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 1040 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

AUTHOR ""

TITLE "Narrative Development in a Multilingual Context"

SUBJECT "Studies in Bilingualism, Volume 23"

KEYWORDS ""

SIZE HEIGHT "220"

WIDTH "150"

VOFFSET "4">

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.





Ludo Verhoeven and Sven Strömqvist

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

Development of narrative production in a multilingual context

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





Ludo Verhoeven and Sven Strömqvist

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

Development of narrative production in a multilingual context

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-





Ludo Verhoeven and Sven Strömqvist

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

Development of narrative production in a multilingual context

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.





Ludo Verhoeven and Sven Strömqvist

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.

Development of narrative production in a multilingual context

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.





Ludo Verhoeven and Sven Strömqvist

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.

Development of narrative production in a multilingual context

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.

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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.

Norwegian–English 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)

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

(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.

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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-

Norwegian–English children’s narratives

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%)

Norwegian–English children’s narratives

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.

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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

Norwegian–English children’s narratives

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

(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,

L2 narratives of Turkish adolescents in Norway

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.

L2 narratives of Turkish adolescents in Norway

(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.

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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:

L2 narratives of Turkish adolescents in Norway

(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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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

L2 narratives of Turkish adolescents in Norway

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.

L2 narratives of Turkish adolescents in Norway

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

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

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

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

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

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

(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.

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

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

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

(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 (p<.05). In the table, signiﬁcant over- and underTable 2. The use of connective adverbs and topic place-holders Swedish L1 Preschool (Age: 6) N: Connective adverb sen ‘then (seq.)’ så ‘so;then’ då ‘then (sim.) först ‘at ﬁrst’

12 n 12 12 12 4

f 141 146 91 7

Topic place-holder så då

12 3

School, grade 1 N: Connective adverb sen ‘then (seq.)’ så ‘so; then’ då ‘then (sim.)’ först ‘at ﬁrst’

11 n 11 11 11 7

23 n 20 11 13 12

p <0.05 f 574 34 95 34

62 7

f 204 149 291 24

11 1

118 1

% 78 5 13 5

OVER UNDER ns ns

737 11 2

% 31 22 44 4

19 n 19 9 16 10

668

total Topic place-holder så då

% 37 38 24 2

385

total

Signiﬁcant over/underrepresentation

Swedish L2

32 2

f

% 724 81 76 8 59 7 36 4

UNDER ns

OVER UNDER UNDER ns

895 9 1

67 1

UNDER ns

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

representation of an element in the data from the L2 children (in relation the the L1 data) is shown as over and under. What is particularly remarkable is the constant but rather slow development towards the L1 usage that can be observed in the data from the L2 learners. In the last recording, all L2 learners have mastered all of sen/så/då at least in principle (with one exception for så). In spite of that, the proportions between the connective adverbs is still not quite native-like after four years (at the group level; some individual L2 learners, however, seem to have acquired a native-like pattern in this subystem.) Sen is still over-represented, even if Table 2. (cont.) Swedish L1 School, grade 2 N: Connective adverb sen ‘then (seq.)’ så ‘ so; then’ då ‘ then (sim.)’ först ’at ﬁrst’

11 n 11 11 11 6

Topic place-holder så då

11 1

School, grade 4 N: Connective adverb sen ‘then (seq.)’ så ‘so; then’ då ‘then (sim.)’ först ‘at ﬁrst’ annars ‘otherwise’

11 n 11 11 11 9 4

18 n 18 17 18 8

310 1

f 381 775 429 41 5

11 9

333 30

f 1,297 572 235 23

Signiﬁcant over/underrepresentation

% 61 27 11 1

OVER ns UNDER UNDER

2,127

% 23 48 26 3 0,3

13 0

132 0

18 n 18 17 18 8 5

f 1,253 704 397 33 5

1,631

total

a

594 577 413 16

% 37 36 26 1

1,600

total

Topic place-holder så då

f

Swedish L2

UNDER

% 52 29 17 1 0,2

OVER (UNDER)a UNDER UNDER ns

2,392 16 12

269 45

ns ns

0.0561

Note : N=the total number of children in the group (with some dropping out over time); n=the number of children who use a certain element; f=the frequency of occurrence of the element; %=percentage in relation to the total number of connective adverbs



 Åke Viberg

there has been a steady decrease in the proportion of sen from grade 1 through grade 2 and up to grade 4. At the discourse level, development can in several respects only be observed as a slow change towards the native norm in terms of proportions with which diﬀerent options are chosen within various subsystems of the target language. This can also be seen in the slow motion of the usage pattern of så as a topic place-holder towards the native norm as represented by the L1 children.

Some aspects of lexical development The core of a narrative consists of a sequence of events. Lexically, events are primarily realized as verbs, which motivates the focus on this word class in this section. One important characteristic of verbal semantic ﬁelds is that they univerally tend to be organized around one or two nuclear verbs, which are typologically unmarked and tend to have the same basic meaning in a wide range of languages. Some of the most important of the nuclear verbs have meanings such as ‘go’ (ﬁeld: Motion), ‘make’ (Production), ‘give’, ‘take’ (Possession), ‘say’ (Verbal communication), ‘see’ (Perception), ‘know’, ‘think’ (Cognition). Lexical development has been treated rather extensively in relation to the learners in the early school-age project. One of the major ﬁndings was that for a very long time the L2 learners tended to favour nuclear verbs (Viberg 1993a 1999). The conclusion was based on a comparison of the frequency of occurrence of individual verbs in the L2 learner data and the L1 control data. Certain nuclear verbs were still overused even after four years. The narratives make it possible to approach this question from another direction, since it is possible to identify certain meanings that more or less all children try to express in some way. For example, the high point of the sandpit story, where the boy ﬂies away with the balloons, is expressed in one way or another by all of the informants in grade 4 in the bilinguals at school project. Table 3, taken from the Lindberg, Juvonen, and Viberg (1991) study, shows the realizations in both Swedish and Finnish. In Swedish, the major alternative is ﬂyga ‘ﬂy’. Among the monolingual speakers, a few more colourful expressions appear as alternatives such as stiga till väders ‘go up in the air’. Flyga ‘ﬂy’ is the most common alternative even among the bilingual informants, but among these there is also a clearly L2-related alternative, namely, gå ‘go’, one of the two nuclear motion verbs. In this case, the verb is used with a clear semantic overextension, since

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

Table 3. The choice of motion verb in ‘‘The boy ﬂew away with the balloons’’ Students in grade 4 Swedish Flyga ‘ﬂy’ Gå ‘go (up)’ Other Finnish Lähteä ‘leave’ Lentää ‘ﬂy’ Mennä ‘go’ Nousta ‘rise’ Other a

