıl bir Familie@s bu? Serkan: Alman Familie@s. Tülay: mhm

but who (are) these (people)? what kind of a family@s (is) this? (a) German family@s mhm

We turn now to look at some examples of reactions to Ilknur’s use of German in Turkish. The term Eis ‘ice cream’ is introduced into conversation by Ilknur in a derived form Eis’çıya ‘to the ice cream man’, which has both derivational and inflectional Turkish morphological markers denoting motion toward the ‘‘professional’’ ice cream seller. She also uses the stem form to denote the food itself. The term Eis is widely accepted in the local community and is thus readily available for derivational processes, though the derived form is also current and likely to have been part of Ilknur’s earlier input, not necessary her original coinage. In the interaction in (9) both forms are accepted and repeated by one of the interviewers, while the other one provides the Turkish translation equivalent for the food term, dondurma ‘ice cream’. (9) Ilknur (5;03a) Turkish: picture book, children and ice cream man Ilknur: sonra da Eis’çıya@s geldiler. and then (they) came to the ice cream man@s Eis@s yediler sonra da. and then (they) ate ice cream@s Tülay: ne yediler? what did (they) eat? Ilknur: Eis@s ice cream@s Filiz: Eis’çıya@s geldiler (they) came to the ice cream man@s Eis@s yediler. (and) ate ice cream@s Tülay: dondurma yediler deg˘il mi? (they) ate ice cream, right? Ilknur: mhm mhm

In (10), Ilknur has just given the Turkish word for ‘Kindergarten’, when she uses the German word Vorschule ‘preschool’ (an institution which does not exist in Turkey), the interviewer responds with a question, as if she did not understand acoustically, but which may be understood as a prompt to say it in the other language. But after Ilknur simply repeats the form, the interlocutor takes the German form up too. (10) Ilknur (5;03b) Turkish: talking about children playing games Filiz:

nerde oynuyorlar bu oyunu? where are (they) playing this game Ilknur: yuvada oynuyorlar. (they) are playing (it) in (the) kindergarten

 Carol W. Pfaff

Filiz:

yuvada. olurmu onlar çocug˘um? biraz büyük gibi geldi bana. Filiz: hm? Ilknur: Vorschule@s çocukları. Filiz: efendim? Ilknur: Vorschule@s çocuklarıdır. Filiz:

Vorschule çocuklarıdır evet olabilir.

in/at (the) kindergarten really, are these children? they seem a little old to me hm? preschool children pardon? they are probably preschool children. they are probably preschool children. yes, could be.

In (11) Ilknur paraphrases ‘nurses’ as hastane kizlar ‘hospital girls’; then, when Filiz asks for the precise lexical item, Ilknur gives the German translation equivalent, Schwester explicitly noting that it is German. Filiz asks for the Turkish translation equivalent, Ilknur hesitates and then Filiz supplies two alternatives Turkish terms hastabakıcı ‘one who takes care of the sick’ or hems¸ire ‘nurse’, her suggested term. (11) Ilknur (6;04) Turkish: picture book, playing doctor Filiz:

bu kız kim? ne giymis¸ böyle üstüne? Ilknur: ähm doktor es¸yası Filiz: doktor es¸yası mı giymis¸ Ilknur: ııh 〈hastane〉 [/] hastane kızların gibisini giymis¸.

who is this girl? what did she put on? um doctor things she put on doctor things? no hospital she put on things like hospital girls wear Filiz: hastane kızların gibi giymis¸. she put on things like hospital girls wear Ilknur: mhm mhm Filiz: ne deriz biz what do we say for these women who work o hastanede çalıs¸an kadınlara? in (the ) hospital Ilknur: ähm ¸sey Almancası um in German thing sister@s Schwester@s. [nurse] Filiz: Türkçesi? in Turkish? Ilknur: Türkçesi ähm + in Turkish um + Filiz: hastabakıcı ya da hems¸ire. ‘‘nurse’’ or ‘‘nurse’’ [alternative terms] Ilknur: mhm mhm Filiz: hems¸ire deriz deg˘il mi? we say ‘‘hems¸ire’’ don’t we?

Co-constructed narratives by Turkish children in Germany 

Responses to Turkish lexical items in German As reported in Pfaff (1991), the children in the KITA Study use Turkish elements in German recordings only in the very earliest states of acquisition, at a time when their conversational abilities are restricted primarily to talk about ‘‘here and now’’ information in the context shared with the interlocutor. Typically, lexical gaps for names of the toys and objects or actions represented in the picture books filled by Turkish items receive explicit comment and expansion. At the outset of our investigation, Serkan was already well beyond this stage in his acquisition of German. There is only one instance, in (12), in which Serkan uses a Turkish word in a German context. This is the term helikopter ‘helicopter’ for the standard German Hubschrauber. The interviewer accepts his use of the Turkish word here without comment, possibly because she knows he knows the German term already, as is evident from his use of it, with nonstandard phonology, in his immediately following utterance. She models the standard phonological form of the German item implicitly in her succeeding question directed to the content. (12) Serkan (4;08) German: playing with toy helicopter and airplane Serkan: und dann kommt das [*] Helikopter@s und die [*] will die [*] Flugzeug kaputt machen. da. und der [*] geht nicht kaputt und da geht [*] Hubsrauber [*] Hubsrauber [*] geht kaputt. Jutta: aha, warum denn? warum geht denn der Hubschrauber kaputt?

and then the helicopter@s comes and wants to destroy the plane. there. and it doesn’t break and there the helicopter helicopter breaks aha, why then? why does the helicopter break?

Ilknur, on the other hand, is still close to monolingual in her production at the outset of our study, though she already comprehends enough German and has enough pragmatic communicative competence to enable her to participate in interactions in German, as in (13), an example of the first recording with Ilknur: (13) Ilknur (3;10) German: picture of train station, interviewer points to clock: Heike: Ilknur:

weißte, was das hier ist? hae? do you know what this is here? huh? saat@s clock@s

 Carol W. Pfaff

Heike:

ja? weißt du auch auf deutsch? Ilknur: saat@s. Heike: das verstehe ich nicht. ich spreche kein Türkisch Martina: guck mal, das ist das selbe wie was die Heike da am Arm hat. Heike: weißt du, wie das auf deutsch heißt? nee? das ist ‘ne Uhr, ne? Martina: das kennste, ne?

Yeah? Do you know it in German too? clock@s I don’t understand. I don’t speak Turkish. look, that’s the same as what Heike has there on her arm. do you know what that’s called in German? no? that’s a watch, right? you know that, right?

Here, though Ilknur produces no German at all, her utterance shows that she comprehends the intention of the question and has the necessary extralinguistic knowledge. The interviewers respond with yes/no questions, explicitly inquiring about her knowledge of German, referring to their lack of knowledge of Turkish. They provide the German translation equivalent and offer another ostensive exemplification of the term. By 4;04, Ilknur has acquired the routine of learning German lexical items, as she shown in (14) in her repetition of Eier ‘eggs’ the translation equivalent of yumurta ‘egg(s)’. (14) Ilknur (4;04) German: picture book, market scene Ilknur: Heike: Ilknur: Heike: Ilknur:

und die yumurta@s wie heissen die? yumurta@s. Eier sind das auf Deutsch. Eier.

and the egg(s)@s what are they called? egg(s)@s ‘‘eggs’’ they are in German eggs

It is interesting to note that, while this translation is semantically and pragmatically equivalent, it is not entirely syntactically equivalent. The Turkish item yumurta is non-specific and unmarked for number, while the German form is the marked plural of Ei ‘egg’. It is impossible to tell exactly how Ilknur interprets the offered equivalent here, but it is possible that such input might lead her (or other Turkish children) to form a mental representation of German (or of this lexical item) in which non-specifics are unmarked for number and that thus conclude the form Eier is a singular form as well. Indeed, there is some evidence from an earlier study of Turkish and Greeks children’s development of German second language (Pfaff 1984: 277) that this may be the case. In talking about a picture of a single egg, over half the

Co-constructed narratives by Turkish children in Germany 

Turkish children used the form Eier in a singular sense, where none of the Greek children did so. In a very few cases, the German interlocutor not only accepted a Turkish lexical item in German but even took it up. In our recordings with the two children discussed here, this happened only in the early recordings with Ilknur at 4;03 and 4;05 with the kin term abla ‘older sister’. In (15) and (16): (15) Ilknur (4;03) German: unpacking dolls from bag Heike: guck mal, hier ist noch mehr drin. Ilknur: ich den abla@s [*] Heike: wer ist das? Ilkbur: abla@s . Heike: ach so, von wem ist die abla@s?

look, there’s more in here I [took] the older sister@s who’s that? older sister@s oh, whose older sister@s is that?

(16) Ilknur (4;05) German: playing house with toys Ilknur: die Frühstück kommt. die abla@s. Heike: ich dachte, das wär die abla@s.

breakfast is coming the older sister@s I thought that [doll]was the older sister@s

In (15), Heike clearly understands that abla expresses a relationship which does not have a single word lexical equivalent in German and uses the Turkish term in her question about the doll’s family. In (16), Ilknur uses abla to refer to two different girl dolls she is using to playing house and Heike is trying to clarify the reference. In any case, at least for the interim, the German adult interlocutor has adopted a cultural ‘‘loanword’’ into her German, an instance of the process change in the superstrate language as a result of the contact situation. It is likely that such lexical adoptions are still quite limited, possibly to German caretakers and teachers of Turkish children and possibly also to their playmates.

Interlocutors’ reactions to ‘‘errors’’ We now turn to responses to nonstandard forms and pragmatically inappropriate utterances. As with mixing, it is beyond the scope of the present chapter to deal with all types of ‘‘errors’’4 found in the two children’s utterances. Here I present and discuss a selection of some of the most typical nonstandard morphosyntactic forms in Turkish and German, insofar as possible presented

 Carol W. Pfaff

on the basis of examples which are parallel for the two children. It should be noted, however, that the selection here does not represent the proportion of error types or their frequencies in the speech of the two children, nor does it address the issue of longitudinal development, which will be the subject of a separate paper. Turkish is well known as a language with very regular and transparent agglutinative morphology. There is, however, some morphophonemic variation and a few irregular nominal and verbal forms. Derivational morphology, widely used by young monolingual children and also by Ilknur, is much less frequent in Serkan’s utterances. In general, Ilknur’s Turkish is much more like that of monolingual Turkish children than Serkan’s, which is quite far advanced toward attrition, though he is far from being a ‘‘semispeaker’’. Their ‘‘errors’’ typically are different in nature. Where she overgeneralizes and overmarks forms, he tends to substitute one marker for another or leave out markers altogether, resulting in structures which sometimes are parallel to German. German is equally well known as a language with opaque fusional morphology and many irregularities in verb stem and plural noun forms. In addition, substitutions of auxiliary verb forms, prepositions and subordinating conjunctions are frequent ‘‘errors’’. Further, the assignment of nouns to grammatical gender classes and marking of case/number/gender on articles and adjectives poses problems even for monolinguals and is notoriously difficult for second language learners. Using the ‘‘wrong’’ article is a stereotypical feature of interlanguage, often commented on by native and nonnative speakers alike. Again, as for Turkish, the two children’s varieties of German differ strikingly. Ilknur tends to omit function words and inflections while Serkan substitutes prepositions and conjunctions. Both children persistently use nonstandard marking not only of grammatical gender items, but also for natural gender, one of the characteristics of Turkish interlanguage discussed in Pfaff (1984). As we will see, the children’s ‘‘errors’’ of these types infrequently become the subject of explicit comment by the interlocutors in either Turkish or German conversations. Instead they are either passed over entirely or implicitly corrected in recasts and expansions of the children’s utterances which advance the narrative flow. At the end of the section, we examine a few pragmatically inappropriate or lexically infelicitous utterances which our ‘‘interviewers’’ are more inclined to ignore than are members of the children’s families.

Co-constructed narratives by Turkish children in Germany 

Responses to ‘‘errors’’ in Turkish Morphophonemic alternation: k~g˘ The first type of ‘‘error’’ to be examined is morphophonemic. Turkish final [k] alternates with ‘‘soft g’’, written g˘, intervocalically, usually realized phonetically as [y] or [γ] or simply as lengthening of the preceding vowel. In (17) and (18) the children suffix case markers beginning with vowels to stems ending in k, but do not make the consonant alternation: Ilknur produces erkeke rather than erkeg˘e ‘boy’-dative and Serkan says ineki rather than ineg˘i ‘cow’-accusative. In (17), the interview digresses from conversation about going to school to mention that she recently saw Ilknur’s cousins with their new baby brother or sister, she isn’t sure about the baby’s gender and asks its name. Ilknur says, ‘‘Mine’’, a girl’s name, triggering talk about the fact that she looks like a boy, (erkek), using the lexical item benziyor ‘resemble’, which takes dative, marked with a vowel suffix. Ilknur’s form is in the right case, but the consonant change is not made, she produces erkeke rather than erkeg˘e here although she usually produces standard forms of case marked nouns, including those with g˘. Here the interviewer simply accepts the form and goes on with the conversation. (17) Ilknur (6;06) Turkish: talking about baby boy or girl Filiz:

çok tatlı bi ¸sey. neydi onun adı? Ilknur: Mine. Filiz: Mine mi? o kız mı? erkek suratlı bi ¸sey. Ilknur: erkeke [*] benziyo Filiz: mhm çok tatlı ama.

a very sweet thing what was his/her name? Mine Mine? is it a girl? a boy-faced thing (s/he) looks like a boy [*] mhm but very sweet

A similar instance is (18) from Serkan, who unlike Ilknur, produces nonstandard case marking more often, though still infrequently, as discussed in Pfaff and Savas¸ (1988). Here the interviewer supplies the standard form in her recast and expansion. (18) Serkan (5;06) Turkish: picture book: men trying to get a cow outside Filiz: n’apııyolar? Serkan: ineki [*]?

what are (they) doing? the cow [acc]?

 Carol W. Pfaff

Filiz:

ineg˘i çıkartıyolar [/] (They) are taking out the cow [/] çıkartmak istiyolar içinden (they) want to take (it) from inside.

Possessives Turkish marks the possessor of an NP not only on the (optional) explicit noun or pronoun denoting the possessor but, unlike German (or English) also as a suffix on the possessed item. As noted in Pfaff and Savas¸ (1988) and Pfaff (1993), the KITA study children Ilknur and Serkan both produce nonstandard marking of possessor on the possessed item, but their errors are characteristically different. In (19) Ilknur overmarks the possessed noun kol ‘arm’, using kolusunu rather than standard kolunu, for ‘(the doll’s) arm’. Such forms have also been observed in the speech of monolingual preschoolers by Ekmekçi (1986), as indicated with our notation ‘‘@c’’, child language form, in the transcript. The interviewer lets it go at first, though she asks a question which might have provided an opportunity for Ilknur to self-correct. After Ilknur repeats the nonstandard form kolusunu in serveral turns, Filiz eventually supplies the standard form kolunu in her scaffolding question. After this intervention, Ilknur responds non-verbally, which may have something to do with the recasting. This kind of reaction on the part of the children in the past may have something to do with the interlocutors reluctance to provide ‘‘corrections’’ of forms which may be stigmatized. (19) Ilknur (4;06b) Turkish: picture book: children fighting over doll break off its arm Ilknur: sonra bebeg˘en [*] then he wants to break off the kolusunu@c koparmak istiyo doll’s arm. Filiz: nesini koparmak istiyo? (the doll’s) what do they want to break off? Ilknur: bebeg˘in kolusunu@c the doll’s arm sonra ordan arkadas¸ bag˘ı yo then (a) friend calls from there Filiz: ne diyo arkadas¸ları? what do her friends say? Ilknur: ’’koparma’’ diyo ‘‘don’t break (it) off (they) say Filiz: haa koparma diyo aah don’t break (it) off (they) say. Ilknur: sonra da ediyolar. and then they do kopardı kolusunu@c (he) broke off its arm Filiz: kim kopardı kolunu? who broke off its arm? Ilknur: [points to picture] [points to picture]

Serkan, rather than overmarking, sometimes uses unmarked forms. In (20), Serkan uses the nonstandard unmarked para rather than standard first


person singular form param, ‘my money’. One of the interviewers responds with a parallel nonstandard unmarked second person singular form in her follow-up question, using para rather than standard paran ‘your money’. The other interviewer provides a standard third person singular form, parası, ‘his/her money’. (20) Serkan (4;10) Turkish conversation about how he and his mother spend money Serkan: o zaman çok benim para [*] var o zaman. Tülay: o zaman senin çok para [*] var? Filiz: annenin parası olunca mı alacak?

then I had a lot of money then then you had a lot of money? when your mother has money will (she) buy (it)?

‘Profession/occupation’ derivation Both children talk about people who engage in some characteristic profession or occupation, which can be formed in Turkish by the productive derivational suffix -CI 5 unless a lexical item with that meaning already exists, thus blocking the derivation. In (21), after Ilknur has not produced a lexical item denoting the occupation ‘barber’, the interviewer introduces the standard form berber ‘barber’. Ilknur then produces the nonstandard, blocked derived form berberci which the interviewer first repeats, but then continues using the standard form. (21) Ilknur (4;05) Turkish: picture book of street scene with barbershop, conversation Filiz: Ilknur: Filiz: … Filiz:

kim kesiyo saçlarını senin? ähm Mürvet’in babası Mürvet’in babası

who cuts your hair? um Mürvet’s father Mürvet’s father

saç kesenlere ne denir?

what are the people who cut hair called? do you know Ilknur? no [sound] you don’t know? have you ever heard ‘‘barber’’? hm? ‘‘barber’’ [with derivational suffix] ‘‘barber’’ [with derivational suffix] is Mürvet’s father a barber? mhm.

biliyo musun Ilknur? Ilknur: çık [sound] Filiz: bilmiyo musun? ‘‘berber’’ duydun mu hiç? hm? Ilknur: ‘‘berberci’’ [*]. Filiz: ‘‘berberci’’ [*]. Mürvetin babası berber mi? Ilknur: mhm.



 Carol W. Pfaff

Serkan, on the other hand, does not use the derivational suffix in (22), but hesitantly produces a nonce-compound instead, itfaiye adamlar ‘fire brigade men’ (which should have had a final vowel adamları). However there is an existing word derived nominal form itfaiyeci ‘fireman’, which the interviewer supplies in her recast confirmation. (22) Serkan (6;04) Turkish: picture book, firemen Serkan ähm itfaiye [/] ähm itfaiye um fire brigade [/] um fire brigade adamlar [*]. men. Tülay: mhm itfaiyeciler bunlar da. mhm and these are firemen.

Adjectives for language names Turkish distinguishes between adjectives representing countries and the names of languages spoken there which are formed with the derivational suffix -CE. These bilingual children sometimes omit the derivational suffix for the language names, possibly transferred from German which uses same form for both. Serkan produces such forms more frequently than Ilknur does. In (23) the nonstandard adjective form occurs twice. After Serkan’s first use of Alman ‘German’ rather than Almanca ‘German [language]’ Filiz provides an explicit correction but the second time, she repeats the incorrect form, leaving it to Tülay to correct it, after which she too returns to the standard form. (23) Serkan (6;07) Turkish: conversation, language use with his German teacher and with an American man he apparently knows Filiz:

e Alman ög˘retmenle ne konus¸uyorsunuz?

Serkan: Alman [*]. Filiz:

Almanca. ‘‘Alman’’ deg˘il ‘‘Almanca’’ konus¸uyorsunuz deg˘il mi?

Filiz:

[intervening turns . . .] nece konus¸uyorsun sen J. ile? Almanca mı, Türkçe mi, Ingilizce mi? ne?

what do you speak with your German teacher German [without language suffix] German [with language suffix] not ‘‘German’’ [no suffix] you speak ‘‘German’’ [with suffix], don’t you? [intervening turns . . .] what [language] do you speak with J? German?, Turkish? English? what?

Co-constructed narratives by Turkish children in Germany 

Serkan: hep Alman [*]. Filiz:

hep Alman.

Tülay: Filiz:

Almanca. o zaman sen çok güzel Almanca konus¸uyosun.

always German [without language suffix] always German [without language suffix] German [with language suffix] then you speak German very well

Ilknur also produces the adjectival form rather than the derived language name, though this is not as characteristic for her. In (24) note that her usage is standard Türkçe for the ‘Turkish [language]’ but nonstandard Alman rather than Almanca for the German language. The interviewer provides a model of the correct form with her question for additional information about language use but does not make an explicit point of the form of the language name. Ilknur, who has previously said that she can use Turkish in class most of the time because there is a Turkish teacher and (all?) the children are Turkish, explains that they have learned that they have to use German when they ask to go to the toilet. Even if her statement is exaggerated, it is indicative of how limited are the domains in which German is normally used by her at this point in her life. (24) Ilknur (6;06) Turkish: conversation about language use in classroom Ilknur: bazıleri [*] Alman [*] konus¸uyom, bazıleri [*] Türkçe. Filiz: Almanca ne zaman konus¸uyorsun? Ilknur: tuvalete gittig˘imiz zaman Almanca konus¸mamız lazım.

sometimes I speak German, sometimes Turkish when do you speak German? when we go to the toilet we have to speak German

Responses to ‘‘errors’’ in German We begin our consideration of responses to ‘‘errors’’ in German with an instance which is parallel to the Turkish examples in the preceding section, the realization of language names and adjectives. This passage from Serkan is cited at length because it illustrates not only the ‘‘error’’ but also his and the interviewer’s expressions of their attitudes toward bilingual language learning, touching on both sides of the problem: difficulties of acquisition of German L2 and loss of Turkish L1.