Same as above

b

Bilingual

Monolingual

N=20 13 5 2 N=20 a 8 8 4 – –

N=10 6 – 4 N=10b 6 – – 3 –

Adults N=10 7 – 3 N=20 7 3 – 1 9

One student does not try to express his opinion

Swedish gå only can be used in the context of going by foot with a human subject. In Finnish, the most frequent alternative is lähteä ‘leave’, a verb without any idiomatic equivalent in Swedish. Even if it is possible to ﬁnd a translation equivalent, Swedish would usually take recourse to other means of expression (such as the particle iväg ‘away’). The verb lentää ‘ﬂy’ only holds second position in Finnish. (I feel the material is too limited to justify a claim that the relatively higher proportion of this verb in the material from the bilingual students represents an inﬂuence from Swedish, even if this suggestion has some general plausibility.) What is more remarkable is that four of the bilingual students use mennä ‘go’, the nuclear verb in Finnish which is their stronger language. This usage seems to represent a simpliﬁcation similar to an L2-related feature even if it appears in their L1. According to the native Finnish researcher (Päivi Juvonen), the bilingual Finnish students in general tended to sound native-like in Finnish with the exception of some subtle diﬀerences. In the material from the early school-age project (Viberg 1993a, 1998a, 1999), it is possible to observe an age-related progression towards a greater stylistic ﬂexibility in the lexical choices of the L1 children. In the car-accident story, for example, it is possible to isolate one event that is expressed by almost all the children, namely how the ambulance personnel put a band-aid on the tree that was hit by the car. The various ways in which this is expressed in the three recordings containing the narration of this video-clip are shown in Table 4. In the preschool recording, the L1 children favour sätta ‘put something so it sticks’, which would be the most straightforward

 Åke Viberg

alternative even in adult usage. In the later recordings, the L1 children more and more favour a more colourful and stylistically marked alternative plåstra om ‘patch up’. This development has many parallels in the other renarrations. (It should be remarked that data from the 10 L1 controls from the Bilinguals at School project have been added in grade 4 to the Swedish L1 students.) The data from the L2 children present in a nutshell a typical L2-related development. In preschool, the most common alternative is the use of the nuclear verb göra ‘make’. Nuclear verbs are often used by L2 learners to replace a wide range of more speciﬁc verbs. Another option is to use a verb that is semantically related to the correct alternative but has a meaning that refers to an action that is very salient and frequent in familiar situations. An example of this is klistra ‘glue (onto something)’, which is used by three learners in preschool. The correct verb, sätta, which is used by only three L2 learners in preschool and in grade 2, is part of a language-speciﬁc, but central, semantic diﬀerentiation pattern in Swedish. Language-speciﬁc diﬀerentiation patterns have a strong tendency to be neutralized by L2 learners. In Swedish, there is a semantic contrast between the three verbs sätta ‘put in a ﬁxed position’, ställa ‘put’, ‘stand’ and lägga ‘put’, ‘lay’, where many languages like English tend to use one verb ‘put’. As has already been observed, the L2 learners at ﬁrst resort to simpliﬁcation and the use of a nuclear verb, in this case göra ‘make’. After that, the major alternative is to let one of the verbs replace the others in language-speciﬁc sets such as sätta/ställa/lägga. As can be observed in Table 4, lägga appears in grade 2 as the most frequent replacement of sätta. Still in grade 4, lägga is as frequent as the correct sätta as a realization Table 4. Verbs used to express: ‘‘They put a band-aid on the tree’’ (Dom satte plåster på trädet)

Swedish L1 Preschool Grade 2 Grade 4 Swedish L2 Preschool Grade 2 Grade 4 a

‘take’

plåstra om sätta lägga göra klistra ‘patch up’ ‘put’, ‘attach’ ‘put’, ‘lay’ ‘make’ ‘glue’

Odd Zero N

1 6 12

5 2 8

3 2a 1

3

12 11 21

5 1

6

1

3 3 7

23 18 18

1

7 7

6 5 2

3 2 1

Age- and L2-related features in bilingual narratives in Sweden 

of the target meaning. The tendency to neutralize the meaning of lägga simply to ‘put’ has many parallels in other parts of the recordings from these times. Frequently, this verb also replaces the verb sätta in another task (sticker activity), where the children put small plastic pictures of various people and objects on a plate background representing a scene such as a hospital or a circus. The L1 children generally used sätta to express the fact that some ﬁgure was put in a ﬁxed position on the plate. The verb lägga is not appropriate in this case, since that verb implies that the ﬁgures were lying loose, which they were not. (Hence the name ‘stickers’). Several of the L2 children were obviously aware of the fact that the ﬁgures stuck to the plate. In the preschool recording, they tended to use the verb klistra ‘glue (onto something)’ to express the same meaning. The acquisition of the semantic diﬀerentiation between sätta/ställa/lägga extends over a long period of time. In the complete set of recordings, lägga is signiﬁcantly overrepresented in grade 1 and grade 2. In preschool, however, lägga had not emerged and most of the L2 children did not use any of the three verbs sätta/ställa/lägga, whereas in grade 4 lägga was still very frequent but not to the degree that it was signiﬁcantly overrepresented (in comparison to the L1 children). Lägga was, however, still rather frequently used with semantic over-extension in grade 4, which indicates that the acquisition of the relevant semantic contrasts were still not completed. The neutralization of the contrasts between the placement verbs and the representation of that subﬁeld by a single verb turned out to be very characteristic even of adult L2 learners in an earlier study, although, in this case, there was variation between the learners with respect to which verb was generalized (Viberg 1996; 1998b). To sum up, there is an age-related development towards the use of increased stylistic ﬂexibility and the use of semantically more loaded and precise alternatives to basic words. In L2, there is a strong tendency to favour nuclear verbs. As demonstrated in Viberg (1993a), this has a parallel in early L1 acquisition, while the neutralization of language-speciﬁc contrasts seems to lack such a parallel and represents an L2-related feature in the narrow sense.

Tense marking The basic functions of the Swedish tense, mood and aspect (TMA) markers correspond relatively closely to the basic English ones except that Swedish

 Åke Viberg

lacks a morphological progressive. The choice between diﬀerent future markers is also rather diﬀerent. In addition, there are several important diﬀerences with respect to the various ‘uses’ of the markers in Swedish and English. In Table 5, the Swedish TMA-system is shown using a verb belonging to the most productive conjugation (For further information, see Viberg et al. 1984, Chapter 9). Table 5. Morphological/periphrastic tense, mood and aspect markers in Swedish Present Imperative Inﬁnitive Past (Plu)Perfect Future Past future/subjunctive

Peter hoppar. Hoppa! Peter vill hoppa. Peter hoppade. Peter har/hade hoppat. Peter kommer att hoppa. Peter ska hoppa. Peter skulle hoppa.

Peter jumps/is jumping. Jump! Peter wants to jump. Peter jumped/was jumping. Peter has/had jumped. Peter will jump. Peter would/should jump.

Since Swedish verbs do not have any agreement markers for person or number, a verb paradigm has rather few forms. There are, however, a number of conjugation classes which will be described in Section 6.2.2. No equivalent to the English progressive is given in Table 5, which only presents obligatory markers. There are, however, a number of periphrastic forms in Swedish which optionally can be used to mark progressive aspect. One alternative is to use the particle verb hålla på (hold + on) coordinated with the main verb as in the following example from a Swedish L1 child in grade 2. (Int.: va läser du för böcker what books do you read?) jaa nu håller ja på å läser en well, now I am reading one som heter ingen e ﬁnare än va joseﬁn which is called Nobody is better than Josephine (KB02, grade 2)

This construction, known as pseudo-coordination, has a number of speciﬁc characteristics setting it oﬀ from regular coordination (Platzack 1979). The postural verbs corresponding to sit, stand and lie can also be used in such pseudo-coordinations to mark progressive aspect:

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

jaa så satt // satt ungen å lekte i sandlådan yes then the kid was playing [lit. sat and played] in the sand-pit så satt mamman å titta på then the mummy was watching [lit. sat and looked on/Ø-marked Past/] (K79, grade 9)

An age-related feature: Switching to narrative present The preschool and early school age children with Swedish as L1 have a strong tendency to use the past tense to mark all verbs that describe events forming part of the plotline. A typical case is shown in Example 5. It should be remembered that the youngest children are ﬁve years old which means that they have passed the age when tense mixing has been observed in narrative contexts in some other languages (Berman and Slobin 1994). Example 5. Narration of the video-clip ‘‘The sand-pit’’. L1 child KB01: Grade 2 (Int. så ska du berätta den också för mej so, can you tell me that one, too) (1) ja va heter de de va en mamma å ett barn yes, what-do-you-call-it it was a mummy and a child (2) å barne lekte i sandlådan and the child played in the sand-pit (3) å mamman satt på bänken å så läste and the mummy sat on the bench and so read [Past] (4) å så kom en farbror å satte sej å fråga and then a man came and sat down and asked [Ø-marked Past] (5) Ifall han ﬁck sitta där if he could [lit. got] sit there (6) å så då viskade han till mannen en sak and then he whispered to the mummy? a thing (7) å så va heter de så blev barnet arg and then what-do-you-call-it the child became angry (8) Kastade sand å va heter de kasta sin spade Threw sand and what-do-you-call-it threw [Ø-marked Past] his spade

 Åke Viberg

(9) så va heter de å sen så kom en annan gubbe then what-do-you-call-it and then (so) another old man came (10) å va heter de // å så gick han förbi and what-do-you-call-it // and then he went by (11) å så sprang den där farbrorn å köpte fyra ballonger till den där lilla pojken and then this man ran and bought four balloons for that little boy (12) å så ﬂög han iväg uppe i luften and then he ﬂew away up in the air (13) å sen va heter de så kysste va heter de farbron / henne på kinden and then what-do-you-call-it (so) kissed what-do-you-call-it the man her on the cheek

All verbs are marked with the simple past tense except the verb in the formula va heter de ’what-do-you-call-it’ which is in the present tense. This expression does not form part of the narrative proper but is rather a speech-planning device (or pause ﬁller). One striking characteristic is that there are very few backgrounded clauses in the versions from younger children. All clauses except the ones in (1)–(3) forming the Orientation belong to the plot-line or foreground in Example 5. Practically all versions contain an Orientation. In some of the narrations in grade 2, there are in addition to that short digressions from the plotline in the form of retrospective remarks marked with the pluperfect or short projections into the future marked with the past future. Such clauses also belong to the background: så då när han hade åkt i luften then when he had gone (up) in the air så pussade mannen mamman (so) the man kissed the mummy (KB06, grade 2) å sen så tog han fram å skulle bjuda han på godis and then he took out and would oﬀer him sweets å då bet han honom i ﬁngret and then he bit his ﬁnger så han skrek / so he screamed (KB12, grade 2)

Age- and L2-related features in bilingual narratives in Sweden 

One characteristic that is rather striking in the narrations by the oldest (16 year old) children in the data-base is the use of narrative (or historical) present. Often the orientation, which primarily contains stative verbs, is narrated in the past tense. The ﬁrst narrative event, the arrival of the man, leads to a switch to present tense as in Example 6. Example 6. Switching to narrative present in the sand-pit story. Grade 9 (K76) (Int. va hände what happened?) (1) ja/ de satt ett litet barn i en sandlåda yeeah, a little child was sitting in a sand-pit (2) och så satt mamman breve på en bänk and its mother was sitting next to it on a park-bench (3) förbi kommer en gubbe å så frågar along comes an old man and asks (4) ifall han får sätta sej breve mamman if he may sit beside the mother (5) hon ﬂyttar på sej she moves. (6) han sätter sej he sits down (7) han har plommonstop he has a bowler hat (8) å ungen slutar gräva and the kid stops digging ETC.

There is some variation with respect to where the switch to present tense occurs as one can see in Example 7, but it usually appears early, shortly after the Orientation.

 Åke Viberg

Example 7. Switching to narrative present in the sand-pit story. Grade 9 (K73) (Int. va hände what happened?) (1) jaa va hände yeeah, what happened? (2) en dam satt på en parkbänk a lady was sitting on a park-bench (3) å en unge satt å lekte i sandlådan and a kid was playing in the sand- pit. (4) å så kom en / herre fram and then a gentleman came along. (5) han har en / stråhatt he has a straw-hat. (6) såg de ut som nästan - It looked like that -almost. (7) eller halmhatt kanske de heter or straw-hat [alternative name] maybe it’s called (8) å han lyfter på hatten å frågar and he lifts his hat and asks (9) ifall han får sitta ner if he may sit down. (10) damen ﬂyttar sej så långt ut på kanten hon kan the lady moves as far out on the edge she can ETC.

The material from each age group is too limited to allow more than tentative conclusions. In grade 4 (age: 10) only one out of 11 L1 children switches compared to 6 out of 12 L1 children in grade 9. Among the 10 adults there is rather wide variation; 5 use present tense throughout, 3 switch from past in the Orientation to present tense with the arrival of the man, while only 2 narrate the whole story in the past tense. Example 7 also contains another feature which seems to represent an agerelated development. In line (3), simultaneity is marked with the Swedish periphrastic progressive en unge satt å lekte i sandlådan ‘a kid was playing in

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

the sand-pit’. Literally, this clause reads: ‘a kid sat and played in the sand-pit’. Even if pseudo-coordinations with a progressive function are found already in some of the recordings with the ﬁve-year-olds, it appears that such constructions are used more frequently with increasing age. In the 11 L1 narrations of the sand-pit story in grade 2, there are two examples of the progressive with a postural verb. In grade 9, 5 out of 10 of the L1 children use the periphrastic progressive and among the adults, 4 out of 10 use it. All uses appear in the Orientation and refer to the activities of the mother and/or the child. With respect to the age-related development, the following picture emerges rather clearly from the material, even if still only parts of the data-base have been systematically analyzed. The younger children (late preschool-early school age) have a strong tendency to stick to simple main verbs in the past tense throughout the story, even if they in principle command the complete adult system. With increasing age, more variation is introduced which also requires more frequent use of the periphrastic tense/aspect markers. Adults exploit the optional markers much more than children. In part, this follows from their more detailed elaboration of the background, where the use of TMA markers crosslinguistically tends to vary more than in the foreground, but the use of the historical present introduces more variation even in the foreground. Simultaneously, there is greater individual variation among adults. Stylistic ﬂexibility is the norm for adults rather than any speciﬁc pattern.