 Carol W. Pfaff

Language name and adjective In German, in contrast to Turkish, the language name is a nominalized form of the adjective related to the country or ethnic group which is distinguished orthographically by being capitalized while the adjective is lower case and may receive further case/number/gender inflections. The nouns denoting person of that nationality, however, have a different form, Türke ‘male Turk’, Türkin ‘female Turk’. In (25), an exchange which starts with Serkan asking why the German interviewers keep coming to the Kita, do they want to learn Turkish? They get on to the subject of language proficiency and learning. Serkan uses the nominal form Türk, as the name of the language. (25) Serkan (5;07a) German: conversation about language proficiency Heike:

es können ja nicht alle so gut Türkisch und Deutsch wie du, die Kinder. Serkan: nee, die können das ganz schlecht machen. Heike: ja, manche können das nur ganz schlecht, weil die zu Hause immer nur Türkisch oder nur Deutsch sprechen Serkan: mhm. Die müssen auch mal Deutsch lernen,Türk [*] und Deutsch. Heike:

ja, ja. Die müssen beides lernen die deutschen Kinder auch n bißchen Türkisch könnten.

Serkan: Das da brauch ich jetzt.

not all the children can speak Turkish and German as well as you. no they can do it very poorly. yes, some can only very poorly because at home they speak only Turkish or only German mhm They have to learn German sometime, Turkish and German. yes, yes They have to learn both and it would be nice if the German children could (speak) a little Turkish too I need that now.

Grammatical gender As noted earlier, the choice of the correct article forms in German is extremely difficult for learners and their nonstandard use is part of the stereotypical interlanguage. In our study, we noted gradual improvement — from null to overt articles, appropriately marking definiteness and case, but with gender marking remaining very problematic (Pfaff 1992, 1994). Despite the fact that native speakers appear to be very conscious of errors involving article forms,


the interlocutors here do not explicitly correct them. In (26) the interviewer models the standard form in her following question, but in (27) the nonstandard form is simply accepted. (26) Ilknur (4;03) German: playing with toy animals Ilknur: Heike:

die Pferd. wo ist das Pferd?

the horse where is the horse?

(27) Ilknur (4;05) German: picture book Ilknur: die Hund geht Haus Martina: mhm

the [fem] dog goes house mhm

In (28) Serkan’s use of feminine form of determiner with a neuter noun, keine Auto rather than the standard kein Auto ‘the car’ is not only accepted, but praised because he has correctly identified what she took out of the bag while he wasn’t looking. Her explicit feedback which refers to the real world context of the game, to the pragmatic rather than to the grammatical form. (28) Serkan (4;05) German: memory game Martina: was hab ich geklaut? Serkan: ach, keine [*] Auto. Martina: richtig!

what did I steal? oh, no car right!

The above examples are all from early recording sessions when we were trying not to make the children uncomfortable in the German conversations. However, the behavior is typical: nonstandard (or null) article forms were mostly accepted without comment, sometimes standard forms were modeled in recasts or questions but there were never any explicit corrections of case/number/gender marking for grammatical gender items in later sessions either. In (29), the interviewers do not correct many nonstandard forms of the auxiliary, of the past participle, or of zero article forms, however in their questions, they do provide the standard case/gender forms mit dem Wolf, das Kind which Ilknur has produced as die Wolf and die Kind. Their modelling certainly does not have any immediate effect, however, as Ilknur continues using her form, die Wolf so it is clear that she has not (yet) internalized a grammar of German which would allow her to infer from the form dem that the lexical item wolf could not be feminine. If she will ever learn this, perhaps after exposure to formal teaching of the paradigm in school, remains to be seen.



 Carol W. Pfaff

(29) Ilknur (5;08) German: Little Red Ridinghood Ilknur:

dann hab [*] die [*] Wolf deht@c [*] da hab[*] die Gross [*]+. Martina: was war mit dem Wolf? Ilknur: die [*] Wolf habt [*] in [*] Haus erein [*]. dann hab [*] die [*] ihre Grossmutter ess [*], dann war die [*] Kind böse, dann hab [*] die / auch [*] ähm, das [*] hab [*] die aufdefress [*] auch [*]. Jutta: wieso war denn das Kind böse?

then the wolf went there (it) had the grand + What was that with the wolf? The wolf had gone into the house then she [wolf] ate her grandmother. Then the child was angry then she [wolf?] also um it [wolf] ate her up too. why was the child angry then?

While Ilknur’s use of die Kind to refer to the girl, Red Ridinghood, looks like it may reflect her acquisition of natural gender marking, the example (30) discussed in the next section makes it clear that she does not have this rule either. Nor does Serkan, whose German is otherwise so much more developed, as shown in (31).

Natural gender As we have seen in the above examples, grammatical gender errors often pass without comment or modeling of the standard forms. Nor are natural gender errors always corrected. As we see in (30), Ilknur’s use of the feminine definite article with a clearly masculine referent, die Mann rather than the standard der Mann, is simply accepted. (30) Ilknur (5;10) German: Snow White Ilknur:

und die [*] Mann kommt an, tötet ein [*] Vogel.

Martina: mhm.

and the man comes kills a bird. mhm.

In (31), when Serkan is telling the story of Snow White to another boy, Satılmıs¸, the interviewer does not interrupt the narrative flow, ignoring all the nonstandard forms, natural and grammatical gender, as well as Serkan’s paraphrase tot geht ‘go dead’ rather than lexical form stirbt ‘die’. She comes in only at the end of the exchange between the children to clarify content of Serkan’s utterance with the lexical item knabbert ‘nibble’ which the other child,

Co-constructed narratives by Turkish children in Germany 

who does not know as much German as Serkan, may not have understood, providing the object am Apfel: (31) Serkan (5;11) German: Snow White, picture book, telling story to another child Serkan:

dann sagt a [=er?] [*], der [*] then he [witch] said he soll ein [*] Giftapfel ihn [*] [queen] should give him geben, daß er [*] 〈tot geht〉@c [Snow White] a poison apple, so that he [Sn Wh] goes dead. Serkan: und dann gibt ihn [*] 0art [*] and then witch gives him [Sn Hexe eine [*], und dann Wh] one and he [Sn Wh] nibknabbert er [*] einmal, gleich bles once, he nibbles right knabbert er [*] away. Satılmıs¸: knabbert er? he [Sn Wh] nibbles? Serkan: ja, guck! yes, look! Jutta: am Apfel. at the apple

Lexical items The interviewers seem to pay more attention to lexical items rather than to grammar. In (32), where Ilknur incorrectly repeats lexical item Pantoffeln ‘slippers’ as Kartoffel ‘potato’ a similar sounding item which she already knows, the interviewer repeats the word in its standard form, but does not correct or model a standard version of the ungrammatical syntax: (32) Ilknur (5;06) German Heike:

Ilknur: Heike:

und der Hund bringt ihm die and the dog brings him the Pantoffeln und die Katze auch. slippers and the cat too. ... dann ähm die Kartoffel [*] bringen. [laughs] Pantoffeln, ja.

... then um (they?) bring the potato. slippers, yes.

Similarly, the interviewers do not respond explicitly to Serkan’s nonstandard grammatical forms in (33), but instead focus on the vocabulary item, pointing out the difference between the term Stuhl ‘chair’ and Hocker ‘stool’. This exchange explores various approaches to definition. Heike rejects his use of ‘chair’ and when he insists, Martina characterizes the stool in terms of its physical attributes — it doesn’t have a back — while Serkan refers to its potential function — to sit on — as if to prove thereby that it is legitimate to call it a ‘‘chair’’. She picks up on his functional characterization, proposing

 Carol W. Pfaff

that one can also stand on a stool when getting something down from a high place. Throughout this passage, it is clear that Serkan is focused on metalinguistic issues but that these concern definitions of the lexical items in question, not the form of his utterances. (33) Serkan (5;09) German: playing with toy furniture, furnishing a house Serkan:

Martina: Serkan: Martina: Heike: Serkan: Martina: Serkan: Martina: Serkan:

Martina:

Serkan: Martina: Serkan:

Martina:

aber weißt du Stuhle [/] was ein Stuhl, wo das is [*] Stuhl is? nee natürlich da [pointing]. ach so. das is aber kein Stuhl. wohl. n Stuhl ohne Lehne, wa? n Hocker. kein Hocker. doch, is n Hocker. wofür braucht man den Hocker denn? für [*] Sitzen? da kann man drauf sitzen. man kann aber auch drauf stehen, wenn man irgendwo ganz weit oben irgendwas holen muss. ‘‘Hocker’’ wofür kann man den noch brauchen? für was, was ganz hoch is wann [*] die was runterholen wollen, da / dann brauchen sie noch des ja

but you know chair [/]what a chair, where that is, chair is? no naturally, there [pointing]. oh But that isn’t a chair Yes [it is] a chair without a back, right? a stool no stool Yes, (it) is a stool. What do you use the stool for then? for sitting? One can sit on it but one can also stand on it when one has to get something down from way up high ‘‘stool’’ What else can one use it for? For something that is very high when they want to get something down, then they need that also yes

Subordination As seen in Serkan’s last turn the example (33) above, subordination is a problem, not so much on account of the verb placement, but the choice of subordinating conjunctions. His use of wann for wenn, which is common in

Co-constructed narratives by Turkish children in Germany 

the speech of other children in his linguistic environment, is persistent in Serkan’s speech. As in (33), interlocutors often simply accept this form. For Ilknur, in addition to the choice of a subordinating conjunction, the syntactic formulation of the subordinate structure seem to pose a problem. In (34) after a very long passage in which the interlocutor did not intervene, she struggles with the purpose clause, choosing weil, the more frequent reason clause conjunction rather than damit, and also seems to be struggling with the syntactic formulation, at which point the interviewer intervenes to reformulate with a standard structure and lexical choice. (34) Ilknur (5;08) German: Little Red Ridinghood, told from memory Ilknur: ‘‘warum is dein Auge so gross?’’ ‘‘why is your eye so big?’’ ‘‘weil dich [/] weil dich [/] dich ‘‘because (I) [/] see you better+ besser sehen+ Heike: ‘‘damit ich dich besser sehen ‘‘so I can see you better’’ kann!’’

Inappropriate or misleading content In the preceding sections we have examined instances in Turkish and German in which the child’s utterance is nonstandard in form. In this section, I want to look briefly at one other type of intervention and correction practiced by interlocutors: instances in which they intervene to ‘‘correct’’ imprecise, erroneous or misleading content rather than focusing on form. These are the kind of interventions familiar from the literature on interactions between monolingual children and their caretakers. In fact, in the following instance, it is members of the children’s families who are the interlocutors who make interventions of this type. In (35), a German recording session which takes place at Ilknur’s home and at which her sister is also present, they are (again) looking at the book, Selim and Susanne which begins with a Turkish boy getting into a fight with German children in his neighborhood. Ilknur already knows this story and proceeds with the narrative. The interviewer ignores her formal errors but intervenes to clarify the reference to the Turkish boy. More strikingly, she accepts the formally correct but pragmatically implausible statement in the direct quote Ilknur offers to account for why the children don’t want to play with him: ‘‘Er versteht nicht was er sagt’’ ‘he doesn’t understand what he is saying’. The older sister, however, intervenes to correct Ilknur’s misleading

 Carol W. Pfaff

quote, explicitly rejecting Ilknur’s form and proposing ‘‘was wir sagten’’ ‘what we said’ instead, a form which is also not exactly appropriate within the narrative context of the direct quote; the form should be present tense wir sagen ‘we say’. The interviewer laughs and Ilknur goes on with the narrative. (35) Ilknur (6;06) German: at home, looking at picture book, sister is also present Martina: Selim und Susanne. Ilknur: hier eines Tages kommt die xx und sagt ‘‘nimm dein Puppe nicht, dann geht sie kaputt.’’ ‘‘na und?’’ und dann der xx steht draussen und dann + . Martina: der türkische Junge, ne? Ilknur: und dann wollen sie Spaziergang x. dann haben sie + ‘‘guck mal ihn, der versteht nich, was er sagt’’ Martina: mhm Sister: nicht ‘‘er sagt’’ Martina: sondern? Sister: ‘‘was wir sagten’’ Martina: [laughs]. Ilknur: und dann haben sie [/] dann hab [*] sie die Puppen ihre Arm kaputt gemacht,

Selim und Susanne. Here one day comes the xx and says ‘‘don’t take your doll, it’ll get broken’’ ’’So what?’’ And then the xx stands outside and then + the Turkish boy, right? and then they want to x (a) walk. then they have ‘‘look at him, he doesn’t understand what he says’’ mhm. Not ‘‘he says’’. Rather? ‘‘what we said’’ [laughs]. and then they / then they broke the doll’s arm

In (36) the interview takes place at Serkan’s home and his mother is present in addition to the interviewer. Although Serkan’s utterance is formally nonstandard, with a nonstandard form of the verb vurarsa rather than vurursa ‘if (s/he) hits’ which takes dative bana ‘me’, ona ‘him/her’ rather than his accusative beni ‘me’ onu ‘him/her’, his mother does not respond to these formal characteristics, but focuses on the content, making the point that violence on either part isn’t necessary. Note that she uses entirely different forms of the verb, producing a structure which is much more complex than any he would


use, and that she does not model the case inflection since the objects are not realized in her utterance. (36) Serkan (6;07) Turkish Serkan: anne J. beni [*] bi vurarsa [*], ben onu [*] öyle vurcam paat@o. Filiz: eyvaah! Mother: ama J. vurmadıg˘ına göre sen de vurmazsın.

mother, if J. hits me one, I’ll hit him/her like that paat@o! oh! But since J. does not hit (you), you needn’t hit (him/her) either

In these two final examples, we see content rather than form is even more the issue for some of the children’s regular conversation partners than it is for our project interviewers.

Conclusions and discussion When evaluating the present findings, it should be kept in mind that the interactions we considered in are of a special type, different in some respects from the childen’s conversations at home, in the neighborhood and at the Kita or school. One major difference between these interactions and those of daily life is the very great attention paid to maintaining the conversation with the individual child, leading the interviewers to employ many global scaffolding devices which support the child’s contributions and make it possible for him or her to continue. A second important difference is the linguistic mode of the interaction. While more or less free in topic and transition between conversational and narrative frames, the researchers who were the primary interlocutors in all of the recordings examined here set the linguistic frame in the monolingual mode. While the monolingual mode is probably typical of the children’s everyday interactions in German in the Kita, in school and in other contexts, it is probably less typical of their everyday interactions in Turkish, at least when conversing with other Turkish/German bilinguals. As has been seen our Turkish interviews also tended naturally toward the bilingual mode at times, at least on the part of the children. Nonetheless, while not completely typical of many of the children’s everyday interactions, our interviews are not totally different either. Particularly with respect to the interlocutors’ responses to language mixing and nonstandard forms and their analysis may suggest ways of approaching of interaction and input which are more commonly encountered.



 Carol W. Pfaff

Our examination of the co-constructed narratives and conversations between adults and bilingual children allow us insight into two aspects of development of language in this migrant community setting: the development of language proficiency in the individual children and the development of ethnic varieties in the bilingual community. As already suggested, the interlocutors’ concern with content rather than form of the children’s utterances is striking. This is particularly obvious for the few documented instances of interventions of the children’s family members, who explicitly comment on the appropriateness of the children’s utterances or recast them in a more acceptable way. Concern with content is also obvious in our interviewers’ responses, as evidenced by the many instances in which the children mixed language or nonstandard forms are simply accepted or even repeated in the process of scaffolding the narrative or conversation, but they very frequently react to the formal aspects of the children’s utterances as well. With respect to their reactions to formal aspects, there appears to be a hierarchy of features which come to the interlocutors’ attention and which receive either explicit comment or modeling of standard forms in recasts and expansions. The use of vocabulary from the other language is the most likely feature to elicit explicit comment from the interlocutors, though it is noteworthy that, in Turkish interactions, it is frequently the child who initiates discussion of the Turkish translation equivalent of a German word, as in Ilknur’s does for the item Schwester ‘nurse’ or Serkan does for pommes frites ‘French fries’. These are clear cases of ‘‘cultural loans’’ which have been incorporated into the Turkish varieties of these children and of the wider population of Turkish/German bilinguals in Berlin, to which our Turkish interviewers belong. Nonetheless, the children here demonstrate their awareness that these lexical items belong to German. For Turkish items used in German, it is usually the German interlocutors who initiate the discussion of translation equivalents; this is the most frequent trigger of explicit metalinguistic comment on their parts, or for modeling of the German form. That this strategy is successful, is clear from the fact that children’s insertion of Turkish in German diminishes rapidly as they adopt a monolingual mode in their interactions with native speakers of German, whom they come to realize expect it and, further, do not usually understand Turkish. We found only one lexical item, the kinship term, ‘older sister’, adopted by a German interviewer, and this was only observed in the earlier conversations when it was used as the name of a toy doll.

Co-constructed narratives by Turkish children in Germany 

With respect to reactions to nonstandard usage or gaps in the language of the ongoing interaction, we again find that it is the lexical aspects that receive explicit comment. One particularly clear example is in the discussion with Serkan of the definition and applicability of the words for ‘chair’ and ‘stool’, which approaches a lexicographical analysis, in which he participates as vigorously as the adults. Apparent lexical gaps, such as ‘nurse’ or ‘barber’ for Ilknur also are the subject of explicit discussion; there appears to be less tendency to discuss lexical gaps when the interlocutors expect the child will not already know the item in his/her weaker language, Turkish for Serkan, German for Ilknur. Morphological aspects, particularly the more lexical derivational morphology are sometimes explicitly commented on, as with the use of adjectives for language names in Turkish, but these are more typically modeled in succeeding turns, though the nonstandard form was also sometimes repeated by the interviewer. Overgeneralized regular forms rather than irregular forms of verbs in German or possessed nouns in Turkish do not receive explicit comment, but they are sometimes modeled in succeeding recasts or expansions of the children’s utterances. Quite often, however, they are also simply passed over and apparently accepted. In contrast, syntactic aspects such as missing elements or nonstandard word order never elicited explicit comment. Nor did nonstandard marking of gender in German, even for natural gender items. Errors of these types were, however, implicitly corrected with standard models in expansions and recasts, but such modeling is inconsistent and was more likely to occur at places where the child has hesitated or explicitly indicated awareness of lack of the item or structure in question. This tendency is, of course, related to the interlocutors’ aim of facilitating rather than interrupting the flow of narrative or conversation. A kind of ‘‘fine tuning’’ of the interviewers’ responses can be observed by comparing how the same interviewers deal with the same phenomena differently for different children, apparently depending on their assessment of the degree of the child’s proficiency. For instance, the use of the adjective for the language name, Alman rather than Almanca in Turkish was explicitly corrected for Serkan but not Ilknur. Similarly, the morphophonemic alternation of k~g˘ was modeled for Serkan but not for Ilknur after the child’s production of a nonstandard form. Whether this sort of fine tuning can be observed to develop longitudinally as the children’s linguistic proficiency develops (or declines) over time remains to be addressed in another paper. We also observe characteristic differences in the type and extent of interlocutors’ interventions in Turkish and in German. These differences can be

 Carol W. Pfaff

traced back to the most fundamental circumstance of the bilingual children investigated here, the fact that they live in Germany, not Turkey and that they are about to enter (or have already begun to face) the essentially monolingual German school system. Thus the necessity for the children to acquire a (reasonably) standard variety of German to be able to benefit from school instruction is clearly acknowledged by everyone in the community. The fact that many do not achieve this level by the time they enter school is a matter of great concern, both for educators and for the children’s families. In contrast, though many adults (and children) find it regrettable, they see that the probable outcome is that the children will increasingly shift to German and that this will result in attrition of Turkish or acquisition of varieties of Turkish that show evidence of extensive contact with German. The extent of such language-related differences in the co-constructed narratives and conversations we have examined is most obvious in comparing the reaction of the interlocutors to the children’s use of lexical items from the other language. When the Turkish-dominant child Ilknur uses Turkish at first to fill gaps in her German, the interlocutors accept this, but generally respond with German translation equivalents. With the exception of the kinship terms which are accepted and, as noted above even sometimes used by the German interlocutors during the initial recording sessions, in other semantic fields, German lexical items are provided, in many cases with explicit mention of the equivalence, and frequently modeled in further turns, sometimes supplying additional encyclopedic information, as we saw with the response to ‘clock’. The child rather quickly follows suit and the number of Turkish words inserted in German declines sharply. In the Turkish recordings, however, the behavior of the adult interlocutors is notably different. German lexical items are generally accepted, and Turkish translation equivalents are rarely introduced. In many cases, the Turkish adult follows the child’s use of German lexical items. (sometimes giving up after several attempts to establish the currency of the Turkish item, as is the case for another KITA child discussed in Pfaff 1998). With respect to the grammatical structures, the difference is less clear. As we have noted, there is little explicit correction in either language and modeling of standard forms is inconsistent in both. Certainly such interlocutor behaviors as we have observed contribute to the establishment of ethnic varieties of the first and second languages in the migrant community here. This is not only so for the minority language, Turkish, for which relatively stable ‘‘contact lects’’ appear to be developing throughout Northwestern Europe, but also for the ‘‘host’’ country languages.