An L2-related developmental path to the Swedish tense-marking system This section will be devoted to two case-studies showing how the basic forms and functions of the Swedish TMA-system are acquired by two learners who are at an early stage in their ﬁrst recordings.

Case study 1: Maria Maria (ﬁctive name) is a Greek girl from the early school-age project, whose development was relatively slow although not exceptional. In the ﬁrst recording when she was six years old, her verb lexicon was limited. Only 26 verb types appeared in the recorded data, which should be compared to 44–91 verb types in the parallel recordings of L1 children. Furthermore, Maria to a very great extent favoured the two nuclear verbs gå ‘go’ and göra ‘make, do’, which comprised 19 per cent vs. 15 per cent of all the verbal tokens in this recording. Most verbs appeared in one basic form throughout the recording. In most cases this form corresponded to the inﬁnitive in L1 Swedish, but a number of





Åke Viberg

present tense forms or bare stem forms also appeared. Example 8 shows the narration from memory of one or more Donald Duck ﬁlms the girl has seen. (A third person s on a verb in the English version is used to mark a present tense ending in Swedish, even if person is never marked in Swedish. When the third person is not marked in the English version, a bare stem or an inﬁnitive form appears in Swedish.) Example 8. Narration of ‘‘Donald Duck’’. Swedish L2 child Maria. Preschool. (Int.: jaha tycker du kalle anka e roli / va brukar han göra för nånting well, do you think Donald Duck is funny? What does he usually do?) (1) han gör // he makes // (2) han gå hem titta / he go home look (3) han titta han mamma // he look he mummy (4) och kalle anka säg till han mamma säg // han säg and Donald Duck say to he mummy say // he say (5) de e ﬁn it is ﬁne (Int.: aha / va ﬁnt / va säger hans mamma då oh, great what does his mummy say then) (6) mm // han säg mm / he say (7) jag ska gå // lite ti // en hus / I shall go / little to // a house (8) ja ska göra // till jobba (mhm) I shall make // to work [INF] (mhm) (9) han jobba / he work [INF] (10) sen han göra // then he make (11) han gå titta (mm) // he go look (mm) //

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

(12) sen han gå hem (mm) äta then he go home (mm) eat (13) kalle anka lite ska gå sova / donald duck a little shall go sleep (14) sen // sen slut then // then end

One year later in grade 1, Maria had mastered the marking of the present tense, at least in principle and the inﬁnitive and a few imperative forms that appeared were also used correctly. Past time reference was also marked but not in accordance with native norms. Instead of the Past tense, the Perfect was used as a generalized marker of past temporal reference. This is clearly illustrated in Example 9 which is a narration of a short video-clip from the ﬁlm based on Astrid Lindgren’s childrens’ book ‘Emil in Lönneberga’. Emil, who is always up to mischief, has found his father’s hat lying on a rock close to the river. His ﬁrst piece of mischief shown in this clip is to take the hat and put it in the water to ﬁnd out whether it can ﬂoat or not. The second one is when he ﬁlls one of his father’s boots with water in order to see if it will leak. In the whole recording, there were 56 Supine forms of 18 verb types, in most cases preceded by har ‘has’. (The main verb is represented in the supine form in the Swedish perfect.) There were only 15 verb tokens with past tense forms realizing 9 diﬀerent verbs (types). The most frequent form was va, the past tense of the copula which is the most frequent Swedish verb. In Example 9, the only past forms are va ‘was’ in (1) and hade ‘had’ in (9). These two verbs have several characteristics in common. From a semantic point of view, they represent states and from a related functional perspective they appear in background clauses, while the perfect forms appear in the foreground or plotline. However, these two verbs also represent the two most frequent verbs in Swedish, and as will be shown in the following case-study, frequency exerts a very strong inﬂuence on acquisition. Example 9. Maria’s version of ‘‘Emil and the boots’’. Grade 1. (1) de va emil it was Emil (2) sen han har hittat he- honoms pappa hatt then he has found he- hims daddy hat





Åke Viberg

(3) sen han har lämnat den inne i vatten (JA) then he has left it inside in water (YES) (4) å sen han harand then he has(5) å sen dom har sjunkt (JA) and then they have sunked (6) å sen domand then they(Int.: va gjorde emil sen what did Emil do then?) (7) sen han har gått å inne i vatten then he has gone and inside in water (8) sen han har trampat i vatten then he has stepped in water (9) sen hans // hans skor dom hade vatten inne then his//his shoes they had water inside (10) sen han har / gått hans pappa hem then he has / gone his daddy home (11) sen han har ee // sen han har gått pappas hem then he has eh //then he has gone daddy’s home (12) å sen han har gjort and then he has made (13) att dom skor ska ha vatten inne (JA) that they shoes shall have water inside (YES) (14) å sen han har gjort ee // han har gjortand then he has done eh // he has done (15) sen pappa har kommit ut then daddy has come out (16) å sen han har tagit en sko sen den andra and then he has taken a shoe then the other (one) (17) sen han har skrik emil then he has scream Emil

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

The primary reason for Maria’s choice of the Perfect form as a generalized past tense marker is probably its salient form. It is in accordance with the analytic tendency often observed in L2 learner language. If this is the correct interpretation, the function of the Perfect in L1 Swedish has been more or less reversed. In native usage, the Perfect is primarily used in background clauses, whereas it appears as the dominant temporal marker in the foreground in Maria’s interlanguage at Time 2. Maria’s generalization of the Perfect has parallels in the recordings with several of the other children in the early school-age project but this characteristic was never found in more than one recording of each child, which means that this stage never lasted more than one year. This also applies to Maria. When she was recorded one year later in grade 2, the use of Perfect as a generalized past had disappeared. She had a grasp of past tense markers in principle with some problems remaining in the choice of the correct allomorph for some speciﬁc verbs. Some of these formal problems could be observed even in the last recording at the end of grade 4, when Maria had a basic command of Swedish grammar and lexicon albeit with occasional errors. Table 6. Summary of Maria’s development during four years. Preschool Verb lexicon

Grade 1

Grade 2

Grade 4

Nuclear verbs, concrete actions;

Varied use of basic verbs;

Broad overextensions

Occasional overextensions

Verbal morphology

Present form mostly correct In principle, one uninﬂected form of each verb

Perfect as a generalized marker of PAST time

Past tense marking grasped in principle, some incorrect forms

ska ‘shall’ for skulle ‘should’ Inversion

None

Connective adverbs

Only sen ‘then’

Almost exclusively sen

Some over-regularized forms: lagde, ’putted’; komde ’camed’ skulle correct in many cases

Variable control

Correct in most cases

Mostly sen

Alternation: så/sen/då





Åke Viberg

Table 6 shows the development over four years of several of the other areas discussed in this chapter. Information on the Swedish inversion rule has been included as well since it is one of the most studied rules in Swedish SLA. The inversion rule requires that the ﬁnite verb always appear in second position in a main clause. If any major constituent other than the subject appears before the verb, the subject must be placed after the verb. This rule developed slowly and had not emerged in preschool. In grade 1, there were a few examples and in grade 2 the control of the rule was variable. The fact that it had been grasped with very few exceptions in grade 4 is a sign that Maria had by that time come rather far in her acquisition of Swedish.

A note on the inﬂectional system of the Swedish verb The acquisition of the Swedish verb inﬂections is inﬂuenced by the system of TMA functions roughly sketched in Table 5 and by the systematic relations between the forms realizing the functions schematically shown in Table 7. This table also includes information about the frequency distribution of the form classes.

Table 7. Form classes of verbs and their frequency distribution in Swedish Productive

Present Imperative Inﬁntive Past Supine

> Lexicalized Regular Short verbs

< er-verbs Weak Strong

Irregular

ar-verbs ramlar ramla ramla ramla(de) ramla(t) ‘fall, tumble’

tror tro tro trodde trott ‘think=believe’

köper köp köpa köpte köpt ‘buy’

springer spring springa sprang sprungi(t) ‘run’

är~e va(r) va(ra) va(r) vari(t) ‘be’

20 33

12 28

0,2 13

Frequency 1. Written Standard Swedish (Allén 1971) Lexical % 67 1 Textual % 25 1

2. Six-year-old Swedish L1 children (from the Preschool project) Lexical % 46 2 19 34 Textual % 11 2 8 79 ( ) segments that are variably omitted in the spoken variety. Person and number are not marked on the verb.

N 4,649 >170,000 263 3,543

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

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

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

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.)

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

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

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

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

Sociocultural aspects of bilingual narrative development in Sweden

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’

Sociocultural aspects of bilingual narrative development in Sweden

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.

Sociocultural aspects of bilingual narrative development in Sweden

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.

Sociocultural aspects of bilingual narrative development in Sweden

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-

Co-constructed narratives by Turkish children in Germany

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

Co-constructed narratives by Turkish children in Germany

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,

Co-constructed narratives by Turkish children in Germany

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

Co-constructed narratives by Turkish children in Germany

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.

Co-constructed narratives by Turkish children in Germany

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<00.001) and the monolingual group (F (2,57)= 25.64, p<00.001). While the Turkish 5-year-olds perform slightly poorer on Component 1 (onset of the plot), the Turkish 7- and 10-year-olds actually “outperform’’ the French monolinguals. This bilingual advantage, however, is not seen in the remaining components. The French monolingual 10-year-olds, in particular, show a higher percentage of usage of the diﬀerent components in comparison to the bilingual 10-year-olds. A comparison of monolingual and bilinguals for the total number of components mentioned per subject shows a slight monolingual advantage for the 5-year-olds (t=1.6, 33df, p<00.10) and no signiﬁcant diﬀerence at seven years of age (t=.83, 34df, NS). There is a signiﬁcant diﬀerence, however, at ten years of age (10-year-olds, t=3.59, 33df, p<00.001). The 10-year-old monolinguals encode more macro-structure components than the 10-year-old bilinguals.

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

Separate comparisons for each component reveals the following results. There is no signiﬁcant quantitative diﬀerence between bilingual and monolingual subjects at any age for the mention of Component 1 (onset of the plot). For Component 2 (unfolding of the plot) there is no signiﬁcant diﬀerence between the monolingual and bilingual 5- and 7-year-olds. However, the diﬀerence is signiﬁcant at ten year of age: more monolingual children encoded Component 2 than bilingual children (P2=4.82, 1df, p<00.05). There is no signiﬁcant diﬀerence between the bilingual and monolingual 5- and 7-year-olds in their encoding of Component 3 (resolution of the plot). However, again at ten years of age there is a signiﬁcant diﬀerence. The monolingual 10-year-olds encode Component 3 signiﬁcantly more than the bilingual children (P2=5.02, 1df, p<00.05). There is no signiﬁcant quantitative diﬀerence between the monolingual and bilingual groups in the mentions of Component 4 (encapsulations), although there is a slight trend toward a monolingual advantage at ten years of age (P2=2.80, 1 df, p<00.10). Table 4. Narrative components in French texts of bilingual French–Turkish children and monolingual French children. Percentage of narrators making explicit mention to each of the four components French bilingual

French monolingual

Age Group N

5 13

7 16

10 15

5 20

7 20

10 20

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

.36 .14 .00 .07

.81 .44 .25 .25

.93 .66 .27 .27

.55 .40 .05 .05

.75 .65 .50 .15

.90 .95 .65 .55

Linguistic structures in L1 vs L2 French As we have seen in the previous section there are few signiﬁcant diﬀerences between the 5- and 7-year old monolingual and bilingual children in terms of total number of components mentioned. Taking each component separately revealed a signiﬁcant diﬀerence only Component 2 (unfolding of the plot) and for Component 3 (resolution of the plot) between the bilingual and monolingual 10-year-olds. Our aim in this section is to examine the linguistic forms used to encode the four components. Do the bilingual and monolingual subjects diﬀer in the structures used, or in the range of structures used to encode each component?

Inﬂuence of L1 Turkish on L2 French narratives 

Component 1: Onset of the plot The structures expressing Component 1 were divided into three categories, illustrated in Table 5. juxtaposition/coordination is considered the least syntactically sophisticated way of recounting the two propositions — the boy looking in the jar, the boy realising that the frog has disappeared. A second way of encoding this episode is through subordination the use of a subordinated complement clause with que (‘that’). juxtaposition/coordination and subordination are mutually exclusive categories. A third category, state of mind, can be added to either category. state of mind refers to attributing a mental state to the boy (and/or the dog) upon the discovery of the missing frog. This implies not just relating two events but imposing a mental state on the boy because of the events in the story. It should be noted that in the French monolingual adult recountings of this story, 64 per cent of the adult narrators included a state of mind clause. Table 5. Structural alternatives for encoding Component 1: onset of the plot juxtaposition/coordination le lendermain matin, le petit garçon la chercha de partout. elle n’était plus dans son bocal (F10;5v) (‘the next morning the little boy looked for her everywhere. she wasn’t in her jar anymore’) il regarda le bocal et ne vit pas la petite grenouille dans son bocal (F10;6f) (‘he looked in the jar and didn’t see the little frog in its jar’) subordination le lendemain matin quand l’ garçon i s’réveille sur son lit i voit qu’ la grenouille est partie (F10;7k) (‘the next morning when the little boy wakes up on his bed he sees that the frog is gone’) state of mind ils ne trouvent plus la grenouille dans le bocal. ils sont peinés (F10;6o) (‘they don’t ﬁnd the frog in the jar anymore. they are hurt)

Table 6 shows the distribution of the various structural alternatives. juxtaposition/coordinationis the preferred structure for all the children. Few subordination structures are found. The French texts of the 5-year-old bilingual children show no subordination at all. The texts of the monolingual children show 3 attempts, but two of the three attempts result in errors. In terms of structures used, juxtaposition/coordination vs. subordina-

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

Table 6. Structures for encoding Component 1: onset of the plot French bilingual Age Group N encoding Component 1/ Total N in Age Group

5 4/13

juxtaposition/coordination subordination state of mind

4

a

0

7

French monolingual 10

5

13/16

13/15

11/20

11 2 0

10 3a 0

8 3 1

7

10

15/20

18/20

12 3 2

13 5 4

Including two errors

tion the bilingual children use the same range of structures as the monolingual children after the age of ﬁve. state of mind is absent from the bilingual texts entirely. Some cases of state of mind clauses are noted in the French monolingual children. The bilingual children are consistent in not encoding state of mind clauses in either their French or their Turkish texts. The monolingual children are beginning to attribute a state of mind to the principal character(s).