While these changes are natural and just what linguists should expect in this language contact situation, from the point of view of children’s opportunities as individuals in society, and for the welfare of society as a whole, it is obviously important for them to have command of standard language, at least enough for education to the level of which they are intellectually capable. One of my implicit goals in this chapter was to demonstrate that, despite their use of nonstandard forms, these children are cognitively very aware and competent in social and linguistic interaction. Although my focus here has been on ‘‘flaws’’ in children’s performance, the children’s language in these examples reveals many areas of proficiency as well as gaps. These include their effective use of global strategies to maintain interactions and achieve the goals of discourse. While greater grammatical proficiency permits the children greater flexibility in leading the conversational interaction in their dominant languages, we see that the children are also able to direct the conversations to their advantage in their weaker language. Further, we have observed their abilities to apply local linguistic strategies, coining lexical items and extending grammatical rules from both languages. Finally, in these interactions, the children demonstrate a strong metalinguistic awareness and interest in language per se and the ability to employ explicit vocabulary searches and hesitations at points of lexical and syntactic difficulty as strategies to elicit assistance in the form of input from the interlocutors. Since, as noted earlier, the type of interaction represented by our ‘‘interviews’’ makes up only a small proportion of the verbal interchanges the children participate in and observe, it is clear one should not expect any direct correlation between what went on in these interviews and the child’s development of linguistic and conversational proficiency. Nonetheless, while these conversations are untypical in the ways noted, I believe they are suggestive of what goes on more generally with other interlocutors with respect to reactions and responses to mixed or ‘‘nonstandard’’ child utterances. Further research into interactions with family members, peers, caretakers and teachers is essential to see what kind of models and feedback characterize the wider input to the children, which they can take in and elaborate creatively using their own innate and acquired linguistic resources.

Notes . The KITA study, investigating the language development of nursery, preschool, school age Turkish/German children was supported by grants from the Freie Universität and the



 Carol W. Pfaff

Deutsche Forschungsgemeinschaft (Pf-201/1–5). The VAK Kita (day-care center), that the children attended is located in the Kreuzberg district of Berlin, which is heavily Turkish. The sample of 33 children included 22 second generation children of migrants from Turkey, 7 German children and 4 children of mixed marriages. An overview of the setting, language policies and the major findings of this and other Berlin SBLA studies is presented in Kardam and Pfaff (1993); for discussion of the development of German second language, see Pfaff (1992, 1994) and for features of Turkish first language development attrition see Pfaff and Savas¸ (1988) and Pfaff (1993). Grammatical aspects of code mixing of various types are treated in Pfaff (1991, 1998) and Pfaff, to appear. The relationship between their proficiency and conversational strategies in both languages of the two children analyzed in the present chapter, including a fine-grained comparison of parallel narratives in both Turkish and German, briefly summarized in Section 3 of the present chapter, is discussed in Pfaff (1994). This work could not have been accomplished without the cooperation of the children, (given pseudonyms here), their parents and the Kita caretakers and staff. I am equally indebted to and grateful to project staff and students who carried out the interviews, made the transcriptions, coded and glossed and participated in the analysis of the material. These include Filiz Kardam, Tülay Savas¸, Ogün Çakarcan and Fügen S¸engün on the Turkish side and Heike MacKerron, Karin Schmidt, Martina Thiel and Jutta Voß and on the German (the ‘‘interviewers’’ appear in the examples here under their own names). I am grateful to all for their cooperation and support but take responsibility for the formulations and interpretations presented here myself. . These excerpts are modified from complete transcriptions in CHAT format, using standard orthography, sometimes representing common colloquial or dialect forms. Nonstandard child language forms are indicated with @c and onomatapoetic sounds with @o at the end of the word. Lexical items from the other language are indicated by @s; Turkish suffixes on German words are preceded by an apostrophe like proper nouns in standard Turkish orthography. Unintelligible syllables are indicated with one ‘‘x’’ per syllable. Nonstandard forms or syntactic structures are indicated by [*] after the relevant item(s). The glosses reproduce only a few of the nonstandard characteristics of the utterances; explanatory material appears in square brackets, omissions (whether grammatical or ungrammatical) appear in parentheses. Some repetitions, stuttering and false starts have been eliminated from the present version. Those that are included are marked [/]. . In fact, from his statements in the course of conversations with him through the years, it is clear that Serkan does feel that it is important for Turkish children and adults to learn German. He is critical of other Turkish children’s lack of proficiency and tells the Turkish interviewers that they need to know German, perhaps as a ploy to induce them to switch languages when they converse with him. He is rather skeptical about the prospect that Germans might also learn Turkish, but acknowledges that he needs to learn more. The topic of language attitudes of children and adults in the community will be taken up in a separate paper. . I use the term ‘‘error’’ here because, though researchers on language development and change would reject this characterization in favor of a term such as ‘‘nonstandard’’, ‘‘interlanguage’’ or ‘‘non-target-like’’, this is how most of the children’s interlocutors conceive of these forms. The point at issue here, whether and if so how, they react to the

Co-constructed narratives by Turkish children in Germany 

children’s use of these forms. It should be noted that ‘‘nonstandard’’ must be distinguished from ‘‘non-target-like’’ because, although the children are exposed to standard colloquial spoken Turkish and German, there are also many nonstandard dialectal and interlanguage forms in their input. This is particularly the case for children in this neighborhood which, as noted earlier has a very high proportion of speakers of rural dialects of Turkish and of varieties of Turkish in various stages of attrition. Similarly they are addressed not only by non-native speakers of German, but also by native speakers of nonstandard urban sociolects. . We have already seen that Ilknur uses this derivational marker in Eis’çiya ‘to the ice cream man’ in (9). Capital letters are customarily used as cover representations for vowel and consonant harmony forms: thus -CI stands for the morphophonemic variants ci, cı, cu, cü/çi, çı, çu, çü of the professional suffix here and similarly for the -CE stands for the variants ce, ca, çe, ça of the language name suffix discussed in the section about adjectives for language names.

References Bamberg, M. (1987). The acquisition of narratives: Learning to use language. Berlin: Mouton de Gruyter. Berman, R. and D. I. Slobin (1994). Relating events in narrative. A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Ekmekçi, Ö. (1986). The developmental errors in the preschool Turkish children’s speech. Proceedings of the Turkish linguistics conference. Istanbul: Bog˘aziçi University Publications. de Houwer, A. (1995). Bilingual language acquisition. In: P. Fletcher and B. MacWhinney (eds.), The handbook of child language. Oxford: Blackwell, 219–50. Kardam, F. and C. W. Pfaff (1993). Issues in educational policy and language development of bilingual children in Berlin. In: S. Kroon, D. Pagel and T. Vallen (eds.), Multiethnische Gesellschaft und Schule in Berlin. Münster/New York: Waxmann., 51–68. Lanza, E. (1997). Language mixing in infant bilingualism. A sociolinguistic perspective. Oxford: Oxford University Press. Pfaff, C. W. (1984). On input and residual L1 transfer effects in Turkish and Greek children’s German. In: R. Andersen (ed.), Second languages. Rowley, Mass.: Newbury House, 271–98. Pfaff, C. W. (1991). Mixing and linguistic convergence in migrant speech communities: linguistic constraints, social conditions and models of acquisition. Code-switching and language contact: constraints, conditions and models. Strassbourg: European Science Foundation, 120–53. Pfaff, C. W. (1992). The issue of grammaticalization in young bilinguals’ second language development. Studies in second language acquisition, 14, 273–96. Pfaff, C. W. (1993). Turkish language development in Germany. In: G. Extra and L. Verhoeven (eds.), Immigrant languages in Europe. Clevedon: Multilingual Matters, 119–46.

 Carol W. Pfaff

Pfaff, C. W. (1994). Early bilingual development of Turkish children in Berlin. In: G. Extra and L. Verhoeven (eds.), The cross-linguistic study of bilingual development. NorthHolland, Amsterdam/Oxford/New York/Tokyo: Koninklijke Nederlandse Akademie van Wetenschappen, 75–97. Pfaff, C. W. (1998). Changing patterns of language mixing in a bilingual child. In: G. Extra and L. Verhoeven (eds.), Bilingualism and migration. Berlin: Mouton de Gruyter, 97–121. Pfaff, C. W. and 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, 351–86. Snow, C. (1995). Issues in the study of input: Fine tuning, universality, individual and developmental differences, and necessary causes. In: P. Fletcher and B. MacWhinney (eds.), The handbook of child language. Oxford: Blackwell, 180–93.

Chapter 7

Influence of L1 Turkish on L2 French narratives Mehmet-Ali Akinci, Harriet Jisa and Sophie Kern

Two distinct, but interrelated levels of analysis have been addressed in research on narrative: cohesion and coherence. Following Hickmann (1995: 201) cohesion refers to the linguistic devices used in the expression of content, while coherence refers to the structure of narrative content. Story grammars, for example, propose representations of underlying narrative structure which it is argued form the cognitive foundations guiding the production and comprehension of narrative texts (cf. Mandler 1978; Mandler and Johnson 1977, 1980; Rumelhart 1975, 1977, 1980; Thorndyke 1977). Despite considerable divergence in the details of story grammars, there is a general consensus that certain elements are essential to a well-formed story: a setting, an initial problem, attempts at a solution to the problem and a resolution (Adam 1985; Labov and Waletsky 1967; Stein 1982; Stein and Trabasso 1981). There is little consensus, however, concerning how narrative structure and narrative cohesion are related. An important study by Thorndyke (1977) presented narrative texts to two groups of subjects. One group heard stories with canonical story structure and another group heard stories with jumbled story structure. As predicted by story grammars, recall was better for subjects who heard canonical stories. Garnham, Oakhill and Johnson-Laird (1982) however, found that recall for the jumbled stories could be influenced through the modification of cohesive structures. By restoring referential cohesion to the jumbled stories recall for jumbled stories improved. Some research has proposed that the development of cohesion and coherence are very closely related (Shapiro and Hudson 1991). Other research has argued that cohesion and coherence are not simply parallel developments, but that the development of story grammars is essential to the development of cohesive devices, such as connectives (French and Nelson 1985). In earlier work on monolingual French children, Jisa and Kern (1995)

 Mehmet-Ali Akinci, Harriet Jisa & Sophie Kern

concluded that the acquisition of narrative structure and the acquisition of syntactic competence are inextricably related. Narrative structure, in particular episodic structure, has been shown to have ramifications on the use of linguistic forms. For example, changes in children’s conceptions of episodic structures can be tracked through the observed changes in relative clause usage in narrative tasks. Relative structures are precocious and frequent in French monolingual children’s narrative texts. However, mature uses of relative structures in storytelling require chunking of a narrative text into episodes (Jisa and Kern 1998). Children’s use of anaphoric pronouns to maintain reference is also related to narrative structure: within episodes anaphoric pronouns are more precocious and more frequent than their use across episodes (Hickmann, Kail and Roland 1995; Jisa 2000). As Hickmann (1995) points out further research is needed to understand the relation between development of linguistic forms and narrative coherence. The vast majority of narrative research has ignored bilingual populations and yet bilinguals, and children on their ways to becoming bilingual, offer an invaluable source of data. In the study presented in this chapter we have attempted to separate narrative structure and the expression of narrative structure. Turkish–French bilingual children were asked to tell a story in Turkish (their home language) and subsequently in French. This allows us to compare their use of narrative structures in the two languages. The subjects of this study are essentially monolingual in Turkish up to the age of three, when they begin attending monolingual French nursery schools. By the age of ten, most of these children show French as their dominant productive language (Akinci 1999). There are three questions which motivate our study. The first question we will address is the following. 1. Are narrative texts produced in French and in Turkish by the bilingual children comparable, or are the texts in one language more “complete’’ than in the other? To examine the first question we will compare the use of macro-structure narrative components (Berman 1988; Berman and Slobin 1994: 46) in Turkish and French texts produced by the bilingual children. We will then compare the bilingual children’s French texts to those produced by monolingual French children in order to answer our second question. 2. Are the French texts produced by the bilingual French–Turkish children and by the monolingual children comparable at all three age ranges?

Influence of L1 Turkish on L2 French narratives

The second question will be addressed by comparing the macro-structure components found in the bilingual children’s texts with those found in monolingual French children’s texts. We will then turn to a closer examination of the linguistic structures used by the bilingual and monolingual children to encode the macro-structure components. Our third question is as follows. 3. Are the linguistic structures used by the French-Turkish bilingual children to encode the different components the same as those used by the French monolingual children in the French texts? To answer the third question we will turn to a more qualitative look at the linguistic forms used by the bilingual and monolingual children to encode narrative components in their stories.

Design of the study The subjects Narrative texts were collected from 5-, 7- and 10-year-old bilingual and monolingual children. Table 1 gives the number of subjects in each group and their mean ages. Four of the bilingual children were born in Turkey and came to France before the age of one year. The remaining subjects were born in France in Turkish-speaking families. The parents of these children were all born in Turkey. None of the mothers work. Sixty percent of the fathers are either unemployed or workers in the construction industry. One fourth of the mothers and 10 per cent of the fathers are illiterate. Sixty percent of the parents received a primary school education. Eleven percent of the mothers and 25 per cent of the fathers have some secondary education. In contrast to Table 1. Turkish–French bilingual and French monolingual subjects. French–Turkish bilinguals French Turkish

French monolinguals

Age Group N

5 13a

7 16

10 14

5 14

7 16

10 15

5 20

7 20

10 20

Mean Age Age Range

5;6 5;1– 5;11

7;6 7;0– 7;11

10;6 10;0– 10;11

5;6 5;1– 5;11

7;6 7;0– 7;11

10;6 10;0– 10;11

5;5 5;0– 5;11

7;5 7;1– 7;10

10;8 10;2– 11;3

a

One 5-year-old refused to tell the story in French



 Mehmet-Ali Akinci, Harriet Jisa & Sophie Kern

the Turkish–French bilinguals, the French monolinguals are from middle class families. The parents’ education level includes from some secondary education to some post-secondary.

Data collection A children’s picture book, Frog, where are you? (Mayer 1969) was used to collect the data. The frog story book consists of twenty-four pages of pictures with no text. The story relates the adventures of a boy and his dog in their search for a runaway frog. During the search, the boy and the dog meet up with a variety of different characters. Their encounters with these other characters yield a series of episodes which make up the story. Following the procedures outlined in Berman and Slobin (1994), narrative texts were collected from bilingual Turkish–French children and monolingual French children. The bilingual children’s Frog stories were collected in Turkish and subsequently in French. A Turkish bilingual researcher (Akinci) showed the children the book. Then, a second Turkish speaker (known to the child) was asked to listen to the child’s story. The children habitually speak Turkish to both adults. The majority of recordings were made in a Turkish cultural center. Some Turkish stories were collected in the children’s homes. Approximately one week later, the children were again shown the book by a researcher (Akinci) and asked to tell the story to a monolingual French speaker. The French monolingual children were recorded for the most part in their homes by a variety of native-speaking French assistants. The children told the story, either to another assistant, or in some cases, to their mother.

Data analysis Error analysis Before presenting the study of narrative components in the Frog story, we will briefly summarise the results of an error analysis of the French texts (Akinci 1999). The appendix lists the types and frequency of errors observed in the bilingual French stories. The error analysis was undertaken to show that the bilingual children master sentence level grammar rather quickly. The majority of the error types observed in the bilingual children are also observed in much younger French monolingual children (Clark 1985; Kern 1997). Some errors may possibly be explained by interference. Among these are the absence of a determiner, a copula and a subject, as well as problems in word order placement

Influence of L1 Turkish on L2 French narratives 

of the object. However, these “interference’’ errors are not frequent and tend to disappear after the age of five. The most frequent error type concerns gender, which is problematic for French second language speakers, regardless of language background. Gender errors, then, cannot be specifically attributed to Turkish. The rapid attainment of sentence level grammar among our Turkish immigrant children is consistent with results reported for other second language learning children (Cummins 1984, 1991; Snow and Hoefnagel-Hohle 1978).

Narrative components Four narrative components were coded in all of the texts, following the procedures given in Berman (1988) and Berman and Slobin (1994: 46). Definitions of the components, as well as examples, are given in Table 2. Texts are coded as containing Component 1 (onset of the plot) when there is explicit mention of the boy realising that the frog is gone. The realisation that the frog is missing triggers the search. To be coded as having Component 2 (unfolding of the plot), the texts must make explicit mention of the search at least three times. Explicit mention of Component 2 expresses the common goal (the search for the frog) which establishes coherence between the various episodes. Verbs, characterised as having a high degree of “control’’ on the part of the subject (Comrie 1981) such as chercher (‘to search for’) and appeler (‘to call for’) were counted as explicit mentions. Regarder (‘to look in or at’) or voir (‘to see’), however, were not considered as explicit. At the end of the story the boy finds a family of frogs and takes one small frog with him. Component 3 (resolution of the plot) is counted in texts where the narrator describes the frog as being the same or as substituting for the frog shown escaping at the beginning of the story. Component 4 (encapsulations refers to mentions of a summary of the search, either prospective or retrospective. Prospective encapsulations summarize the episodes to be related, while retrospective encapsulations summarize the episodes that have been told.

Results Bilingual narrative components Table 3 compares the narrative components used in the French and Turkish texts produced by the bilingual children. Across the three age groups the use of Component 1 remains relatively constant. The only divergence is observed

 Mehmet-Ali Akinci, Harriet Jisa & Sophie Kern

Table 2. Components of the Frog Story COMPONENT 1: ONSET OF THE PLOT Explicit mention must be made of the boy’s noticing that the frog is missing. le lendemain matin quand l’ garçon i s’réveille sur son lit i voit qu’ la grenouille est partie (F10;7k)a (‘the next morning when the little boy wakes up on his bed he sees that the frog is gone’) COMPONENT 2: UNFOLDING OF THE PLOT Explicit mention must be made of the search for the frog at least three times. euh Pierre cherche dans ses bottes. Rouki dans le bocal. ….. Pierre va à la fenêtre et appelle Zizi! Zizi!……. Ils continuent. Ils appellent Zizi! Zizi! …. Pierre cherche dans un trou. …. Pendant ce temps Pierre est monté dans un arbre et regarde dans un trou. …. Pierre monte sur un gros caillou et continue d’appeler. (F10;6o) (Peter searches in his boots. Rouki in the jar. ….. Peter goes to the window and calls Zizi! Zizi! ….. They continue. They call Zizi! Zizi! ….. Peter looks in a hole. …. During this time Pierre climbed up a tree and looks in a hole. ….. Peter climbs up on a rock and continues to call’) COMPONENT 3: RESOLUTION OF THE PLOT The frog taken home at the end of the story must be explicitly described as being the same or as substituting for the frog that the boy lost in the beginning of the story. Pierre et Rouki repartent avec Zizi la grenouille (F10;6o) (‘Peter and Rouki return with Zizi the frog.’) COMPONENT 4: ENCAPSULATIONS Summarizing (prospective or retrospective) of the ongoing search. maintenant le petit garçon poursuit ses recherches (F10;2b) (‘now the little boy pursues his search’) ils la cherchèrent partout (F11;5c) (‘they looked for her everywhere’) Source: Berman 1988; Berman and Slobin 1994: 46 a The numbers following the examples identify the subjects. F refers to French monolinguals, TF refers to Turkish–French bilinguals. The children’s ages are given in years and months. The letter following the age indicates the individual child in the age group.

in the 5-year-olds, who show slightly more encodings of the onset of the plot in French than in Turkish. For Components 2–4, performance in French slightly surpasses performance in Turkish. There are possibly two reasons for this. The first is methodological. As mentioned above, all of the children told the story first in Turkish. It may be that familiarity with the task provides a slightly better grasp of the story for the French production. The second possible reason is that the children have had more experience with this kind of exercise in French than in Turkish, given that all the children go to French

Influence of L1 Turkish on L2 French narratives 

Table 3. Proportion of Turkish–French bilingual narrators making explicit mention to each of the four components in their Turkish and French texts Turkish bilingual

French bilingual

Age Group N

5 14

7 16

10 15

5 13

7 16

10 15

Component 1: onset of the plot Component 2: unfolding of the plot Component 3: resolution of the plot Component 4: encapsulations

.21 .07 .07 .07

.81 .25 .31 .06

.93 .53 .33 .20

.36 .14 .00 .07

.81 .44 .25 .25

.93 .66 .27 .27

school where storytelling and story reading by the teacher are regular activities. The difference between the French and Turkish texts are greater for the 5and 7-year-olds than for the 10-year-olds. There is, however, no significant difference between the total number of components mentioned in the Turkish and the French texts (5-year-olds, W=7.5, 5df, NS; 7-year-olds, W=12, 7df, NS; 10-year-olds, W=10.5, 8df, NS).