Component 2: Unfolding of the plot A variety of structures are observed for the encoding of Component 2 as illustrated in Table 7. On the syntactic level, both simple and complex sentences are used. The simple sentences consist of a single tensed verb and minimally an object, often completed with a locative argument. Also included in the simple category is direct speech. Complex sentences consist of an independent and a dependent clause. The dependent clauses consist of a goal introduced by pour (‘for, in order to’) and a conditional introduced by si (‘if ’). Aspectual verbs and adverbs are another means of encoding Component 2. Adverbs, for example, toujours (‘still’), même (‘even’), encore (‘again’) were observed to encode Component 2. Verbs such as se mettre à (‘puts himself to’) or commencer à (‘begins to’) were considered as aspectual verbs. In addition, the preﬁx re- which indicates reinitiation of an event was considered as attributing reinceptive aspect to a verb. Table 8 shows the distribution of uses of the various structural possibilities. Across all age groups, simple sentences dominate for encoding Component 2. There are no complex sentences in the bilingual 5- and 7-year-olds. By age 10, the bilingual children show considerable use of complex sentences. The 10-year-old monolingual children show more uses of simple sentences

Inﬂuence of L1 Turkish on L2 French narratives 

Table 7. Structural variations for encoding Component 2: unfolding of the plot simple et le petit garçon chercha la grenouille dans un arbre (F10;5c) (‘and the little boy looked for the frog in a tree’) après le garçon il crie “où tu es grenouille? (TF10;11o) (‘after the boy cries “where are you frog? ’) complex il s’agrippe à des branches pour voir si la grenouille est là (F11:02b) (‘he clings to the branches in order to see is the frog is there’) ensuite il regarde dans un trou pour voir si elle y est (TF10;9m) (‘afterwards he looks in a hole to see if she’s there’) aspectual adverbs il appelle toujours la grenouille (F5;0p) (‘he still calls the frog’) et encore le petit chien il cherche dans les arbres (TF7;2g) (‘and again the little dog he looks in the trees’) aspectual verbs il reappelle la petite grenouille (F5;08q) (‘he recalls the little frog’) alors il se mettent à chercher dans la chaussure (F10;10l) (‘so they start to look in the shoe’)

and aspectual items that the 10-year-old bilingual children. The bilingual 5and 7-year-old children use aspectual adverbs, while the monolingual children use both aspectual adverbs and aspectual verbs across all ages. Aksu-Koç (1994: 342–3) comments that the Turkish monolingual children use aspectual and temporal adverbs in cases where Turkish adults show use of Table 8. Distribution of structural variations for encoding Component 2: unfolding of the plot French bilingual

French monolingual

Age group N encoding Component 2/ Total N in age group

5 3/13

7 7/16

7 8/15

5 8/20

7 13/20

10 19/20

simple complex aspectual adverbs aspectual verbs

11

26

28 12

3

2

35 9 3 3

60 1 4 7

90 6 4 5

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

aspectual verbs (with an inﬁnitive complement) such as basla (‘to start’), çik (‘to set out’), and dwam (‘to continue’) to express instantiations, reinstantiations and continuation of the search. It should be noted that the bilingual children’s Turkish texts also show an absence of aspectual verbs but considerable use of aspectual adverbs (Akinci 1999). In summary, across all ages it appears that the bilingual subjects use a more restricted range of structures than the monolingual children.

Component 3: Resolution of the plot Three alternative structures were noted as encoding Component 3. (1) and (2), below, illustrate how the reiterative preﬁx re- and a possessive determiner are used to identify the frog as being the same as at the beginning of the story. An alternate way of reidentifying the frog is by attributing a proper name to the frog at the beginning of the story and maintaining it throughout. (3) illustrates this strategy, as well as, the use of re-. (1) il retrouve sa grenouille (F10;3t) ‘he reﬁnds his frog’ (2) il voit une famille de grenouille et récupère sa grenouille (TF10;11o) ‘he sees a family of frogs and recuperates his frog’ (3) Pierre et Rouki repartent avec Zizi la grenouille (F10;6o) ‘Pierre and Rouki releave with Zizi the frog’

Table 9 shows the distribution of various structures for encoding Component 3. There are few bilingual subjects who encode Component 3. However, the structures used are the same as those used by the monolingual subjects. No bilingual subject and only one 10-year-old monolingual subject used a proper name. Table 9. Structures for encoding Component 3: resolution of the plot Age group N encoding Component 3/ Total N in age group rePossessive Proper name

French bilingual

French monolingual

5 1/13

7 4/16

10 4/15

5 1/20

7 10/20

10 13/20

1

2 2

2 3

1 1

4 10

5 10 1

Inﬂuence of L1 Turkish on L2 French narratives 

Component 4: Encapsulations Component 4 was divided into two categories. Examples of each category are given on Table 2. The ﬁrst category consists of the use of partout (‘everywhere’), a locative expression. The second includes more expanded phrase length alternatives such as the example given on Table 2, maintenant le petit garçon poursuit ses recherches (F10;2b) (‘now the little boy pursues his search’). Adult versions of this second category (il leur est arrivé plein d’aventures au cours de cette recherche (F20g), (‘lots of adventures happened to them in the course of this search’)) are often found at the beginning of the story and announce the series of episodes to come in the story. Child versions are found at the beginning of episodes and establish links between one episode and the other. Table 10 shows the distribution of the two diﬀerent categories used to encode Component 4. Only one bilingual 5-year-old encoded Component 4. Both bilingual and monolingual 7- and 10-year-olds used the locative adverb partout (‘everywhere’). Only monolingual 7- and 10-year-old used more expanded phrase length forms for encoding encapsulations. Table 10. Structures for encoding Component 4: encapsulations French bilingual

French monolingual

Age group N encoding Component 4/ Total N in age group

5 1/13

7 4/16

10 4/15

5

Partout (‘everywhere’) Others

1

4

4

7 3/20

10 11/20

3 2

15 3

Conclusions and discussion The ﬁrst question which motivated our study concerned the macro-structure components encoded by the Turkish–French bilinguals in their two languages. Our analysis revealed no signiﬁcant diﬀerence between the Turkish and French texts in terms of the total number of macro-structures encoded. Our second research question asked whether or not the bilingual and monolingual subjects diﬀered in the total number of macro-structure components expressed. No signiﬁcant diﬀerences were observed between the 5- and 7-year-old groups. However, the diﬀerence between the 10-year-old groups revealed a signiﬁcant monolingual advantage. A separate analysis of each