Narrative components in L1 vs L2 French Table 4 compares the bilingual French texts with the monolingual French texts. The figures for the bilingual children are repeated in Table 4 for convenience of comparison. Both the bilinguals and monolinguals show development toward the encoding of more narrative components. The interaction of age and total number of components was significant for both the bilingual group (F (2,39)=14.40, p il va la prendre, ‘he’s going to take it’ [= the frog, (la grenouille), feminine]) preposition The major error consists of overgeneralising the preposition dans (‘in’). puis il regarde dans la fenêtre (TF7;6b) (>il regarde par la fenêtre, ‘he looks in the window’) il met dans sa tête (TF5;6h) (>il le met sur sa tête, ‘he put (it) in his head’) object missing in obligatory transitive context et le chien il peut pas enlever (TF5;6h) (>et le chien il peut pas l’enlever, ‘and the dog he can’t take off ’) determiner missing garçon il rigole (TF5;11m) (>le garçon il rigole, ‘boy he laughs’) chien il part (TF5;11m) (>le chien il part, ‘dog he leaves’) subject missing il a aussi trouvé des grenouilles. est content (TF5;11f) (>il a aussi trouvé des grenouille. il est content, ‘he also found some frogs. is happy’) regarde gurba. après tomb/e/ là. après parti. (TF5;5q) (>il regard la grenouille. après il a tombé là. après il est parti, ‘look frog. after fall there. after left.’) auxiliary missing The past perfect (passé composé) is formed by the auxiliary (either avoir (‘have’) or être (‘be’)) plus the past participle. The auxiliary carries tense and agrees with the subject. il pas pris l’autre grenouille (TF5;11f) (>il a pas pris l’autre grenouille, ‘he not take the other frog’) agreement Agreement in the present tense is rarely audible in spoken French. There are some verbs, however, which make an audible difference between the 3rd person singular and plural. le garçon dorment (TF5;6i) (>le garçon dort, ‘the boy sleep’) les mouches elle suit le chien (TF5;6i) (>les mouches elles suivent le chien, ‘the flies they follows the boy’) infinitive The verb in French carries tense and agrees with the subject. il dire viens viens (TF5;8n) (>il dit viens viens, ‘he say(INF) come come’) la petite fille sortir (TF5;8n) (>la petite fille sort, ‘the little girl leave(INF)’)

 Mehmet-Ali Akinci, Harriet Jisa & Sophie Kern

copula (être) missing la petite fille il debout (TF5;8n) (>la petite fille elle est debout, ‘the little girl he standing up’) la petite fille là-bas comme ça (TF5;8n) (>la petite fille est là-bas comme ça, ‘the little girl over there like that’) word order: object misplaced Object clitics are placed before the verb in French. il allait piquer lui (TF5:6a) (>il allais lui piquer, ‘he was going him to sting’) existential elle est (‘she is’) used as il y a (‘there is’). The existential form in French consists of the masculine pronoun il, an oblique clitic y, and a tensed form of avoir (‘have’). après elle est une pierre (TF5;10d) (>il y a une pierre, ‘after she is a rock’,=there is a rock) après elle est une maison (TF5;10d) (>il y a une maison, ‘after she is a house’=there is a house) past participle The past participle accompanies the auxiliary in the perfect tense (passé composé). The error consists of using the present tense form instead of the past participle. le chien il a prend la ballon (TF5;6h) (>le chien il a pris le ballon, ‘the dog he tooked the ball’) reflexive missing The reflexive pronoun precedes the verb in French. la grenouille est en train de sauver (TF7;5e) (>la grenouille est en train de se sauver, ‘the frog is saving’)

References Adam, J.-M. (1985). Le texte narratif: Précis d’analyse textuelle. Paris: Fernand Nathan. Akinci, M.-A. (1999). Dévelopment des competences narratives des enfants bilingues turcfrançais en France âgés de 5 à 10 ans. Doctoral dissertation. Université Lumière-Lyon 2. Aksu-Koç, A. (1994). Development of linguistic form: Turkish. In: R. Berman. and D. I. Slobin (eds.), Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum, 329–85. Berman, R. (1988). On the ability to relate events in narrative. Discourse Processes. 11, 469–97. Berman, R. and D. I. Slobin (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Clark, E. (1985). The acquisition of romance, with special reference to French. In: D. I. Slobin (ed.), The crosslinguistic study of language acquisition, Volume 1: The data. Hillsdale, NJ: Lawrence Erlbaum , 687–782.

Influence of L1 Turkish on L2 French narratives 

Comrie, B. (1981). Language universals and linguistic typology. Chicago: University of Chicago Press. Cummins, J. (1984). Bilingualism and special education: issues in assessment and pedagogy. Clevedon, Avon, England: Multilingual Matters. Cummins, J. (1991). Interdependence of first- and second-language proficiency in bilingual children. In: E. Bialystok (ed.), Language processing in bilingual children. Cambridge: Cambridge University Press. French, L. A. and K. Nelson (1985).Children’s acquisition of relational terms: some ifs, ors, and buts. NY: Springer Verlag. Garnham, A., J. Oakhill and P. Johnson-Laird (1982). Referential continuity and the coherence of discourse. Cognition, 11, 29–46. Heath, S. B. (1982). Protean shapes in literacy events: Ever-shifting oral and literate traditions. In: D. Tannen (ed.), Spoken and written language. Norwood, NJ: Ablex. Heath, S. B. (1984). Ways with words: Language, life, and work in communities and classrooms. Cambridge: Cambridge University Press. Hickmann, M. (1995). Discourse organization and the development of reference to person, space, and time. In: P. Fletcher and B. MacWhinney (eds.), Handbook of language acquisition. Oxford: Blackwell. Hickmann, M. M. Kail and F. Roland (1995). Cohesive anaphoric relations in French children’s narratives as a function of mutual knowledge. First Language, 15, 277–300. Jisa, H. (2000). Increasing cohesion in narratives: A developmental study of maintaining and introducing subjects in French. Linguistics, 38, 591–620. Jisa, H. and S. Kern (1995) Discourse organization in French children’s narratives. In: E. Clark (ed.), Proceedings of the 26th Annual Child Language Research Forum. Stanford: Stanford University Press, 177–88. Jisa, H. and Kern, S. 1998. Relative clauses in French children’s narratives. Journal of Child Language, 25, 623–52. Kern, S. (1997). Comment les enfants jonglent avec les contraintes communicationnelles, discursives et linguistiques dans la production d’une narration. Unpublished doctoral dissertation, Université Lumière-Lyon 2, France. Labov, W. and J. Waletsky (1967). Narrative analysis: Oral versions of person experience. In: J. Helm (ed.), Essays on the verbal and the visual arts. Seattle: University of Washington Press. Mandler, J. M. and N. S. Johnson (1977), Remembrance of things passed: Story structure and recall. Cognitive Psychology, 9, 111–51. Mandler, J. M. and N. S. Johnson (1980). On throwing out the baby with the bathwater: A reply to Black and Wilensky’s evaluation of story grammars. Cognitive Science, 4, 304–12. Mandler, J. M. (1978) A code in the node: The use of a story schema in retrieval. Discourse Processes, 1, 14–35. Mayer, M. (1969). Frog, Where are you? New York: Dial Press. Rumelhart, D. E. (1975) Notes on a schema for stories. In: D. G. Bobrow and A. M. Collings (eds.), Representation and understanding: Studies in cognitive science. New York: Academic Press, 211–36.

 Mehmet-Ali Akinci, Harriet Jisa & Sophie Kern

Rumelhart, D. E. (1977) Understanding and summarizing brief stories. In: D. Laberge and S. J. Samuels (eds.), Basic processes in reading: Perception and comprehension. Hillsdale, NJ: Lawrence Erlbaum, 265–303. Rumelhart, D. E. (1980) On evaluating story grammars. Cognitive Science, 4, 313–16. Shapiro, L. R. and J. A. Hudson (1991). Tell me a make-believe story: Coherence and cohesion in young children’s picture elicited narratives. Developmental Psychology , 27, 960–74. Snow, D. E. and M. Hoefnagel-Hohle (1978). The critical period for language acquisition: Evidence from second language learning. Child Development, 49, 1114–28. Stein, N. L. (1982) The definition of a story. Journal of Pragmatics, 6, 487–507. Stein, N. L. and T. Trabasso (1981). What’s in a story: Critical issues in story comprehension. In: R. Glaser (ed.), Advances in the psychology of instruction, Hillsdale, NJ: Lawrence Erlbaum, 213–67. Thorndyke, P. W. (1977) Cognitive structures in comprehension and memory of narrative discourse. Cognitive Psychology, 9, 77–110.

Chapter 8

Development of temporal relations in narratives by Turkish–Dutch bilingual children Jeroen Aarssen

For a language learner, acquisition of temporal relations involves the acquisition of a clearly defined set of forms for expressing these relations. The acquisition of this set as a whole is not instantaneous, but should be seen as the outcome of a developmental process. Therefore, certain forms will, in the course of time, serve new functions, while new forms are acquired for expressing old functions (Slobin 1973). Bilingual children simultaneously acquiring two languages with diverging temporal systems, are faced with the task of keeping track of the right system in the right language. In this respect, studying temporal features of narratives means investigating the order and the way in which learners tackle these acquisitional problems. The focus of this chapter will be on the following research questions: i.

How do bilingual children temporally organise their narratives? Which tense is used to ‘anchor’ their narratives in Turkish and which in Dutch? Are tense shifts motivated or erratic? ii. How do bilingual children express the temporal relation of simultaneity at different ages, in both Turkish and Dutch? What are the functions of the expression of simultaneity? How does the ability develop to conceptualise and express different types of simultaneity? To answer these questions, the narratives of bilingual children in seven age groups (4 to 10 years of age) in Turkish and Dutch were analysed. There are twenty informants per age group, which makes a total of 140 narratives in Turkish and 140 in Dutch.

 Jeroen Aarssen

Temporal organisation of narratives This chapter1 focuses on both local and global dimensions of the temporal organisation of narratives. The first objective is to obtain insight into the global organisation of the narratives. It will be investigated which tense children choose for presenting the events. One important criterion for the well-formedness of a narrative is the consistent selection of one tense throughout the story, the anchoring tense. It will be looked at how (and if at all) the informants temporally chain clauses in their stories, that is, whether they maintain the temporal frame adopted as introductory tense. The second objective is to give a detailed account of the way in which children express simultaneous events. To be able to make claims about and comparisons of the system of temporal relations in Turkish and Dutch, the expression of a temporal concept, rather than of temporal forms, will be the basic unit of analysis. The concept of simultaneity (which can be described roughly as the relation between two events or states taking place at the same time) was adopted as the starting point in the analysis of temporal relations. There are a number of reasons for adopting this particular concept. First, simultaneity is a core notion of temporality (the other two being before and after relations). Second, simultaneous events are perceptually accessible for children, even as young as age 4. Narrators (of all ages) may feel the need (f.i. triggered by pictures) to express that certain events happen at the same time. However, since simultaneity is usually more marked than expressions used for anteriority or posteriority, narrators may avoid expressing it (di Luzio 1994: 249). Third, the concept can be expressed by means of a large variety of forms, which holds especially true when two languages as different as Turkish and Dutch are analysed in combination. In the expression of simultaneous events, grammatically advanced principles such as subordination (in Dutch) or the formation of converbs (in Turkish) are involved, in combination with discourse principles (such as deviations from the natural order of events). Fourth, the concept of simultaneity can be expressed both at the local level (chaining two adjacent utterances) and at the global discourse level (connecting different episodes or scenes). The analyses are inspired by the research on temporal relations in L1 acquisition of German and Turkish by Aksu-Koç and von Stutterheim (1994). The underlying framework of time is the one developed by Klein (1993, 1994) in his research on adult second language acquisition.

Temporal relations in Turkish–Dutch narratives

Klein (1994: 25) proposes a temporal framework which consists of the Time of Utterance (TU), Time of Situation (TSit) and Topic Time (TT). TU is the actual time of speech, TSit is the time interval in which the event takes place and TT is the time for which a certain claim is made. Whereas TSit refers to the infinite part of an utterance, TT is the time which corresponds to the finite component. For instance, when a child is building a narrative on the basis of the frog story, he may use an utterance such as: (1) The frog escaped from the jar TSit=frog escaping from jar TT=the finite component denoting the time interval for which the information given by the infinitive component is true (PAST)

This means that tense imposes a temporal constraint on the situation described in the infinitive component: the particular situation can only be true for some given time.

Informants and data collection Data have been collected at primary schools in different parts of the Netherlands. All schools have a high proportion of pupils with a home language other than Dutch. Table 1. Number of bilingual informants Informants

Language(s)

N

Data collected at age

Bilinguals Cohort T1 (longitudinal) Cohort T2 (longitudinal)

Turkish and Dutch Turkish and Dutch

20 20

4567 8 9 10

The age range of the bilingual informants is from 4 (from the first months of school contact onwards) until age 10. The design of the study is pseudolongitudinal, i.e. two consecutive cohorts of 20 informants have been followed during a longer period of time: the younger cohort (T1) in four rounds (from age 4 to age 7) and the older cohort (T2) in three rounds (from 8 to 10). This means that the subsequent age groups of both cohorts taken together correspond to the first 6 (out of 8) grades of the Dutch primary school system, called basisschool. In the basisschool two years of kindergarten (grade 1 and 2) are included.



 Jeroen Aarssen

The narrative data were collected by means of a picture story book ‘Frog where are you?’ (Mayer 1969). An important reason to use this story was that it has already been used in earlier research (see for instance Bamberg 1987 and especially Berman and Slobin 1994a), which makes cross-linguistic comparison possible. The reason to use picture stories for eliciting narrative data, was that a ‘common content’ was provided for all children in all age groups. The story is about a boy and his dog who have captured a pet frog in a jar. The frog escapes. The boy and the dog go out and try to find it. After a series of adventures, they find it again. The frog story has considerable length, consists of a number of events, and allows the informants to express a wide range of topics.

Analysis The first analysis was done by reading through all transcripts line by line, while coding for anchoring tense in all clauses expressing foreground events (that is, the introductory tense and tense shifts that could either be classified as unmotivated or motivated). What may influence the choice for present or past is the fact that the assignment given by the researcher before the retelling starts, also contains tense. Since this tense might be guiding the selection of an anchoring tense by the informants, this assignment was always consistently given in present tense (‘can you say what happens’ and not ‘can you say what happened’). It should be noted here that typical stage setting at the beginning of the narrative is regarded as background information and therefore not taken into account in the analyses. Therefore, what is called ‘introductory tense’ here is not in all cases similar to ‘the first tense marked verb’. The informants can start to narrate foregrounded events in present tense (either following the temporal frame provided by the researcher, or independently of it), or in past tense. It was looked at whether whichever time chosen as introductory tense was maintained throughout the story. It is clear that only tense maintenance in foregrounded events is of interest here. There are in fact two kinds of unmotivated tense shifts. The first one is when children start off in one tense and somewhere in the middle unexpectedly switch to the other tense. In the second type the larger part of the narrative is in one tense, with tense shifts occurring only in some single clauses. For the latter type, it seemed necessary to set a minimum number of shifts, since

Temporal relations in Turkish–Dutch narratives 

these might be slips of the tongue only, and not clear-cut signs of the inability to organise discourse. The fact that single unexpected shifts occur in adult speech as well, supports this claim. The minimal number was set at five cases: if single unmotivated tense shifts occurred more than five times, the narrative was labelled as having ‘mixed’ tense. This number is substantially stricter than the criterion used by Aksu-Koç (1994: 333), which was 25 per cent or more. Motivated tense shifts were, of course, considered as part of the story-line, but not counted as a deviation from the anchoring tense. In addition, a computerised analysis was performed on the narratives in text file format to track down those specific devices that might express simultaneity. These forms were located and presented with parts of the context (in terms of clauses before and after the clause containing the target form). The output files were scrutinised in order to distinguish those forms that explicitly or implicitly expressed simultaneity from those forms that had different functions in the text.

Results Temporal anchoring of narratives One way of organising a narrative is to maintain an anchoring tense throughout the text. The number of narratives in which there is not one consistently favoured tense (hence called ‘mixed tenses’) is expected to decrease when children grow older. Younger children will start to use present tense forms for anchoring, since past tense forms are not available at very young ages. Older children will gradually use past tense forms more often. There does not seem to be a default tense for third person narratives. Earlier frog story research showed that monolingual English, German and Spanish adults predominantly use present tense in their stories (Berman and Slobin 1994b: 132, Bamberg 1994: 194, Sebastián and Slobin 1994: 244). On the other hand, adult monolingual speakers of Hebrew show no clear preference (Berman and Neeman 1994: 291). Aksu-Koç (1994: 334) found that, although most monolingual Turkish children and adults use the -iyor present, the Turkish adults show no clear preference. Turkish has two past tense forms: -di and -mis¸. The latter has the function of evidential past, and is also used as narrative modality.

 Jeroen Aarssen

In Table 2 the anchoring tenses in foregrounded events in the Turkish data of the bilingual informants are presented. Table 2. Anchoring tenses in L1 Turkish narratives of Turkish–Dutch bilingual informants, per age group Age Present Past -di -mis¸ Mixed Pres/Past

Cohort T1 (N=20)

Cohort T2 (N=20)

4 8 – – 12

8 6 2 5 7

5 12 – – 8

6 6 4 3 7

7 7 3 2 8

9 10 3 2 5

10 – 7 6 7

This table shows that the majority of the 4-year-old bilingual informants make use of mixed (present and past) tense in their Turkish narratives. There are no instances at all of narratives with past tense only. At age 4 there are parts in present tense and also parts in past tense within the same narrative. These shifts in tense are in most cases ‘unmotivated’: there is no explicit need for presenting the particular event as temporally different from the other events. At age 5 the informants mainly choose the present as the anchoring tense in their narratives, but still a large number of mixed tense narratives is found. From age 6 onwards, children gradually acquire the ability to use past tense as the dominant tense (both -di and -mis¸). In the Turkish narratives of the 7-year-olds, 60 per cent uses mixed past and present tenses. There is, however, also a tendency to use the past tense more often than the present tense as anchoring tense. The Turkish narratives of the 8-year-olds show less examples of unmotivated tense shifts, and more present-only texts. Still, past tense narratives, both with -di forms and -mis¸ forms, seem to gain ground. Half of the 9-year-olds uses present tense as anchoring tense in their Turkish narratives. However, none of the bilingual 10-year-olds uses present as anchoring tense. This means that all informants use past tense forms, although 7 out of 20 still make unmotivated tense shifts. Table 3 presents the distribution of anchoring tenses in the Dutch narratives of the same children. In the Dutch narratives a substantial number of informants at age 4 and 5 does not use finite verb forms at all, or only in combination with infinitives. Four children do not use any verb forms at all. They merely sum up the protagonists and other concrete elements (bed, jar, tree, etc.) in the pictures. Three other 4-year-old informants only use infinitives instead of finite verb forms, as in example (2).

Temporal relations in Turkish–Dutch narratives 

Table 3. Anchoring tenses in L2 Dutch narratives of Turkish–Dutch bilingual informants, per age group Cohort T1 (N=20) Age Present Past Mixed Pres/Past Mixed (other) Infinitives only No verb forms

4 7 1 – 6 3 3

5 14 1 1 3 – 1

Cohort T2 (N=20) 6 13 2 4 1 – –

7 12 2 5 1 – –

8 10 6 4 – – –

9 5 9 6 – – –

10 3 13 4 – – –

(2) kikker pakken. jij ook kikker pakken. hij pakken jongetje kikker. frog take-inf. you too frog take-inf. he take-inf boy-dim frog. en jongetje allemaal kikker zo doen and boy-dim all frog like-this do-inf ‘take frog. you too take frog. he take boy frog. and boy all frog do this’ [I˙smail, age 4;11, in Dutch]

These two types of stories (‘no verb’ and ‘infinitive only’) are illustrative for the way in which young informants give static descriptions based on single pictures in the frog booklet. Infinitives are typical means for the description of a static situation, similar to merely mentioning characters and objects without presenting them as actors in the story. From age 5 and 6 onwards, the present tense becomes the most prominent anchoring tense, and past tense forms start to appear in the data. First, they appear rather randomly (mixed Present/Past) and unmotivated. From age 8, however, they start to become more prominent, becoming the most frequent anchoring tense at age 9 and 10.

The expression of simultaneous events Simultaneity can be described as the relation between two events, processes or states that share a value on the time axis (Aksu-Koç and von Stutterheim 1994: 396). The expression of simultaneity may have the function of presenting backgrounded events or backgrounded information. Given the fact that the default structuring principle in narratives is that of sequentiality, some contexts demand explicit marking of simultaneity. In fact, three types of simultaneity should be distinguished:

 Jeroen Aarssen

i.

simultaneity of events or states in the foreground (two characters are involved in different actions) ii. simultaneity of events or states within the background iii. simultaneity of events or states across foreground and background. Expressions of simultaneity may be triggered by simultaneity on the level of perception: when in one picture two situations occur (the frog story has a substantial number of these). This level is important because the way in which the informants conceive of the task (as a dynamic description of related events or as static description of isolated pictures) may have an impact on the way they express simultaneous relations.

Table 4. Explicit and implicit means for expressing simultaneity in Turkish and Dutch Turkish

Dutch

Explicit converbs -ince ‘as soons as’ -ken ‘while’ -ip ‘and then’ -erek ‘by V-ing’ verbal nouns -dig˘i zaman ‘at the time of V-ing’ -dig˘inde ‘at V-ing’ adverbial means bu arada/o arada ‘in the meantime’ öte yandan ‘on the other hand’ o zaman ‘at that time’ aynı anda ‘at the same time’ particle de ‘too’

adverbs ook ‘too’ conjunctions en ‘and’ toen ‘when’ (+past) als/wanneer ‘when’ (+pres) terwijl ‘while’ adverbial means tegelijk/tegelijkertijd ‘in the meantime’ intussen ‘meanwhile’ op dat moment ‘at that moment’

Implicit complement clauses -dik, ki ‘that’ causal adv. clauses -dig˘i için ‘because of ’ çünkü ‘for’ adverbs hala ‘still’ hemen ‘immediately’ tam ‘just’

causal conjunctions want ‘for’ omdat ‘because/since’ complement clauses of ‘whether’ hoe ‘how’ adverbs net ‘just’ nog ‘still’ altijd ‘always’

Temporal relations in Turkish–Dutch narratives 

Turkish and Dutch differ to a large extent in the way simultaneity can be expressed. A short overview of means for expressing simultaneity in Turkish and Dutch is given below (see Table 4). Turkish expresses simultaneous relations by means of various devices, mainly through verb morphology: converbs and nominalised forms of the verb. There are some lexical means as well. Converbs are nonfinite verb forms, attached to the verb stem, which take their temporal specification from the main verb. There are four converbs that may encode actions simultaneous with the action expressed by the main verb. The converb -ince means ‘as soon as’ and indicates immediate succession (see example 3). The converb -ken, ‘while’, has a more explicit simultaneous meaning, and the action is typically from an imperfective perspective, i.e. seen as an ongoing process (see example 4). Both -ince and -ken specify the TT of the event in the main clause: they define the time span during which the main clause event takes place. (3) -ince og˘lan geyig˘i görünce hemen tutuyor boy deer-acc see-ince immediately grab-prog.3sg ‘when the boy sees the deer, he immediately grabs him’ (4) -ken og˘lan uyur-ken kurbag˘a kaçmıs¸ boy sleep-ken frog escape-m.past.3sg ‘while the boy was sleeping the frog escaped’

Two other converbs, -ip ‘and (then)’ and -erek ‘by V-ing’, function to integrate two situations as either immediately successive or simultaneous. The converb -ip (example 5) serves to package constituents of an event into a larger event (Aksu-Koç 1994: 347). The meaning of -erek (example 6) is manner of action or instrumentality. (5) -ip çocuk kalkıp kurbag˘ayı aramaya bas¸ladı boy get.up-ip frog-acc search-inf-dat begin-d.past.3sg ‘the boy got up and started to look for the frog’ (6) -erek gülerek kavanozdan çıktı laugh-erek jar-abl get.out-d.past.3sg ‘he got out of the jar laughing’

 Jeroen Aarssen

There are two types of nominalised verb forms: the first one is -dig˘i zaman (verb + dik + poss followed by postposition zaman), meaning ‘at the time of V-ing’. The second one is -dig˘inde (verb + dik + poss + loc), meaning ‘at V-ing’. The situation expressed in these adverbial clauses is seen as perfective. These morphological means express simultaneity at a local level (connecting adjacent utterances). Turkish has adverbial means as well, which represent simultaneity at a more global level. These devices may mark a topic shift, and are frequently used to relate episodes. Bu arada/o arada ‘in the meantime’ and öte yandan ‘on the other hand’ express a simultaneous relation between two episodes. They typically introduce a new situation as foregrounded. O zaman ‘at that time’ and aynı anda ‘at the same time’ may be used to present two isolated situations as simultaneous. The standard topic/focus particle de (or its allomorphs da, te and ta), meaning ‘too’ can be used to express simultaneity. Aksu-Koç and von Stutterheim (1994) consider it to be a compensatory device in children’s speech. This label, however, is highly normative and does not fit well in their classification. Aksu-Koç and von Stutterheim (1994) make a threefold classification of means for expressing simultaneity: explicit lexical means, implicit means and compensatory devices. Contrary to this classification, de will be regarded in this chapter as an explicit form for marking simultaneity. The subjective interpretation of whether it is a compensatory device or not, is not for the researcher to decide upon. Finally, various forms do not explicitly encode the temporal relation of simultaneity, but their meaning contributes to the interpretation of two situations as being simultaneous. This implicit simultaneity can be expressed by means of complement clauses, causal adverbial clauses meaning (-dig˘i için ‘because of ’ and çünkü ‘for’). Other adverbial means such as hala ‘still’, hemen ‘immediately’ and tam ‘just’, can be used to attach an imperfective meaning to an utterance, making the event or state unbounded. When such an utterance is preceded or followed by an utterance without a specific temporal reference, both events might be interpreted as having a relation of simultaneity. Compared with Turkish, Dutch is rather limited as regards the available means for expressing simultaneity. In many ways it resembles German (see Aksu-Koç and von Stutterheim 1994: 402 ff.). Dutch does not have any morphological means for expressing simultaneity, other than the sparingly used present participle, as in hij kwam lopend ‘he came walking’. Instead, Dutch uses conjunctions and adverbs. The adverb ook ‘too’ can express simultaneity, depending on the context. Since the semantics of ook are similar to

Temporal relations in Turkish–Dutch narratives 

those of German auch, it can be inferred that Aksu-Koç and von Stutterheim would consider it to be a compensatory marker, as was the case with de in Turkish. The objection against their label of ‘compensatory marker’ is even stronger when it comes to ook. In this chapter, ook is regarded as (and will turn out to be) an important explicit adverbial device in Dutch for expressing simultaneity. The coordinating conjunction en ‘and’ may express simultaneity of both coordinated clauses. Temporal subordinating conjunctions are toen ‘when’ (+past), als/wanneer ‘when’ (pres) and terwijl ‘while’. The adverb (en) toen ‘(and) then’ is comparable to Turkish o zaman, in that it may function to switch perspective to unrelated but simultaneous actions. Only in the meaning ‘at that time’ it expresses simultaneity. The prototypical use of (en) toen in children’s narratives, however, is the expression of successive actions. In this way, it can be regarded as an easy filler form or discourse marker to start an utterance. Among the adverbs and adverbial phrases that express simultaneity are tegelijk/tegelijkertijd ‘in the meantime’, intussen ‘meanwhile’, and op dat moment ‘at that moment’. These adverbs express simultaneity at a global level, relating simultaneous episodes. In many ways simultaneity can be inferred from the context, from the inherent meanings of the specific clauses. Among those devices that do not encode simultaneity explicitly but imply it from the context are causal conjunctions (want ‘for’ and omdat ‘because/since’), and complement clauses with of ‘whether’ or hoe ‘how’. Utterances can be made temporally unbounded by means of expressions which have a durative or imperfective meaning. If this utterance is followed or preceded by an utterance that does not have any temporal reference, the two will be interpreted as simultaneous. Such expressions are adverbs like net ‘just’, nog ‘still’, altijd ‘always’ etc.

Simultaneity in Turkish Table 5 presents a quantitative overview of devices for expressing simultaneity in the Turkish data of the bilingual informants. The first figure in each column gives the absolute number of occurrences of that particular form in the data of a given age group. The second figure indicates the number of informants that used the form. On the whole there are not many instances in the Turkish data in which simultaneous events are expressed. Moreover, a large number of implicit means, such as the adverbial phrases öte yandan ‘on the other hand’ or bu

 Jeroen Aarssen

Table 5. Distribution of expressions of simultaneity in L1 Turkish narratives of Turkish–Dutch bilingual informants, by age Cohort T1 (N=20) Age Explicit means a. converbs V+ince V+ken V+ip V+erek b. adverbial clauses V+dik+de V+dig˘i zaman c. adverbials o zaman d. conjunctions ve e. particles de/da/te/ta Implicit means a. compl. clauses V+DIK ki b. adverbial clauses V+dig˘i için çünkü c. adverbs hala tam

Cohort T2 (N=20)

4

5

6

7

8

9

10

1/1 9/2 1/1 –

– – – –

1/1 – 2/1 1/1

– 1/1 2/2 –

1/1 11/7 1/1 1/1

4/4 8/3 3/3 –

5/4 14/8 6/4 4/3

– –

– 2/1

– –

– –

1/1 1/1

– –

1/1 3/3

12/3

6/4

15/6

15/5

8/4

9/3

3/1

1/1

–

2/2

–

1/1

–

5/4

44/13

10/7

32/16 33/12

81/20

46/18

51/18

– –

– –

– 3/2

– 3/2

1/1 5/3

1/1 7/4

– 6/4

– –

– –

– –

– 2/1

2/2 2/2

– –

– 2/2

– –

– –

– –

– 1/1

1/1 1/1

– –

– –

arada ‘in the meantime’, or the adverb hemen ‘immediately’ do not appear in the data at all. For all age groups the adverb o zaman and the particle de are the most frequent forms. As they grow older, children make use of more and more forms to express simultaneity. The fact that within the group of 5-yearolds less devices for the expression of simultaneity in Turkish are found, can be related to earlier findings that 5-year-olds present a more global story line, which results in shorter stories (Karmiloff-Smith 1985). In the Turkish data of the 4-year-old bilingual children no implicit means are found. Besides the adverbial o zaman and the particle de/da, the converb -ken is used nine times. This form meaning ‘while’ in the target language was used correctly in three cases, as in example (7):

Temporal relations in Turkish–Dutch narratives 

(7) çocuk uyurken kurbak içinden çıkıyor child sleep-ken frog inside-poss-abl get.out-pres ‘while the child is sleeping, the frog gets out from inside’ [Rabia, age 4;11, in Turkish]

One informant, however, uses the converb -ken six times in combination with the verb git- ‘go’. The action expressed by the main verb diyormus¸ ‘he was saying’ is simultaneous to the going of the actor, whereas the going of the actor is being presented as a continuous action, as the following example shows: (8) sonra da giderken 〈anne [/] anne ben oraya after da go-ken 〈mother [/] mother I there-dat gidecem> [‘‘] diyormus¸ go-fut-1sg> [‘‘] say-past.prog ‘while he was going he said 〈mother, mother I’m going there〉 [‘‘]’ [Figen, age 4;4, in Turkish]

In adult speech one would expect the converb -ip to appear in this and similar contexts, meaning ‘he went and said’ or ‘he went off saying . . .’. The same informant also uses the converbs -ince (the only occurrence) -ip and -ken at the age of 4. She is apparently making efforts to use different converbs in different contexts, although the story she created at age 4 is hardly coherent. It is all the more striking then that at age 5 she does not use any of these forms at all. At age 6 the converb -erek appears for the first time in the data, as well as complement clauses with ki, which is the first implicit device found in the Turkish data. Complement clauses with ki appear in most cases after a clause with the main verb bak- ‘look’. This is in fact standard use of the form, which means ‘I looked and saw that . . .’ (Lewis 1986: 212). The interpretation, however, of the utterance in (9) is that ki means ‘whether’ and not ‘that’, following the pictures in which the frog is actually gone: (9) baktılar ki kurbag˘a orda look-d.past-3pl whether frog there ‘they looked whether the frog was there’ [I˙smail, age 6;9, in Turkish]

Some 7-year-olds use the implicit form ki in complement clauses of the verb bak- ‘look’ as well. They use it, however, in the standard way, with the meaning ‘look and see that . . .’:

 Jeroen Aarssen

(10) oraya baktı ki kurbag˘a kaçtı there-dat look-d.past ki frog escape-d.past.3sg ‘he looked there and saw that the frog had escaped’ [Murat, age 7;0, in Turkish]

In this example there is simultaneity of Topic Time of both events and not on the level of the Time of Situation: the result of the second event (the frog having escaped=being gone) is presented as simultaneous with the first (the frog is gone at the time the boy looked). Another example illustrates that at age 6 the ability emerges to present events in such a way that the order of mention is not the order of events. In (11) a tense shift from present to past can be seen. The -mis¸ past has a clear perfective meaning. The result of the action expressed by the verb carrying the past tense morpheme -mis¸, is simultaneous with events expressed by the verbs with the present tense morpheme -iyor (which is the anchoring tense). Although there is no simultaneity of the actual time spans in which the events take place, the Topic Time of both events is identical. (11) geç kalkıyorlar. bakıyorlar. o zaman late get.up-pres.3pl look-pres.3pl that time kaçmıs¸ escape-m.past.3sg ‘they get up late. they look. then he has escaped’ [Selim, age 6;1, in Turkish]

The conjunction ve is primarily used for enumeration of actions (two actions by one protagonist). Only in two cases this conjunction is used for simultaneity, namely when it connects two actions by two protagonists, as in example (12), where it is combined with the particle da that marks a shift of topic: (12) köpek gidiyor ve og˘lan da oturuyor dog go-pres.3sg and boy da sit-pres.3sg ‘the dog goes and the boy is sitting’ [Sibel, age 6;2, in Turkish]

This interpretation is strongly based on the information derived from the picture book. It would have been impossible to claim that either simultaneity or sequentiality was intended, if there were no picture to support this claim. From age 8 onwards, children start to use converbs more frequently. The converb -ken is used 11 times by seven 8-year-olds. What is denoted

Temporal relations in Turkish–Dutch narratives 

by the verb taking the -ken suffix, are temporally unbounded events, as in example (13): (13) çocuknan köpek uyurken kurbag˘a kaçmıs¸ child-with dog sleep-pres-ken frog flee-m.past.3sg ‘while the dog and the child were sleeping, the frog fled’ [Muhammed, age 8;9, in Turkish]

Explicit forms used for the first time by 8-year-olds (both of them, however, only once) are the temporal adverbial clause with -dig˘inde ‘at the time when’ in example (14) and the complement clause with -dik in example (15): (14) sonra da # çıktıg˘ında ali kalktı after da # get.out-dig˘inda Ali stand.up-d.past.3sg ‘and then when he got out, ali stood up’ [Yurdagül, age 8;10, in Turkish] (15) sonradan bilmedi nerede oldug˘unu after-abl know-neg-d.past.3sg where be-dik-poss-acc ‘and then he didn’t know where he was’ [Eray, age 8;10, in Turkish]

The converb -ince ‘as soon as’ is in several cases used as a kind of temporal stepping stone, especially when attached to the verb bak- ‘look’ in combination with the main verb gör- ‘see’. The meaning then changes from ‘as soon as X, then Y’ to a semantic cluster ‘X and Y’: (16) bakınca böyle kurbag˘a ailesini görüyor look-ince thus frog family-poss.3sg-acc see-pres.3sg ‘he looks and thus sees the frog family’ [Zühal, age 9;2, in Turkish]

In other cases, -ince is used as a positional temporal adverb: (17) ondan sonra sabah olunca yatmıs¸ there-abl after morning become-ince sleep-m.past.3sg ‘and then, when it became morning, he slept’ [Özlem, age 9;6, in Turkish]

The converb -ken appears eight times in the data of the 9-year-olds. There is one example in which -ken is used, where -erek would be expected. In example (18) the subject of kos¸uyorken and gelmis¸ is the same (the deer):

 Jeroen Aarssen

(18) o zaman 〈bir tane〉 [//] bir ¸sey hızlı kosuyorken that time 〈one〉 [//] a thing quickly run-prog-ken gelmis¸ came-m.past.3sg ‘then one [//] a thing came running quickly’ [Ramazan, age 9;0, in Turkish]

Within the age group of the 10-year-olds, all explicit forms found in the entire data set are used. All four converbs are used, but still only by a minority of the informants. Even at this age examples can be found of -ince ‘as soon as’ used as if it were -ip ‘and then’ (19) and vice versa of -ip used as -ince (20): (19) kaçınca 〈her ¸sey〉 [//] og˘lan yere düs¸üyor flee-ince 〈every thing〉 [//] boy ground-dat fall-pres.3sg ‘〈everything〉 [//] the boy flees and falls to the ground’ [Ali Osman, age 10;7, in Turkish] (20) arılar arkasından gidip &ço çocuk [/] çocuk düs¸müs¸ bee-plur behind-poss-abl go-ip &chi child [/] child fall-m.past.3sg ‘as soon as the bees went after him (=dog), the child fell’ [Muhammed, age 10;8, in Turkish]

Simultaneity in Dutch Table 6 presents an overview of devices expressing simultaneity in the Dutch data of the bilingual informants. As in Table 5, the first figure gives the absolute number of occurrences of that particular form in the data of a given age group. The second figure indicates the number of informants that used the form. The top half of the table contains all explicit devices for expressing simultaneity that were found in the data; the bottom half lists those devices that imply simultaneity in a given context. The first obvious result that can be derived from Table 6 is that, in general, simultaneity is expressed in Dutch in only a small number of cases, explicitly or implicitly. A second observation is that obvious means for simultaneity, such as the conjunction terwijl ‘while’, or adverbs such as gelijk/tegelijkertijd ‘at the same time’ are not used at all. The conjunction en and the adverb ook are the most frequent in all age groups. Within the age group of the 4-year-olds, only few examples of simultaneity were found. Children at age 4 do not yet express simultaneity at all levels. The most obvious exception, however, is the explicit device ook ‘too’, Ook is used

Temporal relations in Turkish–Dutch narratives 

Table 6. Distribution of expressions of simultaneity in L2 Dutch narratives of Turkish–Dutch bilingual informants, by age Cohort 1 (N=20) Age Explicit means a. conjunctions toen als en b. adverbs (en) toen ook Implicit means a. causal conj. want omdat b. compl. clauses of hoe c. adverbs net nog opeens

Cohort 2 (N=20)

4

5

6

7

8

9

10

– 1/1 7/6

– 2/1 8/6

– – 9/6

6/4 – 20/11

4/3 3/3 3/1 – 24/14 20/8

– 1/1 23/11 18/9

– 18/9

– 21/14

5/5 7/5 – 30/16 18/13 23/12

1/1 1/1

– –

– –

1/1 1/1

– 2/2

– –

5/4 –

– –

1/1 –

– –

– –

1/1 –

1/1 –

3/1 –

1/1 – –

– 6/2 –

– 1/1 –

– 2/1 –

– 5/4 –

– 5/4 4/4

– 2/1 7/4

17/7 4/3 21/11

in a construction parallel to the one in the previous clause and refers to activities of two protagonists, either as subject or as object (21) and (22). (21) die jongen kijkt naar kikker. die hond kijkt that boy look-pres-3sg to frog. that dog look-pres-3sg ook naar kikker also to frog ‘that boy looks at frog. that dog also looks at frog’ [Levent, age 4;6, in Dutch] (22) hert ging vallen doen hem. # hond ook deer went fall-inf do him. # dog too ‘deer went to make him fall. dog too’ [Salim, age 4;8, in Dutch]

These examples, as well as other occurrences of ook, suggest that 4-year-olds are capable of expressing simultaneity on the level of perception (that is, when two situations appear in one picture). Less frequently than ook, the coordinating conjunction en ‘and’ is used for expressing simultaneous relations.

 Jeroen Aarssen

Occasionally, the 4-year-olds use implicit forms. Of these implicit forms, the conjunctions want ‘for’ and omdat ‘because’ indicate a backgrounded state or event: (23) hij heb@ um stil [‘‘] gezegd. want uh 〈hondje gaat〉 [/]# he has er hush [‘‘] say-pres.part for er 〈dog go-pres-3sg〉 [/] hondje gaat dur in dog-dim go-pres-3sg there in ‘he has said hush [‘‘], for the dog goes in it’ [Funda, age 4;5, in Dutch] (24) hij gaat hem prikken omdat hij kijkt naar hem he go-pres-3sg him sting-inf because he look-pres-3sg to him ‘he will sting him, because he looks at him’ [Sezgin, age 4;8, in Dutch]

Within the group of 5-year-olds, the adverb toen ‘then’ appears for the first time, although only once, as a device for expressing simultaneity. The subordinate conjunction toen ‘when’ first appears at age 7: (25) toen ze in donker was toen ging de kikker when she in dark is-past.3sg then go-past.3sg the frog zo dur uit met zo voet like-this there out with like-this foot ‘when she was in dark then the frog went out like-this with foot like-this’ [Rabia, age 7;7, in Dutch]

It was seen above that 7-year-olds use causal adverbial clauses in Turkish. In Dutch, the causal conjunctions want ‘for’ and omdat ‘because’ appear again in the data (both forms were used once by 4-year-olds). The informant who used them, is again the girl who introduced converbs in her Turkish narratives rather early and who started to use the Turkish çünkü ‘for’ at age 7. Examples (26) and (27) show both cases. In (27) there is again simultaneity of Topic Times: the effect of the event ‘‘frog being gone’’ is simultaneous with the state ‘‘boy not happy’’. (26) en die hond ging gauw rennen want dat jongen and that dog go-past.3sg quickly run for that boy was op het kangoeroe be-past.3sg on the kangaroo ‘and that dog started to run quickly for that boy was on the kangaroo’ [Figen, age 7;0, in Dutch]

Temporal relations in Turkish–Dutch narratives 

(27) het jongen is niet blij omdat de kikker weg the boy be-pres.3sg not glad because the frog away is gegaan be-pres.3sg go-pres.part ‘the boy is not glad because the frog has gone away’ [Figen, age 7;0, in Dutch]

The 9-year-olds use among other things the implicit device opeens ‘suddenly’. It expresses the ‘closing of the time gap’ between two situations. In this way, the informants manage to express a relation of simultaneity or immediate succession of two events. The interpretation of opeens expressing simultaneity is suggested strongly by the picture, in which both events are depicted. An example is given in (28): (28) en de hond zit nog steeds te spelen. opeens and the dog sit-pres.3sg still to play-inf. Suddenly komen er bijen uit de huisje van de honing come-pres.3pl there bee-plur out.of the house-dim of the honey ‘and the dog is still playing. Suddenly bees come out of the little house of the honey’ [Cemal, age 9;9, in Dutch]

The conjunction toen is very frequent in the Dutch data of the 10-year-olds. The adverb (en) toen, on the other hand, disappears. Whereas younger children used ook to indicate similar actions or states by different protagonists, at age 10 there is an example in which the first state is not explicitly mentioned, but can be derived from the clause containing ook: (29) 〈o mijn kikker is weg〉 [‘‘] zeg ron. en pietje 〈o my frog be-pres.3sg away〉 [‘‘] say-pres-0 ron. and pietje kijk ook verbaasd look-pres-0 also surprised ‘〈o my frog has gone〉 [‘‘] says ron. and pietje also looks surprised’ [Eray, age 10;9, in Dutch]

Furthermore, the 10-year-olds use a small number of implicit devices, such as the causal conjunction want ‘for’, complement clauses with of ‘whether, if ’, the adverbs nog ‘still’ and opeens ‘suddenly’. The subordinate conjunction als ‘when’ appears four times. There is one example of als erroneously used with past tense (that is, as if it were toen):