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

component revealed no signiﬁcant diﬀerences between the 5- and 7-year-olds for any component. The 10-year-old bilinguals, however, show a delay compared to monolinguals in the encoding of Component 2 (unfolding of the plot) and Component 3 (resolution of the plot). Our analysis of the forms used to express the diﬀerent components revealed the following results. The preferred structure of all the children for encoding Component 1 (onset of the plot) is the juxtaposition or coordination of simple clauses. There is a slight delay in the attempted uses of subordinate structures among the 5-year-old bilinguals. While French monolinguals are approaching French adults in the attribution of a state of mind to the principal character, no bilingual subject expressed a state of mind. All subjects show a preference for simple constructions in the encoding of Component 2 (unfolding of the plot). A larger variety of structures, however, is observed among the monolingual children across all ages. For Component 3 (resolution of the plot) the monolingual children show considerable more uses of a possessive determiner than the bilingual children. For the expression of Component 4 (encapsulations) monolinguals were observed to use, in addition to the locative adverb partout (‘everywhere’), other more expanded structures. The bilingual children used only the adverb. The delay of the bilingual children in comparison to the monolingual children is most marked at tens years of age, the age at which the vast majority of the bilingual children are no longer producing clause level errors. All of the narrative components can be encoded by simple clause constructions. One may ask why the bilingual 10-year-olds do not continue on in development as do the monolingual children. The error analysis (Appendix) reveals that the Turkish–French bilingual children attain proﬁciency in clause level grammar of French rather quickly. The study of the structures used by the bilingual and monolingual subjects to encode the diﬀerent narrative components shows that the monolingual’s preferred encoding strategies are available to the bilingual children. We can illustrate this through the example of the possessive determiner used for encoding Component 3 (resolution of the plot). When a monolingual child encoded this component, s/he used either the possessive determiner only, or the possessive determiner and a reiterative preﬁx. The error analysis shows that gender errors on the determiner for the Turkish–French bilingual children are frequent for the 5-year-olds and steadily decreases with age. One of the 10-year-old Turkish–French bilingual children did make a gender error on the determiner in the encoding of Component 3. Nevertheless, we considered

Inﬂuence of L1 Turkish on L2 French narratives 

his production as encoding Component 3. The absence of a determiner is an infrequent error observed in our Turkish–French children : two 5-year-old, one 7-year-old and one 10-year-old show this error (Appendix). The majority of the Turkish–French bilingual children use the possessive determiner elsewhere in their texts: over half (7/13) of the 5-year-olds, 69 per cent (11/16) of the 7-year-old and 100 per cent of the 10-year-olds. It would be diﬃcult to conclude, then, that the Turkish–French bilingual children do not encode Component 3 because they do not have access to the preferred structure (the possessive determiner) of the monolingual children. If there had been a diﬀerence in the macro-structures encoded in the two languages, for example if the texts in Turkish had shown more macro-structure components, we might have argued that the diﬀerence observed in the French bilingual and monolingual texts could be attributed to language competence. However, the comparison of the Turkish and French texts produced by the bilingual children revealed no signiﬁcant diﬀerence in the number of macro-structure components encoded. It would appear, then, that the delay observed in French for encoding Component 3 among the bilingual children is not attributable to a lack of linguistic means in French. The fact that the structure (the possessive determiner) is available to them in French, coupled with the fact that there is no diﬀerence between their Turkish and French texts, would argue that the delay is attributable to a delay in macro-structure development. We argue, however, that the bilingual delay observed in French, particularly in the 10-year-olds, should not be attributed to their bilingualism. We propose that the diﬀerence between the bilingual and monolingual subjects is due to diﬀerences in the amount of exposure to literacy-related activities. Learning to use a language in narrative contexts requires a certain amount of exposure to those contexts. Children acquire community norms for both the form and the purpose of narratives through their early experience in their communities (Heath 1982, 1984). All of our subjects are exposed to narrative texts in French school. Our monolingual subjects, however, are all middle class children for whom classroom narrative activities are reinforced at home: bedtime stories and storybook reading are reported as being everyday home activities. Our bilingual children have very little experience with this type of activity in their home language, Turkish. Half of the parents report that they never read or tell stories to their children. The other half reports that they only occasionally engage in this kind of activity. Eight percent of the Turkish fathers and 26 per cent of the mothers are illiterate. It would be an error, then, to

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

attribute the bilingual French–Turkish subjects’ weakness in their French narratives solely to their bilingualism. Further research is needed to ascertain the kind of narrative experience our bilingual subjects have in their homes.

Appendix: Errors in French in Turkish–French bilingual Frog Stories Table 1: Frequency of sentence level errors in the French texts of French–Turkish bilingual Frog stories. Number of children per group in which errors were produced at least two times. Age Group

5-year-olds 7-year-olds 10-year-olds

N

13

16

15

gender – determiner gender – anaphor and subject clitic gender – adjective gender – object clitic preposition object missing determiner missing subject missing auxiliary missing agreement inﬁnitive copula (être) missing word order: object misplaced existential: elle est → il y a (‘there is’) past participle reﬂexive se missing

12 7 7

8 8

3 2

5 4 2 2 2 2 2 1 1 1 1

2 7 1 3 1

1 1 1 1 1

1

2

Types of error: Deﬁnitions and examples Listed below are the clause level errors observed in the French texts of the Turkish–French bilingual children. Italics indicate the form used by the child. ‘>’ indicates the target form. Whenever possible the error types in French were translated into equivalent error types in English. gender French has two genders, masculine and feminine. Gender is marked on determiners, pronouns, adjectives, and clitics. Determiner le chaussure (>la chaussure, ‘the shoe’), le chèvre (>la chèvre, ‘the goat’), la garçon (>le garçon, ‘the boy’)

Inﬂuence of L1 Turkish on L2 French narratives 

Anaphor and subject clitic le garçon elle va à l’eau. et puis même le chien elle va à l’eau (TF5;10d) (>le garçon il va à l’eau. et puis même le chien il va à l’eau, ‘the boy she goes to the water. and then even the dog she goes to the water’) Adjective elle est content (TF5;10d) (>elle est contente, ‘she is happy’) Object clitic il va le prendre (TF7;11h) (> 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

Bilingual core group

Monolingual control group

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

Monolingual control group

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

Monolingual control 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

Monolingual control 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

Bilingual core group

Monolingual control group

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

Monolingual control group

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

Bilingual core group

Monolingual control group

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

Monolingual control group

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

Bilingual core group

Monolingual control group

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

Bilingual core group

Monolingual control group

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<.01). Once again, this suggest that children who are able to use more complex clauses in Papiamento are also able to do this in Dutch. Table 5. Mean clause length in Papiamento and Dutch per grade Mean clause length Papiamento Mean clause length Dutch