 Jeroen Aarssen

(30) als kees wakker werd zag hij 〈dat ze〉 [//] as kees awake become-past.3sg see-past.3sg he 〈that she〉 [//] dat hij niet er was that he not there be-past.3sg ‘when kees woke up, he saw 〈that she〉 [//] that he was not there’ [Özlem, age 10;7, in Dutch]

Conclusions and discussion In this chapter two issues relating to temporality in narratives were considered: the development of temporal anchoring of texts, and the development and use of devices expressing simultaneity of events. The use of infinitives only (as opposed to inflected verb forms) in some narratives of young children, and the absence of verb forms in some other narratives, provide evidence for the claim that children at ages 4 and 5 do not relate successive or simultaneous events into a coherent narrative, but instead tend to give static descriptions based on isolated pictures in the frog booklet. Moreover, the stories in which the children mix present and past tense, also support this claim. However, within the group of 5-year-old bilinguals some attempts to temporally organise texts can be seen. There is, for instance, a significant rise (compared to the 4-year-olds) in number of (en) toen and (en) dan ‘(and) then’ used as means for indicating sequentiality. At a later age there is a development towards the steady use of one favoured tense to anchor the narrative. The majority of the bilingual children at age 10 prefer past tense in both their Turkish and their Dutch narratives. This development in temporal organisation of the children’s narratives, is an indicator of a growing command of discourse. At an early stage children connect single utterances (at a local level) and at a later age they acquire the skills to structure their texts hierarchically (at the global text level). In both the Turkish and the Dutch data of the 4- and 5-year-olds, there are many occurrences of spatial adverbs referring to the picture book context (the spatial deictic adverbs burda/orda in Turkish and hier/daar in Dutch). The children do not relate the events within the pictures or across a series of pictures, but merely use deictic orientation to give a static description of what they see in the picture. This is, in turn, related to a change in function of simultaneity, from the level of perception (single pictures) at an early stage, to more advanced levels at which simultaneity is used for adding background information to foregrounded events, or for relating simultaneous episodes. Apparently, at age 4 and 5, many

Temporal relations in Turkish–Dutch narratives 

children have not yet reached the final stage of development of temporal systems (cf. Weist 1986: 387) in which TU, TSit and TT can represent three different time spans on the time axis. Young children seem not yet able to comprehend the concept of simultaneity of TSit’s or TT’s. The ability to distinguish between the abstract notions of TU, TSit and TT is a prerequisite for the ability to conceive of two situations as simultaneous. Young children might, however, be able to distinguish simultaneity on the level of perception. In conclusion, there is a development from chaining utterances in which children rely on spatial deictics (here/there) towards a gradual expanding set of devices to build hierarchical structures in the text. Older children develop the skills to organise their texts temporally and learn to distinguish different functions of simultaneity with respect to foregrounding and backgrounding. In this phase, they acquire and start to use those expressive devices that mark these distinctions. Simultaneity was expressed at the local level only. Adverbials meaning ‘meanwhile’ or ‘on the other hand’ were not found at all in the data. In fact, it is not surprising that no clear cases of simultaneity at the global level were found. The informants retell the story picture by picture by looking at them. This procedure results in a more sequential ordering of events at the expense of potentially simultaneously represented events. As for the Dutch data, this result is similar to the findings of Aksu-Koç and von Stutterheim (1994) for German. Only adults (and two 9-year-olds) use adverbial means for relating different episodes. In the Turkish data, however, they found some more frequent cases of simultaneity at the global level. Some devices that are used for expressing simultaneity appear to acquire new functions across time. At first, children use a limited set of forms with broad meanings. Gradually, they expand this set of forms and there are shifts in meaning of some forms. For instance, the Turkish converb -ip is used by some young informants to express a wide range of temporal relations (‘while’, ‘after’, ‘when’). With the progression of age, a development towards a more restricted use can be seen (two verbs expressing one ‘combined’ action), reflecting adult use. Some devices are used less frequently in later age groups than at earlier age. This means that children learn to abandon certain forms used for certain functions, because these forms are inappropriate or unconventional. Examples are Turkish o zaman and Dutch en toen both meaning ‘and then’, and the expression meaning ‘too’ (the Turkish particle de and the Dutch adverb ook). These devices are among the first forms used for the expression of simultaneous events, but decrease in number as the children

 Jeroen Aarssen

grow older. At a later age, children reserve these forms for expressing sequentiality and express simultaneity by means of other devices, such as conjunctions (in Dutch) and converbs (in Turkish).

Note . The research presented in this chapter is part of a larger project on the acquisition of cohesive devices by Turkish–Dutch bilingual children in The Netherlands (see Aarssen 1996), supported by the Linguistic Research Foundation (grant No. 300–174–002), which is funded by the Netherlands Organisation for Scientific Research, NWO.

References Aarssen, J. (1996). Relating events in two languages. Acquisition of cohesive devices by Turkish–Dutch bilingual children at school age. Ph.D. thesis, Tilburg University (=Studies in Multilingualism, Vol. 2, Tilburg: Tilburg University Press). Aksu-Koç, A. (1994). Development of linguistic forms: Turkish. In: Berman and Slobin (eds.), 329–85. Aksu-Koç, A. and C. von Stutterheim (1994). Temporal relations in narrative: simultaneity. In: Berman and Slobin (eds.), 393–455. Bamberg, M. (1994) Development of linguistic forms: German. In: Berman and Slobin (eds.), 189–238. Berman, R. A. and Y. Neeman (1994) Development of linguistic forms: Hebrew. In: Berman and Slobin (eds.), 285–328. Berman, R. A. and D. I. Slobin (eds.) (1994a). Relating events in narrative: A cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Berman, R. A. and D. I. Slobin (1994b) Development of linguistic forms: English. In: Berman and Slobin (eds.), 127–87. di Luzio, A.(1994). Temporal reference and narrative structures in Italian and German by Italian migrant children in Germany. In: G. Extra and L. Verhoeven (eds.), The crosslinguistic study of bilingual development. Amsterdam: Royal Netherlands Academy of Arts and Sciences, 219–52. Karmiloff-Smith, A. (1985). Language and cognitive processes from a developmental perspective. Language and Cognitive Processes, 1, 1, 61–85. Klein, W. (1993). The acquisition of temporality. In: C. Perdue (ed.), Adult language acquisition: Cross-linguistic perspectives, Vol. II: the results. Cambridge: Cambridge University Press, 73–118. Klein, W. (1994). Time in language. London: Routledge. Lewis, G. L. (1986). Turkish grammar. Oxford: Clarendon Press. Mayer, M. (1969). Frog, where are you? New York: Dial Press.

Temporal relations in Turkish–Dutch narratives 

Sebastián, E. and D. I. Slobin (1994). Development of linguistic forms: Spanish. In: Berman and Slobin (eds.), 239–84. Slobin, D. I. (1973). Cognitive prerequisites for the development of grammar. In: C. A. Ferguson and D. I. Slobin (eds.), Studies of child language development. New York: Holt, Rinehart and Winston, 175–211. Weist, R. M. (1986). Tense and Aspect. In: P. Fletcher and M. Garman (eds.), Language acquisition. Cambridge: Cambridge University Press, 356–74.

Chapter 9

Temporality issues in Moroccan Arabic and Dutch Petra Bos

Temporal relations refer to the anchoring of events to a given reference time. Tense oppositions in narratives not only function to locate events relative to the moment of speech, but also as organizers of narrative structure. Aspectual markers create the possibility of giving additional meaning to the action or situation expressed by the verb. Some languages offer the ability to use both tense and aspect, some languages only are able to differentiate between different tenses, and other languages almost uniquely use aspectual markers to express the relationship between the verb in question and the rest of the utterance. Another common device for expressing temporal relations within and between sentences is the use of temporal adverbials. In most languages, this is a productive linguistic device, used from a very young age onwards. The focus of this chapter is on these three aspects of temporal reference: tense, aspect and temporal adverbials. The degree of linguistic and cognitive complexity of these three aspects is not the same across languages. With respect to the acquisition of linguistic means for temporal reference, we can expect a possibly universal principle to account for the lack of complex temporal features in the language use of early learners. This principle, called the principle of natural order (Klein 1994: 45), states that if there are two subsequent related events, the reference to the earlier event is made first, i.e., the order of mentioned events corresponds with their order of occurrence. This means that, in the process of acquisition, learners of a language tend to state events in a chronological order, with no complex conjunction necessary. Learners will therefore prefer sequence (1) to sequence (2): 1. The frog was in the jar. The boy forgot to close the jar. The frog got out off the jar. 2. The frog was in the jar. He got out off the jar because the boy forgot to close the jar.

 Petra Bos

Apart from this possibly universal principle, there are language-specific aspects that play a role in the acquisition and use of temporal features by language learners. In this chapter we would like to investigate what principles drive young bilingual language learners when they are asked to retell a story, and, as a consequence, are forced to use temporal expressions. Will they adhere to different principles in L1 than in L2? And will their use of L2 be influenced by their knowledge of L1 (transfer) or will they adhere to universal language learning principles such as the principle of natural order?

Design The informants and the data that are used in this chapter, are derived from a large research into the development of bilingualism among Moroccan bilingual children in the Netherlands at school-age. Other topics that were treated in this research were: anaphoric reference, relative clauses (experimental data) and topic continuity (semi-spontaneous data). The group of informants consisted of bilingual Moroccan children living in the Netherlands. Their L1 was Moroccan Arabic and their L2 Dutch. They were aged 4 until 11 and the data were collected on a pseudo-longitudinal basis. Control groups consisted of their monolingual Dutch classmates and monolingual children living in Morocco (see Bos 1997). The data we will base ourselves on in this chapter, are transcripts with semi-spontaneous speech. Informants are bilingual children of the core group and children of the control groups. For this chapter we take a sub-set of the original data-set, i.e. the 5-, 7- and 9-year-olds. Of each group of 25 informants we have 25 transcripts per age, except for a few missing transcripts from Morocco. This presents us with a data-set of almost 300 transcripts.

Table 1. Number of transcripts involved in the analysis of temporality in Moroccan Arabic and Dutch Moroccan Arabic Group

Age 5

Dutch

7

9

5

7

9

Core group of bilinguals Moroccan control group Dutch control group

25 24 –

25 24 –

25 23 –

25 – 25

25 – 25

25 – 25

Total

49

49

48

50

50

50

Temporality in Moroccan Arabic and Dutch 

The data in the Netherlands were collected on a pseudo-longitudinal basis. For this subset this means that the 5-year-olds and the 7-year-olds are the same children. The 9-year-olds are part of the older group. The data in Morocco were collected cross-sectionally. The data consist of retellings of the famous Frog story by Mayer (1969). In this story a boy and a dog have a frog in a jar. The frog escapes and the boy and the dog go looking for the frog. The boy and the dog both encounter all kinds of adventures on their way and in the end they find a (the?) frog and many more other frogs.

Research questions In narratives, devices for anchoring tense and specifying aspectual features are almost always used. The extent to which this happens and the way in which it happens differs for each language user (especially young language learners) and the possibilities within each language. In the case of a retelling, as was performed by our informants with the frog story, the story-tellers are obliged to connect different actions and situations that are presented in the pictures and to put them in some kind of temporal frame. In order to make an analysis of temporality markers that our informants used in Moroccan Arabic and Dutch, the following questions were raised: 1. Are there any differences between the bilingual and monolingual children regarding the use of temporal markers in Dutch? And are there any developmental patterns to be observed? 2. Are there any differences between the bilingual and monolingual children regarding the use of temporal markers in Moroccan Arabic? And are there any developmental patterns to be observed? 3. Are there any universal developmental strategies and/or language-specific strategies that the children use? Can any influences of transfer be found? We will first look briefly into the grammatical systems of Moroccan Arabic and Dutch, in order to see what kind of temporal marking can be expected to be found in the data.

Moroccan Arabic In Moroccan Arabic there are no clear tense markers, while there are many aspect markers. Instead of referring to past or present, the different forms of

 Petra Bos

the verb in Moroccan Arabic refer to the incompletion (the prefixed form) or completion (the suffixed form) of an action or a situation. Incompleteness and completeness often refer to present and past, respectively, but not necessarily so. Four verb forms can be distinguished (cf. Caubet 1993a: 31ff). Different kinds of combinations of these forms are also possible, and will be specified at the end of this section. The four forms in question are: • • • •

bare prefixed form (used in marked cases); particle ka- (or ta-) + prefixed form (usually denoting incompleted action/ situation); suffixed form (usually denoting completed action/situation); active participle (usually denoting incompleted action/situation, but denoting completed action/situation if occurring in combination with a suffixed form).

The concepts prefixed and suffixed concern the conjugation of verbs for person. For the first form, the stem of the verb gets conjugated for the different persons by means of prefixations of phonemes. For the second form, the same conjugation is used, but with an additional prefixation of a particle, which is either ka- or ta-, dependent on regional varieties.1 The suffixed form is conjugated by means of suffixation of phonemes to the verb stem. We give an example of what the singular forms look like in the different conjugations by using the verb stem šreb (‘to drink’). The active participle has two forms in the singular: one for masculine and one for feminine. The meaning is comparable to that of the -ing form in English. Person

Prefixed form

ka + prefixed form

Suffixed form

Active participle

1 2fem 2masc 3fem 3masc

ne-šreb t-šerb-i te-šreb te-šreb ye-šreb

ka-ne-šreb ka-t-šerb-i ka-te-šreb ka-te-šreb ka-ye-šreb

šreb-t šreb-ti šreb-t(i) serb-at šreb-ø

šarba (fem) šâreb (masc)

The four forms presented above (and their combinations) are used to denote temporal aspect. As mentioned before, the prefixed form is used in very specific cases, as in proverbs, eventualities, vague future, orders or wishes and optatives. In such cases, the prefixed form usually refers to an incompleted action or situation, either concomitant or non-concomitant. It can also be

Temporality in Moroccan Arabic and Dutch 

used in combination with other verb forms in order to refer to modality, and in combination with different particles, denoting unmarked future and near future. It is used in proverbs, for potentiality/eventuality, vague future, order/wish, as optative, in circumstantial sentences, for unmarked future, near future. The combination of the particle ka- + prefixed form almost always refers to the incompleted action, either concomitant (comparable to the durative or progressive) or non-concomitant. For verbs of motion ka- + prefixed form is always non-concomitant. This form can also refer to aspects such as habitual, iterative, aorist and general truths. The suffixed form of the verb is commonly used for completed actions or situations that are non-concomitant, but also sometimes for completed actions that are concomitant. Other uses are pseudo-performatives, hypothetical clauses, concessive clauses, circumstantial clauses, and perfect. The active participle, finally, can be used as perfect/completed action or situation (concomitant), as actual/incompleted action or situation (concomitant), as aorist, as prospective (incompleted action, concomitant) and as past durative.

Dutch Whereas in Moroccan Arabic there are no clear tense markers and many aspect markers, Dutch shows the reverse picture: there are no clear aspect markers in Dutch morphology and many tense markers. This does not mean that there are no means in Dutch to link aspectual features to an action or situation, but they are not expressed by the verb form itself. The different categories for the different tenses are as follows: Present: simple present present perfect Past:

simple past past perfect

The perfect tenses are created by the combination of an auxiliary (either hebben, ‘to have’ or zijn, ‘to be’), either in the present or in the past tense, plus the past participle. To give the reader an idea, the forms for the singular are presented here. The infinitive is drinken (‘to drink’) and the verb stem is drink. The conjugation in other tenses than simple present is a so-called ‘strong’ one, which means that a vowel change takes place (in this case i → o):

 Petra Bos

Person

Simple present Present perfect

Simple past Past perfect

1 2 3fem 3masc

drink drinkt drinkt drinkt

dronk dronk dronk dronk

heb gedronken hebt gedronken heeft gedronken heeft gedronken

had gedronken had gedronken had gedronken had gedronken

Aspectual markers in Dutch are realised through the use of modal verbs, the use of compound verbs and prepositions: ik ben aan het lopen (I am on the walk = ‘I am walking’), ik ga lopen (I go walk, ‘I am going to walk’), ik begin alvast te lopen (I start already to walk = ‘I am going ahead, you catch up with me’), etc.

Anchor time We will first look at the anchor time in the retellings of the children, taking this as a starting point for comparison. Anchor time refers to the general tense pattern of the retellings. Most children adhere to the use of one tense for the whole retelling. For Moroccan Arabic, we looked at forms that either referred to completed actions or situations or forms that referred to incompleted actions or situations. We called these forms the accompli and the inaccompli respectively, copying the French terminology used by Caubet (1993b). For Dutch we looked at the past and the present tense forms used in the retellings. We decided that if over 80 per cent of the utterances made by an informant was made with the use of one kind of forms, that would be characterized as the anchor tense. If percentages were below 80 per cent, the retelling was put under mixed. For Moroccan Arabic this provides us with the categories ‘accompli’, ‘inaccompli’ and ‘mixed’ and for Dutch the categories are ‘past’, ‘present’ and ‘mixed’.

Moroccan Arabic For Moroccan Arabic, the transcripts of the bilingual core group and the Moroccan monolingual control group were taken into account. The anchor time was established on the basis of the above-mentioned 80 per cent criterion. The outcome is shown in Table 2.

Temporality in Moroccan Arabic and Dutch 

Table 2. Anchor time in Moroccan Arabic (N=25) Age 5 7 9

Bilingual core group

Monolingual control group

Inaccompli Accompli Mixed

Inaccompli Accompli Mixed

18 20 8

6 5 17

1 – –

17 12 9

3 12 14

4 – –

There were hardly any children that used only the inaccompli as the basic anchor time for their retellings. Only among the 5-year-olds in the bilingual group was this done by one child and in the monolingual group by four children. We then observe that the 9-year-olds all ended up with most stories having accompli as the anchor time, but for the monolingual children this trend seemed to start slightly earlier than for the bilinguals. The bilinguals adhered to mixed retellings longer than the monolinguals.

Dutch In Table 3 the results are shown in Dutch for the bilingual and the monolingual children. If we compare the Moroccan Arabic accompli with the Dutch past tense, we can say the same about Dutch as we did for Moroccan Arabic. Both groups of informants ended up with past tense reference as anchor time. But here the monolinguals made much more use of the present tense, whereas the bilingual children composed more mixed retellings. If we compare the development in the L1, Moroccan Arabic, of the bilingual group to their L2, Dutch, we see that the informants have highly comparable results. The only difference being the fact that in Moroccan Arabic the informants make less use of the inaccompli than they do of the present tense in Dutch. But in both languages they move towards a trend of using a

Table 3. Anchor time in Dutch (N=25) Age 5 7 9



Present

Past

Mixed

Present

Past

Mixed

9 7 3

5 6 20

11 12 2

14 17 9

4 5 15

7 3 1

 Petra Bos

verb form that represents the past (accompli and past tense), which can be seen as a universal trend in narratives.

Aspectual features Moroccan Arabic We have put all occurring verb forms in the retellings in Table 4. The classification presented was made on the basis of verb forms. We have seen that many functions can be attributed to the different verbs, depending on the context of the sentence. It is impossible to determine the exact aspectual function of each verb form, because this would mean going over more than 10,000 verb forms with a group of native speakers to determine the exact aspect of the verb. Unfortunately this cannot be dealt with in the context of this chapter. We decided to make a division of prefixed, ka+ prefixed and suffixed forms and then to attach a specific, inaccompli and accompli aspect to them respectively. The same holds for active participle (inaccompli) and suffixed form + active participle (accompli). For the modal verbs, we also had as a basis suffixed form (accompli), prefixed form (inaccompli) and g˙adi (future particle). From Table 4, we can see that the monolingual children made more use of the verb forms that have a specific aspectual meaning (bare prefixed form) than the bilingual children. The bilingual children mostly used the ka+prefixed form Table 4. Verb forms in Moroccan Arabic retellings (N=25) Bilingual core group Age

5

‘Tense’ Prefixed (inacc) Ka+prefixed (inacc) Suffixed (+ka+prefixed) (acc)

7

35 16 267 218 660 538


9

5

7

9

16 62 741

70 193 427

47 136 579

44 108 629

Active participles Active participle (inacc) Suffixed+active paticiple (acc)

52 6

26 6

16 8

87 16

66 33

53 36

Modality Prefixed+(ka+)prefixed (inacc) Suffixed+(ka+) prefixed (acc) G˙adi+(ka+) prefixed (fut)

3 40 9

3 28 2

54 -

15 117 13

18 139 5

8 186 6

1,072 837

897

938

Total

1,023 1,070

Temporality in Moroccan Arabic and Dutch 

or the suffixed form (with or without the combination with a prefixed form). For the active participle the same can be said. It has certain special functions that cannot always be expressed by the prefixed or suffixed form and can therefore be an important device for narrators. The monolingual children made much more use of it (either in a bare form or in combination with a suffixed form) than the bilingual children did. The same can also be said for the use of modal verbs, whether in combination with a prefixed form, a suffixed form or the future particle: the monolinguals used them much more often than the bilinguals.

Dutch For Dutch we have constructed a similar table as we did for Moroccan Arabic, in which the different tenses of the verb forms are presented. Although there is no real aspect in Dutch, it is possible to attach aspectual meanings to verb forms with the aid of devices such as modal verbs, compound verbs and prepositions. Other ways of expressing, for example, a durative action, emerged, such as those in examples (1) and (2): (1) toen gingen ze zoeken zoeken zoeken zoeken. then they went to search search search search. (Iliass, bilingual Moroccan boy, 7 years old, in Dutch)

Table 5. Verb forms in Dutch retellings (N=25) Bilingual core group


Age

5

7

9

5

7

9

Tense Present Past Present perfect Past perfect

365 163 64 44

370 257 38 54

128 613 21 46

513 221 16 18

662 250 21 26

296 447 15 34

‘Aspect’ aan het /Present aan het /Past gaan /Present gaan /Past

12 5 112 172

14 2 34 202

2 6 7 161

11 8 81 35

20 3 45 45

6 14 21 89

14 8

13 12

2 28

18 11

22 11

7 28

959

996

1,014

932 1,105

957

Modality Present Past Total

 Petra Bos

(2) toen ging ie roepen roepen. then he went to call call. (Iliass, bilingual Moroccan boy, 7 years old, in Dutch)

We have, however, limited ourselves in Table 5 to the verb forms discussed earlier, including also tense and modality, just as in the previous section. We see that for the so-called aspectual markers there was a clear difference between past and present references. The same holds for the modal forms: over time, there was a decrease of present and an increase of past tense reference. There was not such a large difference between the monolingual and the bilingual informants. For the basic forms the same can be said, although the bilinguals made more use of the past than the monolinguals, who made relatively more use of the present tense as anchor time. The table also shows that the bilingual children made much more use of gaan-present(+infinitive) or gaan-past(+infinitive) than the monolingual children. It is not clear if these occurrences should be seen as an expression of inchoative aspect. For instance, the same ‘overuse’ has been observed by De Ruiter (1989) and he made the very plausible suggestion that this might be a strategy to avoid derivations of the main verb. We have to state that because of the large differences between Moroccan Arabic and Dutch in the field of tense/aspect, it is hard to make comparisons, but some comparisons can be made on the basis of the strategies the children seemed to use. If we compare the way the bilingual children performed in their L1 to the way they performed in L2, the following can be noticed. First, we see that the bilingual children overgeneralize the use of the Dutch verb gaan (‘to go’) in combination with another verb form. On the other hand we see that they underuse the use of modality verbs in Moroccan Arabic. Therefore, it seems that in Dutch, in order to avoid difficult grammatical verb forms, they overuse a certain construction and in Moroccan Arabic, again probably to avoid less frequent and therefore more difficult verbs, they just avoid to use these forms at all. The second point we noticed, is the fact that a large majority of the verb forms the children use in Moroccan Arabic is somehow denoting the accompli. Whereas in Dutch the children first use more present forms than past forms and at the age of 9 this has changed in that they use more past forms than present forms. This should probably not be seen as a specific feature of bilingual development, but as a language specific feature of Moroccan Arabic and Dutch respectively. The fact that the monolingual children (both Moroc-

Temporality in Moroccan Arabic and Dutch 

can and Dutch) show the same patterns, can be seen as evidence that this should be considered as a language-specific feature.

Temporal adverbials Temporal adverbials are used to link sentences together temporally, or to make a temporal link within one sentence. Some of them are widely used in story telling, others are more complicated and therefore less used. For instance, the word combination ‘and then’ at the beginning of a sentence can be used to connect a whole series of sentences, provided they are told in a chronological way. By means of the available computer program, we sampled out all the sentences that contained a temporal adverbial or a temporal conjunction. The only temporal adverbials found in the texts were either positional adverbials, indicating that the action took place after or at the moment of speech (‘and then’, ‘thereafter’, ‘now’), or contrastive adverbials (such as ‘not yet’, ‘still’, ‘just’). No instances of positional adverbials referring to an action before the moment of speech (‘before’) were used by our informants, nor frequentative adverbials (‘always’, ‘never’, ‘often’), nor durational adverbials (‘until’, ‘in’, ‘during’) (cf. for an elaborate description of categories: Starren 1996).

Moroccan Arabic As there were no durative or frequentative temporal adverbials in our data-set, we singled out all occurrences of positional and contrastive temporal adverbials. These are presented in Table 6. We see that there were quite some differences between the bilingual group and the monolingual group. We will discuss them one by one. The temporal device men be,d (‘after that’ / ‘and then’) was hardly used by the bilingual group. The monolingual group used this device much more frequently, but not to an extent that it could be seen as their basic connector for a retelling with a chronological order. The bilingual children used all kinds of variants of sa,(intranslatable particle, most probably related to sa,a: ‘hour’, ‘moment in time’) for this purpose, like sa,tek, sa,antek, sa,anti, etc. Of these words, most variants could not be identified by native adult speakers of Moroccan Arabic. The monolingual children also used these forms, but to a much smaller extent. In the data-set of the bilingual children, there are frog stories in which each sentence

 Petra Bos

Table 6. Occurrences of temporal adverbials in Moroccan Arabic transcripts (N=25) Bilingual core group


5

7

9

5

7

9

Positional (men) be,d sa,ha ra iwa ,awed ,ad

1 142 17 – – 54 77

– 81 2 – 1 25 41

14 62 3 3 37 10 24

20 5 92 42 53 – –

2 63 238 48 19 – –

15 14 47 7 27 – –

Contrastive ma-zal baqi ,awed ,ad

– 1 – –

– 7 – –

– 2 – –

3 1 6 2

1 1 2 6

2 3 5 1

starts with a variant of sa,. For the monolingual data-set this is not the case. In the monolingual data-set, we found another particle that fulfilled the role of much used chronology marker, i.e., ha. The particle ha (as well as the particle ra-), however, indicates that the moment of speech and the topic time coincide (cf. Caubet 1994a). This is in contrast to sa,, which puts the topic time before the moment of speech. We present a few examples of utterances with these adverbials: (3) men beεd [/] men be,d dik ž-žru tah . u men be,d therrsat l-u [//] herrsat l-u ž-žaža dyal-u. u men be,d dik l-,ayel qebt-u. then [/] then that dog fell. and then his glass broke (for him) [//] broke (for him). and then that boy grabbed him. (Siham, bilingual Moroccan girl, 9 years old, in Moroccan Arabic) (4) u sa,tek ž-žru ka-yet,elleq fe-š-šežra. u sa,tek n-nh el xeržu. u sa,tek eh herbu. and then the dog is hanging in the tree. and then the bees came out. and then er they fled. (Ouidan, bilingual Moroccan boy, 9 years old, in Moroccan Arabic) (5) l-weld ha huwa n,es. l-kelb ha huwa fuq-u. the boy(, there he) is sleeping. the dog(, there he) is on top of him. (Sami, monolingual Moroccan boy, 5 years old, in Moroccan Arabic)

Temporality in Moroccan Arabic and Dutch 

The 9-year-old bilinguals started to use iwa (‘thus’) as a connector, whereas the younger bilingual children did not use this device at all. The monolingual children used it from a young age onwards: (6) iwa u bqa yg˙uwwet. iwa u tah u huwa u kelb-u. iwa u had l-g˙zala teyyh athum. iwa u tah u fe-l-ma u g˙erqu. well, and he started to scream. well, and they fell, he and his dog. well, and this deer made them fall. well, and they fell into the water and drowned. (Amina, monolingual Moroccan girl, 5 years old, in Moroccan Arabic)

We found that contrastive temporal adverbials, such as ma-zal (‘still’), baqi (‘yet’), ,awed (‘again’) and ,ad (‘just’) are less used by the bilinguals than by the monolinguals. Examples are: (7) u lqa ,awed wah ed axur. and again he found another one. (Ahlam, monolingual Moroccan girl, 5 years old, in Moroccan Arabic) (8) ha huwa ,ad taleε here he is just (starting to) climb up. (Kawtar, monolingual Moroccan girl, 7 years old, in Moroccan Arabic)

For the bilinguals, we see an overgeneralized use of ,ad and ,awed. These words, which usually mean ‘just’ and ‘again’ respectively, seem to be used by the bilinguals as a sort of chronological marker, as in: ‘and then this happened, and then that’ (identified by Caubet as markers in retellings: “série d’événements à l’aoriste du récit’’ 1994b: 177). This is why we put ,ad and ,awed under “contrastive’’ for the monolinguals, but under “positional’’ for the bilinguals because that is the function they seem to give to these words. This holds to a lesser degree for ,awed than for ,ad, by the way. Especially the 9-year-old bilinguals seemed to start to use ,awed in the same way as the monolinguals do. Here, too, we present some examples: (9) u ,ad xelliw š-šeržem meh lul. ,ad ka-yfettšiw u ka-yfettšiw u ka-yfettšiw. u ,ad huma ma-žebru-ha-ši. and then they left the window open. then they were looking and looking and looking. and then they did not find her. (Iliass, bilingual Moroccan boy, 7 years old, in Moroccan Arabic)

 Petra Bos

(10) ,awed wah ed l-me,za hezzat-u. ,awed bqat [//] kanet [//] &kat bg˙at teyyh u fe-l-ma. u ,awed tah . ,awed lqaw hadik kikker (= Dutch for frog). then a goat picked him up. then she started [//] went [//] &to wanted to make him fall into the water. and then he fell. then they found that frog. (Bilal, bilingual Moroccan boy, 7 years old, in Moroccan Arabic)

The adverbials mentioned in the above section can all be used at the beginning of a simple clause. If we look at conjunctions that cause the use of subordinate clauses (h it, mh it, li’anna, (,la) h eqqaš: ‘because’, baš: ‘in order to’), we see that such conjunctions were less used by the bilinguals than by the monolinguals, and this goes especially for baš (see Table 7). Table 7. Occurrences of conjunctions. that cause subordinate clauses in Moroccan Arabic transcripts (N=25) Age h it mh it li‘anna (,la) h eqqaš baš



5

7

9

5

7

9

– 1 – – 5

– – – – 1

– – – – 6

– – 1 – 19

– – – 2 16

4 – – – 14

Example: (11) u l-kelb bg˙a yetleε mhi t xaf men hadik eh eh l-far. and the dog wanted to climb because he was afraid of that er er mouse. (Nahid, bilingual Moroccan girl, 5 years old, in Moroccan Arabic)

There is another group of temporal markers that we can take into consideration. These are markers of simultaneity (melli, menni, mnin: ‘while’, ‘when’). They require coordination of clauses. Table 8. Occurrences of markers of simultaneity in Moroccan Arabic transcripts (N=25) Bilingual core group Age melli menni mnin


5

7

9

5

7

9

1 – 2

– 1 –

6 – –

5 – –

13 – 2

6 2 14

Markers of simultaneity are also made more use of by monolingual children

Temporality in Moroccan Arabic and Dutch 

than bilingual children. The monolingual data-set contains 42 occurrences of markers of simultaneity as opposed to 10 in the bilingual data-set. Examples: (12) u mnin faqu tellu ,la l-bwata lli kanet fe-ha ž-žrana. and when they woke up, they looked at the box in which the frog had been. (Youssouf, monolingual Moroccan boy, 9 years old, in Moroccan Arabic) (13) mnin nad u ma-lqaw-š d -d efd a,a. when they got up they did not find the frog. (Laila, monolingual Moroccan girl, 7 years old, in Moroccan Arabic) (14) ha huwa melli tah ha huwa ga, therres hadak. here when he fell he completely destroyed that (=jar). (Kawtar, monolingual Moroccan girl, 7 years old, in Moroccan Arabic)

On the whole, we can state that the monolingual informants did not use these complex markers very often, but they did use them more frequently than the bilinguals. Adults are expected to use more of these complex markers (contrastive adverbials, subordinate conjunctions, markers of simultaneity), and the observed phenomenon can be seen as a developmental feature. The bilingual children used these complex markers even less than the monolingual children, which can be seen as a consequence of the slower pace at which they seem to develop throughout the period of acquiring two languages.

Dutch We now turn to the Dutch data-set of both the bilingual core group as well as the monolingual control group. Table 9 presents the occurrences of temporal adverbials in both Dutch data-sets. Also in these data-sets, we found only positional and contrastive adverbials and no durative or frequentative ones. The first thing that attracts the attention is the apparent overgeneralization of toen (‘then’) by the bilinguals and the underrepresentation of dan (‘then’) where monolinguals used both toen and dan. Also hardly any occurrences of daarna (‘thereafter’) and nou (‘now’) can be found in the data-set of the bilinguals. This was also the case in Moroccan Arabic. The bilingual children seemed to make more use of general markers of chronology than that they drew the attention to whether two actions had a sequential order or that the moment of speech and topic time coincided; see examples (15)–(18).

 Petra Bos

Table 9. Occurrences of temporal adverbials in Dutch transcripts (N=25) Age



5

7

9

5

7

280 38 5 –

307 38 – 1

216 1 – 1

157 187 80 3

147 218 193 52 22 7 8 3

1 1 – 10

4 – 2 3

4 – 2 3

8 2 3 1

9

Positional (en) toen (en) dan nou daarna Contrastive nog nog niet nog steeds bijna

12 1 1 3

14 1 8 8

(15) en toen wou tie op een steen klimmen. toen was tie op een steen klimt. toen was die hond ook klimt. and then he wanted to climb on a stone. then he had climbed on a stone. then that dog had climbed too. (Khaled, bilingual Moroccan boy, 5 years old, in Dutch) (16) en dan vallen ze in het water. en dan horen ze de kikker. en dan gaan [/] gaan ze over de boom. en dan zien ze hem. en dan nemen ze weer die kikker mee. and then they fall into the water. and then they hear the frog. and then they go [/] go over the tree. and then they see him. and then they take that frog with them. (Mouhcin, bilingual Moroccan boy, 7 years old, in Dutch) (17) en nou die jongen is wakker. and now that boy is awake. (Tarik, bilingual Moroccan boy, 5 years old, in Dutch) (18) daarna gaat het jongetje in een boom kijken. after that the boy goes to look in a tree. (Marieke, monolingual Dutch girl, 7 years old, in Dutch)

The unmarked way of using toen in Dutch is in utterances with past tense reference and the unmarked way of using dan is in utterances with non-past tense reference. The fact that the bilinguals used toen much more often also

Temporality in Moroccan Arabic and Dutch 

relates to the fact that the anchor time of their stories was mainly mixed or past; for the monolinguals this was present and past. It is therefore logical that the monolinguals made more use of adverbials that work in combination with sentences in present tense than the bilinguals. However, it is very plausible that this is not the only factor involved here. Results of another approach are presented in Table 10. This table displays the number of occurrences of combinations of toen in present-tense sentences and dan in past-tense sentences. This table must be read as follows: in the data-set of the 5-year-old bilingual group, there were 280 occurrences of toen, 36 of which were in combination with an utterance in present tense. This is 13 per cent, as opposed to 3 per cent in the data-set of the 5-year-old monolingual group. Table 10. Occurrences of ‘‘present tense plus toen’’ and ‘‘past tense plus dan’’ in Dutch transcripts (% in parentheses) (N=25) Bilingual core group Age toen + Present dan + Past


5

7

9

5

7

9

36/280 (13) 12/38 (32)

23/307 (8) 4/38 (10)

1/261 (0) 0/1 (0)

5/157 (3) 6/187 (3)

8/147 (5) 4/193 (2)

3/218 (1) 3/52 (6)

We see that all the percentages of the non-standard devices toen + present tense and dan + past tense were higher for the bilinguals than for the monolinguals (apart from the 0 per cent for 9-year-olds in dan + past, but there was only one occurrence), but this does not account for the great differences between toen and dan in Table 9. We therefore have to conclude that the bilingual children overgeneralized toen as a clause linker at the beginning of the sentence. It seems as if they made less use of conjunctions that require subordination than monolinguals, a strategy we also witnessed for Moroccan Arabic. Examples: (19) en toen [/] toen trekt ie z’n kleren aan en z’n jas. and then [/] then he puts on his clothes and his coat. (Deborah, monolingual Dutch girl, 5 years old, in Dutch) (20) toen doet ie die raam open. then he opens that window. (Hassna, bilingual Moroccan girl, 7 years old, in Dutch)

 Petra Bos

(21) en dan keek Jan achter de boomstam. and then Jan looked behind the treetrunk. (Chet, monolingual Dutch boy, 9 years old, in Dutch) (22) dan ging tie wakker worden. then he went to wake up. (Samir, bilingual Moroccan boy, 5 years old, in Dutch)

In the case of contrastive adverbials (nog (steeds) ’still’, nog niet ‘not yet’, bijna ‘almost’), we see that also in this domain, the monolinguals made much more use of them than the bilinguals did, just as we saw for Moroccan Arabic. Examples are: (23) en eh die uil kwam nog steeds achter hem aan. and er that owl was still following him. (Said, bilingual Moroccan boy, 9 years old, in Dutch) (24) en toen viel hij bijna om. and then he almost fell over. (Yasmina, bilingual Moroccan girl, 9 years old, in Dutch)

There were many occurrences of bijna in the data-set of the 5-year-old bilinguals, but most of them were used with a non-standard meaning, as we can see in the following example, where the informant uses bijna in order to make a distinction between an action-in-progress and a completed action: (25) en deze is bijna gevallen, helemaal. and this one has almost fallen [=is falling], completely [has fallen]. (Oussama, Moroccan boy, 5 years old, bilingual, in Dutch)

For conjunctions that refer to a causal relation between clauses, the results are set out in Table 11. Table 11. Occurrences of conjunctions that cause subordinate clauses in Dutch transcripts (N=25) Age want omdat dus



5

7

9

5

7

9

4 – –

1 – –

6 3 13

13 2 –

16 – 2

19 4 8

Temporality in Moroccan Arabic and Dutch 

The coordinating device want (‘for’) requires no inversion (such as verb– subject or auxiliary–past participle, depending on the syntax of the utterance), the subordinating device omdat (‘because’) does. This difference in complexity might account for the difference between the number of occurrences of both conjunctions. Dus means ‘therefore’ or ‘so’. There seems to be an overgeneralized use of dus for the bilingual 9-year-olds. For the rest, not many of these complex conjunctions can be found in the data-set of the bilinguals. There were more occurrences in the data-set of the monolinguals, although these numbers were not substantial either. We present some examples: (26) en die hond hij kan niet heel hoog springen want hij is nog klein. and that dog he cannot jump very high because he is still small. (Hassna, bilingual Moroccan girl, 7 years old, in Dutch) (27) en toen viel die naar beneden omdat dat dier kwaad was. and then that one fell down because that animal was angry. (Dimmy, monolingual Dutch boy, 5 years old, in Dutch)

Markers of simultaneity also occur in the Dutch data-sets, as can be seen in Table 12. Table 12. Occurrences of markers of simultaneity in Dutch transcripts (N=25) Age toen . . . (toen) terwijl



5

7

9

5

7

9

2 –

1 –

11 –

1 –

3 –

7 2

Most occurrences are of the kind toen . . . (toen) (‘when . . . (then)’) and there are only two occurrences of terwijl (‘while’). There were few differences between the bilingual and monolingual group. We see that the use of these complex markers was quite rare in both data-sets. Examples are: (28) toen dat hert weg &ree rende, hangde hij nog aan dat gewei. when that deer &dro ran away, he was still hanging in those antlers. (Shirley, monolingual Dutch girl, 9 years old, in Dutch) (29) toen hij slaapt toen ging de aap weglopen. when he sleeps then the monkey went to walk away. (Iliass, bilingual Moroccan boy, 5 years old, in Dutch)

 Petra Bos

(30) toen ging ie [//] de hond in de [/] in de pot [/] in de pot zoeken terwijl de jongen zich aan zit te kleden. then went he [//] the dog to search in the [/] in the jar [/] in the jar while the boy sits to dress himself. (Danny, monolingual Dutch boy, 9 years old, in Dutch)

As a rule, the bilinguals made more use of temporal adverbials that are used at the beginning of simple sentences, whereas the monolinguals made more use of conjunctions that demand more complex sentences or even subordination. These results fit in with the concept of the Basic Variety, proposed by Dietrich, Klein and Noyau (1995: 15), with the difference that adult L2-learners may fossilize at this point, whereas for young L2-learners this usually is not the case.

Conclusions and discussion In this section we will sum up the differences and similarities found for the different developmental aspects of temporal reference. If we look at the use of tense and aspect, we see that all young language learners, whether in their first or second language, slowly change from a child-like performance to an adultlike performance.

Anchor time In both languages under consideration, the standard way of (re)telling stories by adults seems to be by means of reference to actions that have happened in the past. As regards tense this would generally mean the use of the accompli in Moroccan Arabic and the use of past tense in Dutch. We suppose that our informants are on their way to standard adult performance in their respective languages, therefore their usage of time anchoring will be developing towards this adult-like distribution. The children of the younger age groups (age 5 and 7) in Moroccan Arabic mostly alternated the accompli and the inaccompli in their retellings without clear motivation. The children of the older age group (age 9) showed a preference for the accompli, coming closer to the adult norm. This applied to both monolinguals and bilinguals. For Dutch, almost the same can be said: the younger children (age 5 and 7) often mixed tenses (the bilingual children) or used only present tense (the monolingual children) as anchors for their stories.

Temporality in Moroccan Arabic and Dutch 

The older children (age 9) preferred past tense (the bilingual children even more so than the monolingual children). We see that the 5- and 7-year-old monolingual children produced fewer stories with mixed tenses, which might indicate that they became more consistent in their use of tense at an earlier age than the bilingual children. This was true for both languages.

Aspectual features In Moroccan Arabic, for both bilingual and monolingual informants, there was a decrease in occurrences of the bare prefixed form. This is a form with very specific meanings and it is most probably overgeneralized by the younger children. Over time, the most frequently used forms were the unmarked prefixed and suffixed forms. The use of active participles decreased over time. The younger children used them for expressing the progressive aspect, whereas the older children used the ka- + prefixed form to express this meaning. We also see that the monolingual children had a much more even distribution of the use of different (combinations of) forms. Their capacity to use all (combinations of) forms that are available in their language seemed more in balance than that of their bilingual peers.

Temporal adverbials As regards the use of temporal adverbials, the bilingual children seemed to be less able than the monolingual children to establish subtle temporal relations that are characteristic for good story telling. They strongly overgeneralized the use of certain forms, usually the ones that do not require clause subordination or inversion of the basic word order. The monolingual children made more and more standard-like use of these complex adverbials, conjunctions and particles. This indicates that there is a difference between bilingual and monolingual children on a higher level, i.e., a level that does not show in everyday communication. This holds for both languages. The data point to the concept of a Basic Variety, discussed by Dietrich, Klein and Noyau (1995). The language production of these bilingual children seemed very fluent and nativelike when listened to. But when analysed, it turned out to be less sophisticated than that of monolingual children. In general, all children strongly adhered to the principle of natural order for reporting events. Not many children used complex retelling devices to

 Petra Bos

move back and forth in reporting the events of the frog story. Part of this is caused by the fact that the children told the story picture by picture while the pictures were lying in front of them, in a chronological order. But even in scenes where it would be rather natural to either look back or forth in time, this was not done. This became very clear from the fact that there was not one occurrence of a positional adverbial referring to ‘before’ in the whole data-set.

Note . From this point on we will refer to this prefixation only by ka-, but this should always be read as ka- or ta-.

References Bos, P. H. F. (1997). Development of bilingualism. A study of school-age Moroccan children in the Netherlands. Studies in Multilingualism 8. Tilburg: Tilburg University Press. Caubet, D. (1993a). L’Arabe Marocain. Tome I. Phonologie et Morphosyntaxe. Paris-Louvain: Éditions Peeters. Caubet, D. (1993b). L’Arabe Marocain. Tome II. Syntaxe et Catégories Grammaticales, Textes. Paris-Louvain: Éditions Peeters. Caubet, D. (1994a). Deixis, aspect et modalité. Les particules hâ- et râ- en Arabe Marocain. Paris: INALCO, 139–149. Caubet, D. (1994b). La particule εâd en Arabe Marocain. Actes des premières journées internationales de dialectologie arabe de Paris. Paris: INALCO, 173–184. Dietrich, R., W. Klein and C. Noyau (1995). The acquisition of temporality. Amsterdam: John Benjamins. Klein, W. (1994). Time in Language. New York: Routledge. Mayer, M. (1969). Frog, where are you? New York: The Dial Press Ruiter, J. J. de (1989). Young Moroccans in the Netherlands. An integral approach to their language situation and acquisition of Dutch. Ph.D. thesis, University of Utrecht. Starren, M. (1996). Temporal adverbials as a blocking factor in the grammaticalization process. In: Proceedings of the CLS opening of the the academic year 1996–1997. Tilburg/Nijmegen: CLS.

Chapter 10

Bilingual narrative development in Papiamento and Dutch Ria Severing and Ludo Verhoeven

In the present study the production of narratives in Papiamento and Dutch of children by elementary school children on the island of Curaçao was examined. Curaçao is part of the overseas territory of the Netherlands and therefore has Dutch as the official language. Papiamento is the main language of communication and a Creole language, evolved from an Afro–Portuguese Pidgin spoken in West Africa, with new vocabulary acquired from Spanish and Dutch (Ferrol 1982; Maurer 1988). Papiamento-speaking children in Curaçao must thus acquire Dutch as a foreign language at school. The use of Papiamento in the classroom is extremely limited. Outside school the production and reception of Dutch tends to be very limited. Over the past decades, the frustration of the majority of the school children, their parents, and their teachers with regard to economic waste in education and the use of Dutch as the language of instruction has been greatly discussed. The limited usefulness of Dutch in the region and the low educational achievement of the children have also been discussed at length. Each year, roughly one quarter to one third of the elementary school population does not pass and the number of dropouts is alarming. A switch to the use of the mother tongue has still not occurred, however. One reason is widespread ambivalence with regard to the schooling system, which is clearly linked to the forms of higher education in the Netherlands. Papiamento is nevertheless now being officially taught in elementary school and has also been used illegally and unsystematically as the medium of instruction for quite some time. One might expect the historical colonial contempt of Papiamento by the economically and politically more powerful groups to be shared by the users of the language and to effect their use of the language as a consequence. Surprisingly, this is not the case. Although the vernacular is not completely standardized and codified, it has gradually

 Ria Severing and Ludo Verhoeven

become the major medium of communication in for newspapers, television, and radio. The movement towards (greater) independence for Curaçao has also been accompanied by promotion of the Creole language and a rejection of the Dutch language, as a symbol of the former colonial status of the island. Proficiency in Dutch, which is still the official language, no longer guarantees greater social prestige, although it does indicate a certain level of education and thus societal success. The most common languages in business and industry at present are English and Spanish. The position of the Dutch language is only strong in those domains where the language is still required legally, such as in court, official documents, and education. Since the eighties, the language policies have gradually changed. Papiamento is acquiring a more important position within the educational system and being standardized. As already mentioned, moreover, Papiamento was introduced as an official subject for elementary school instruction in 1986. Students are taught 30 minutes of Papiamento a day. For the remainder of the day, the students are taught in Dutch by teachers with Papiamento as their mother tongue in the vast majority of cases. This means that the children do not hear Dutch spoken by native speakers and, from a linguistic point of view, one might argue that the Dutch language of instruction is being increasingly colored by Papiamento, which may actually lead to a subvariety of Dutch. In order to gain greater insight into the narrative development of children in this postcolonial context, empirical studies of both L1 and L2 language acquisition are needed. The empirical data on the submersion of Caribbean children in a Dutch foreign language school curriculum show such submersion to not be very successful. More than 70% of the children do not succeed in finishing elementary school without class repetition of at least one grade. At the same time, the number of children being referred to schools for special education continues to grow. Studies of the language development of children learning both Papiamento and Dutch have recently been undertaken. In one study (Narain and Verhoeven,1994; Narrain 1995), both the Papiamento and Dutch language development of 80 children 4 to 6 years of age, living on the island of Curaçao, was examined. The language data were collected at three measurement points: at the beginning of kindergarten, after one year of kindergarten, and after two years of kindergarten. The relationship between language proficiency and various socio-cultural background characteristics was also examined. The results showed the development of Dutch to be much slower than the development of the L1, Papiamento, and the productive abilities in Dutch to develop

Narrative development in Papiamento and Dutch 

much later in particular. Another finding was that a number of the sociocultural background characteristics substantially predicted both the first- and second- language proficiency of the children. The prediction of first- language proficiency was nevertheless somewhat lower than the prediction of secondlanguage proficiency, which may be explained by the fact that the attitudes towards the home culture and L1 proficiency of most (first generation) Antillian families do -not substantially differ. In another recent study (Severing and Verhoeven 1995; Severing 1996), both the Papiamento and Dutch language proficiency of grade 5 children living in Curaçao were assessed. In order to do this, a number of language- comprehension tasks relating to the lexicon, syntax, -and semantics were used, on the one hand, and a worddecoding task on the other hand. The results showed the children’s level of language comprehension to be clearly better in Papiamento than in Dutch, while their level of decoding was clearly better in Dutch than in Papiamento. The degree of language-comprehension proficiency and decoding proficiency in the two languages was also found to be related to both sociolinguistic factors, as language contact and such background characteristics, as length of residence, and family size. The production of narratives in by children learning both Papiamento and Dutch has yet to be investigated. In the present study, the production of narratives in both Papiamento and Dutch by children 4 to 12 years of age living on the island of Curaçao will be explored. An attempt will be made to find an answer to the following questions. 1. What global differences in the Papiamento and Dutch narratives can be detected at different age levels? 2. What syntactic devices do the children use to build a narrative in the two languages? 3. And what devices do the children use for topic continuity in the narratives in the two languages?

Design of the study Subjects In selecting the informants, a more or less homogeneous sample was drawn from a group of children meeting the following criteria: Antillian by birth, Papiamento as their mother tongue, and middle-class background. A group

 Ria Severing and Ludo Verhoeven

of 102 children was randomly selected from the kindergarten (mean age 5 years, 2 months), grade 2 (mean age 8 years, 2 months), grade 4 (mean age 10 years, 4 months) and grade 6 (mean age 12 years, 3 months) classes in three schools. Children repeating the grade at the time of selection were excluded from the sample. The numbers of boys and girls in the three subsamples were more or less equal. Table 1 shows the number of subjects as a function of grade and sex. Table 1. Numbers of subjects as a function of grade and sex Boys Girls Total N

Kindergarten

Grade 2

Grade 4

Grade 6

9 9 18

14 16 30

7 19 26

11 17 28

Data collection Two separate sessions were undertaken for data collection: in Papiamento and Dutch with interval of two weeks between the sessions. All of the data collection took place in a separate room within the schools attended by the children in question. The children were given The Frog Story booklet, which they had never seen before. They were then asked to go through the booklet and prepare a story about the frog to be recorded for other children who did not know the story to listen to. The session only began after the child indicated that he or she was ready. To imitate the natural story-telling situation, the child was asked to sit down in front of the experimenter and keep the booklet closed during narration. In such a manner, it was also attempted to avoid the provision of simple description of the pictures in isolation and/or the use of deictic pointing rather than anaphoric references. However, the results of a pilot -study showed the kindergarten children to be unable to do this. The kindergarten children were therefore allowed to keep the book open on the table in front of them while telling the story. Both, the older and younger children were given as little feed-back as possible. Only when a very long period of silence occurred were the children encouraged to continue the story. All of the narratives were taped and transcribed for further analysis using CHILDES (MacWhinney 1991). Each text was divided into clauses, as defined by Berman and Slobin (1994).

Narrative development in Papiamento and Dutch 

Data-analysis Global text characteristics The number of monitoring devices used by the children in their narrative productions was first compared. Such devices as: corrections, slips of the tongue, repetitions, false starts, and restarts can be viewed as indicators of the monitoring behavior of children while telling stories (Verhoeven 1989). For each text, the percentage of the clauses containing one or more of these monitoring indicators was computed. It was predicted that the children would need to monitor their speech more in Dutch as a second language than in Papiamento (cf. Klein 1986), and that the degree of monitoring would generally decrease with age. The number of clauses in each text was taken to be a global measure of text length. It was predicted that the children would be able to tell longer stories in Papiamento, their mother tongue, than in Dutch and that the text length would increase as a function of age in both languages. The coherence of the children’s texts was also evaluated. In order to do this, the number of underlying propositions (with a maximum score of 28 for The Frog Story) was counted for each of the texts produced by the child (see Trabasso and Rodkin 1994). In addition, the overt expression of 15 major coherence relations between propositions was scored for each of the texts produced by the child; this included the expression of various cause-effect, cause-goal, problem-goal, problem-solution, situation-cause, and opposition relations. The internal consistency of these measures was found to be reasonably high. The Cronbach’s alpha for the number of propositions was .83 in Papiamento, and .88 in Dutch; the Cronbach’s alpha for the number of coherence relations expressed was .84 in Papiamento and .87 in Dutch. Syntactic devices The mean clause length was calculated as a global measure of the syntactic complexity of the narratives. It was predicted that the children produce longer clauses in Papiamento, their mother tongue, than in Dutch and that the mean clause length would increase with ages. The use of different types of conjunctions in the two languages was also evaluated. It was generally predicted that more conjunctions would be used as the child’s age progressed (cf. Flores d’Arcais 1978). The use of coordinate and subordinate conjunctions was also examined in the two languages. With regard to coordination, Papiamento uses I, anto, and ku (‘and’) for coordinate

 Ria Severing and Ludo Verhoeven

conjunctions, and pero, loga,and ma (‘but’) for adversative conjunctions. Dutch uses en (‘and’) for coordinate conjunction; maar (‘but’) for adversative conjunction; and want (‘for’) for causal conjunctions. With regard to subordination, the two languages both use a broad range of conjunctions. The most important ones for Papiamento are: ora, prom (‘before’), prom ku (‘until’), te, te ku (‘while’), te ora, tanten, mitras, si (‘when’) for temporal conjunctions; pasobra, pesei, dor ku, dor, pa (‘because’) for causal conjunction; o (‘or’) for comparison; si, ku (‘that’) for grammatical conjunctions; and si (if ’) for the expression of conditional relations. Dutch uses toen (‘then’), na (‘after’), voordat (‘before’), totdat (‘until’), terwijl (‘while’), als (‘if/when’), and wanneer (‘when’) for temporal conjunction; omdat (‘because’) for causal conjunction; dus (‘so’) for consecutive conjunction; om, opdat (‘so that’) to mark purpose; of (‘or’) as a comparative conjunction; and dat (‘that’) for grammatical conjunction.

Topic continuity devices The way in which the children referred to the main characters in The Frog Story (i.e., the boy and the dog) was also analyzed for the two languages. The references were divided into the following three categories (cf. KarmiloffSmith 1985, 1986). • • •

Introduction: the first mention of a referent. Maintenance: maintenance to the referential form in the previous clause. Switch: reintroduction of a referent not mentioned in previous clause.

The children could refer to the characters in their narratives using the following main set of referential devices. Linguistic device

Papiamento

Dutch

Full noun Indefinite determiner + nominal Definite determiner + nominal Demonstrative determiner + nominal Possessive pronoun + nominal Personal pronoun

Juan un kachó e mucha hòmber e sapu e su kachó mi e, el e, el nan

Jan een hond de jongen die kikker zijn hond ik hij zij zij

‘John’ ‘a dog’ ‘the boy’ ‘that frog’ ‘his dog’ ‘I’ ‘he’ ‘she’ ‘they’

Some important cross-linguistic differences should be noted at this point. Papiamento has a number of the features typical of many other Creole

Narrative development in Papiamento and Dutch 

languages. Nouns are not inflected (see 1 and 2). The realization of the plural always occurs by addition of the particle nan (see 3). (1) Mi ta buska mi sapu. (‘I am searching my frog’) (2) E sapu ta di mi. (‘The frog is mine.’) (Mi = I, my, mine) (3) Sapunan (‘Frogs’) Kasnan (‘Houses’) Muchanan (Children)

There is also a difference between the use of full noun phrases in Papiamento and Dutch. In Papiamento, reference to a specific noun or known object can be made using an NP without a definite or indefinite determiner. In similar cases in Dutch, use of a determiner is always required.

Results Global textual characteristics Table 2 presents the mean percentages clauses containing indicators of monitoring, and the mean number of clauses per narrative in Papiamento and Dutch for each grade. Table 2. Mean percentage clauses containing indicators of monitoring and mean number of clauses per text in Papiamento and Dutch for each grade

Mean % monitoring Mean number of clauses

Kindergarten

Grade 2

Grade 3

Grade 4

Pap. Dutch

Pap. Dutch

Pap. Dutch

Pap. Dutch

14% 10% 35 29

18% 40% 35 53

18% 49% 47 51

12% 23% 51 47

It can be seen that the use of monitoring devices in Papiamento is more or less constant the four grades. In Dutch, however, we find a sharp increase in the middle grades followed by decrease in grade 6. With respect to the number of clauses per narrative, it can be seen that the number is consistently higher for Papiamento than for Dutch, which shows the narratives to be longer in Papiamento than in Dutch across all grades. A sharp increase in text length occurred between grades 2 and 4 in both languages. Table 3 shows the mean number of propositions (maximum of 28) and mean number of coherence relations (maximum of 15) expressed in the Papiamento and Dutch narratives per grade. As can be seen, children express more Propositions and Coherence

 Ria Severing and Ludo Verhoeven

Table 3. Mean numbers of propositions and coherence-relations in Papiamento and Dutch narratives as a function of grade Kindergarten Grade 2 Mean propositions Papiamento 16.67 Mean propositions Dutch 12.33 Mean coherence relations Papiamento 6.17 Mean coherence relations Dutch 3.83

16.87 12.70 7.10 4.67

Grade 4

Grade 6

21.23 19.08 10.42 9.08

21.93 20.43 10.50 10.21

relations in their Papiamento narratives than in their Dutch narratives. It is thus clear that the child is capable of detecting and expressing the relevant propositions and coherence relations in the mother tongue but has trouble expressing them in Dutch as a second language. In both languages the children become more proficient with age. By grade 4, moreover, the differences between the two languages start to get smaller. In Figure 1, the results for a compound are presented: the average sum score for the number of content elements propositions and coherence relations for each language as a function of grade level. Analysis of variance revealed a significant grade by language interaction and also significant main effects of both grade and language. 35

Percentage

30

Papiamentu

25 20

Dutch

15 10 5 0 K

G2 Grade

G4

G6

Figure 1. Compound coherence measure for Papiamento and Dutch as a function of grade

In Table 4, the correlations between the different coherence scores for the two languages are displayed. It can be seen that the number of propositions and number of coherence-relations in Papiamento and Dutch are highly related. This means that children who are also typically able to tell a well-structured story in Papiamento are also able to do this in Dutch.

Narrative development in Papiamento and Dutch 

Table 4. Correlations between number of propositions and number of coherence relations expressed in Dutch and Papiamento narratives

Propositions Dutch Propositions Papiamento Coherence Dutch Coherence Papiamento a

Propositions

Coherence

Dutch

Papiamento

Dutch

Papiamento

1.00 0.73a 0.92a 0.68a

1.00 0.66a 0.93a

1.00 0.64a

1.00

p < .01

Syntactic devices in Papiamento and Dutch Table 5 presents the mean clause length in Papiamento and Dutch per grade. It can be seen that the average clause length in Papiamento is larger than in Dutch, and that the mean clause length progresses in both language as the children get older. The average clause lengths for the two languages were also highly related (r=.62, p

Narrative Development in a Multilingual Context (Studies in Bilingualism)

Aspects of Bilingualism in Wales (Multilingual Matters)

Sign Bilingualism: Language development, interaction, and maintenance in sign language contact situations (Studies in Bilingualism)

Beyond Narrative Coherence (Studies in Narrative)

Social Policy in a Development Context

Key Issues In Bilingualism and Bilingual Education (Multilingual Matters 35)

Bilingualism in Development: Language, Literacy, and Cognition

Bilingualism in Development: Language, Literacy, and Cognition

Beyond Bilingualism: Multilingualism and Multilingual Education (Multilingual Matters)

Studies in Japanese Bilingualism (Bilingual Education and Bilingualism, 22)

Narrative Interaction (Studies in Narrativity)

Selves and Identities in Narrative and Discourse (Studies in Narrative)

Language Attrition: Theoretical perspectives (Studies in Bilingualism)

Bilingualism and National Development

Bilingualism: Beyond Basic Principles (Multilingual Matters)

Beyond Bilingualism: Multilingualism and Multilingual Education (Multilingual Matters)

Multilingual Communication (Hamburg Studies in Multilingualism)

Comparisons in Human Development: Understanding Time and Context (Cambridge Studies in Social and Emotional Development)

Spanish in Four Continents: Studies in Language Contact and Bilingualism

Spanish in Four Continents: Studies in Language Contact and Bilingualism

Child Development in Context: Voices and Perspectives

Bilingualism in Society and School (Copenhagen Studies in Bilingualism, Vol 5)

Incomplete Acquisition in Bilingualism: Re-examining the Age Factor (Studies in Bilingualism, Volume 39)

Incomplete Acquisition in Bilingualism: Re-examining the Age Factor (Studies in Bilingualism, Volume 39)

Research Interviewing: Context and Narrative

Targeting Development (Routledge Studies in Development Economics)

Identity in Narrative: A Study of Immigrant Discourse (Studies in Narrative, V. 3)

Bilingualism in Society and School (Copenhagen Studies in Bilingualism, Vol 5)

Menander: A Rhetor in Context

Concept Development Studies in Chemistry

Narrative Development in a Multilingual Context (Studies in Bilingualism)

Aspects of Bilingualism in Wales (Multilingual Matters)

Sign Bilingualism: Language development, interaction, and maintenance in sign language contact situations (Studies in Bilingualism)

Beyond Narrative Coherence (Studies in Narrative)

Social Policy in a Development Context

Key Issues In Bilingualism and Bilingual Education (Multilingual Matters 35)

Bilingualism in Development: Language, Literacy, and Cognition

Bilingualism in Development: Language, Literacy, and Cognition

Beyond Bilingualism: Multilingualism and Multilingual Education (Multilingual Matters)

Studies in Japanese Bilingualism (Bilingual Education and Bilingualism, 22)

Narrative Interaction (Studies in Narrativity)

Selves and Identities in Narrative and Discourse (Studies in Narrative)

Language Attrition: Theoretical perspectives (Studies in Bilingualism)

Bilingualism and National Development

Bilingualism: Beyond Basic Principles (Multilingual Matters)

Beyond Bilingualism: Multilingualism and Multilingual Education (Multilingual Matters)

Multilingual Communication (Hamburg Studies in Multilingualism)

Comparisons in Human Development: Understanding Time and Context (Cambridge Studies in Social and Emotional Development)

Spanish in Four Continents: Studies in Language Contact and Bilingualism

Spanish in Four Continents: Studies in Language Contact and Bilingualism

Child Development in Context: Voices and Perspectives

Bilingualism in Society and School (Copenhagen Studies in Bilingualism, Vol 5)

Incomplete Acquisition in Bilingualism: Re-examining the Age Factor (Studies in Bilingualism, Volume 39)

Incomplete Acquisition in Bilingualism: Re-examining the Age Factor (Studies in Bilingualism, Volume 39)

Research Interviewing: Context and Narrative

Targeting Development (Routledge Studies in Development Economics)

Identity in Narrative: A Study of Immigrant Discourse (Studies in Narrative, V. 3)

Bilingualism in Society and School (Copenhagen Studies in Bilingualism, Vol 5)

Menander: A Rhetor in Context

Concept Development Studies in Chemistry

Recommend Documents