Kindergarten

Grade 2

Grade 4

Grade 6

6.38 5.41

7.39 7.13

8.18 7.75

8.44 7.76

In Table 6 the mean percentages of coordinate and subordinate conjunctions are presented for the two languages per grade. In Figure 2, the mean scores are represented graphically. Table 6. Mean percentages of coordinate and subordinate conjunctions in Papiamento and Dutch per grade Mean coordination Papiamento Mean coordination Dutch Mean subordination Papiamento Mean subordination Dutch

Kindergarten

Grade 2

Grade 4

Grade 6

6.83 5.67 1.33 0.50

7.20 9.70 1.97 0.70

14.42 17.19 6.46 3.00

12.29 16.64 6.21 3.96

It can be seen that the children generally used many more coordinating devices than subordinating devices. Furthermore, there is a gradual increase in

 Ria Severing and Ludo Verhoeven

Percentage

20 15 Mean coordinator Dutch Mean coordinator Papiamento Mean subordinator Dutch Mean subordinator Papiamento

10 5 0

K

G2 Grade

G4

G6

Figure 2. Mean percentages of coordinate and subordinate conjunctions per grade

both types of conjunctions as the children get older. A one-way ANOVA with as the independent variable proved highly signiﬁcant. (p<.001). This means that the use of conjunctions signiﬁcantly increased with age. A sharp increase can again be seen between grades 2 and 4 in both languages. The percentage of both coordinate and subordinate conjunctions is larger in Papiamento than in Dutch although the total percentage of conjunctions used by the children in the Papiamento and Dutch languages was signiﬁcantly related (r=.50, p<.01). That is, those children who use clause linking devices tend to do this in both of the languages. Table 7 provides an overview of the coordinate conjunctions used most frequently by the children. In both Papiamento and Dutch the constructions with ‘and’ were used most frequently and particularly by the three youngest age groups. Table 7. Mean percentages speciﬁc coordinate conjunctions used in Papiamento and Dutch per grade Kindergarten

Grade 2

Grade 4

Grade 6

Papiamento I ‘and’ anto ‘and’ pero ‘but’ logá ‘but’ ma ‘but’

4.34 90.54 4.51 0.00 0.61

14.63 79.92 3.93 1.52 0.00

31.00 44.60 21.10 0.95 1.97

28.41 57.05 12.78 1.08 .23

Dutch en maar want

98.77 0.00 1.23

91.66 7.90 0.44

79.00 19.74 1.26

78.89 18.68 2.43

‘and’ ‘but’ ‘because’

Narrative development in Papiamento and Dutch 

Table 8. Mean percentages speciﬁc subordinate conjunction used in Papiamento and Dutch per grade Kindergarten

Grade 2

Grade 4

Grade 6

0.00 0.00 28.03 0.00 0.00 2.27 0.00 12.00 13.65 4.55 18.21 21.24

.90 .90 52.68 0.00 2.66 2.00 3.99 6.67 17.67 0.00 9.31 2.32

0.00 0.00 43.02 .30 2.88 4.15 0.00 11.34 12.32 2.21 10.97 9.20

4.02 1.04 43.39 1.43 1.79 1.40 0.00 9.52 14.08 2.05 7.58 12.25

0.00 0.00 66.35 0.00 0.00 0.00 0.00 33.35 0.00 0.00 0.00

2.77 0.00 62.50 0.00 0.00 2.77 0.00 9.00 4.17 9.45 8.34

29.05 7.55 8.43 0.00 1.95 0.00 4.05 12.28 6.88 7.14 21.36

47.24 4.33 2.86 2.38 1.43 .45 1.42 16.40 4.65 9.34 7.89

Papiamento tanten (‘while’) promé ku (‘before’) ora (‘when’) te (‘till’) te ku (‘until’) te ora (‘until’) pesei (‘for’) pasobra (‘because’) pa (‘to’) kaminda (‘where’) si (‘if ’) ku (‘that’) Dutch toen (‘when’) terwijl (‘while’) als (‘when’) wanneer (‘when’) omdat (‘because’) zodat (‘so that’) dat (‘that’) om (‘to’) waar (‘where’) dat (‘that’) of (‘if ’)

Table 8 provides an overview of the subordinate conjunctions used by the children. The most frequently occurring subordinate conjunctions were: the temporal conjunctions als and toen in (Dutch) and ora (Papiamento along with), the purposive connectives om in (Dutch) and pa (in Papiamento). In Papiamento, the children were also found to use the causal conjunction pasobra and the connectives si and ku quite frequently.

Topic continuity in Papiamento and Dutch Introduction of a main character In Table 9 an overview of the diﬀerent devices used to the boy in Papiamento and Dutch is presented.

 Ria Severing and Ludo Verhoeven

Table 9. Mean percentages of forms used to introduce boy in Papiamento and Dutch per grade Papiamento 2

Dutch

Introduction

K

4

6

Name Indef det+NP Def det+NP Dem det+NP Pron e/hij (‘he’) Pron e/zij (‘she’) Pron dem esei/die (‘that’)

0.00 0.00 0.00 22.29 20.0064.29 88.00 59.26 66.6732.14 12.00 18.52 0.00 0.00 0.00 0.00 6.67 3.57 0.00 20.00 0.00 0.00 0.00 0.00 6.67 0.00 0.00 0.00

K

2

4

0.00 0.00 3.85 20.00 78.57 92.30 60.00 14.28 3.85 0.00 3.57 0.00 20.00 0.00 0.00 0.00 3.57 0.00 0.00 0.00 0.00

6 17.80 75.00 3.57 3.57 0.00 0.00 0.00

From Table 9, it can be seen that some of the youngest children use a pronoun to introduce the boy in their narratives. This was particularly true for Dutch. Some other children appear to use a deictic strategy even when the booklet is closed in front of them. An example is given below. Richard (5 years): Esaki i esaki ta wak e sapu. ‘This and this are looking at the frog.’

In the vast majority of the narratives, a full noun phrase (i.e., the standard linguistic device) is used to introduce the main protagonist in the story. In Figure 3, the mean percentage introductions with a NP in Papiamento and Dutch is presented for the diﬀerent grades. Inspection of Table 9 also shows the majority of the youngest children to use a deﬁnite determiner with a NP to introduce the boy. Prototypical examples are given below. 100 Percentage

95

Papiamentu

90 Dutch

85 80 75 K

G2 G4 G6 Grade Figure 3. Percentage introductions of the boy with a NP in Papiamento and Dutch per grade

Narrative development in Papiamento and Dutch 

Dustin (5 years): E mucha hòmber ke drumi. ‘The boy wants to sleep.’ Rhona (5 years): De jongen en de hond slapen. ‘The boy and the dog are sleeping.’

With age, however, the data show a clear increase in the use of indeﬁnite determiners with a NP to introduce the boy. In Table 10, an overview of the forms used to introduce the dog by the children in Papiamento and Dutch per grade is presented. Table 10. Mean percentage forms used to introduce the dog in Papiamento and Dutch per grade Papiamento 2

Dutch

Introduction

K

4

6

Full NP Indef det+NP Def det+NP Dem det+NP Poss pro+NP Pron e/hij (‘he’)

6.67 0.00 0.00 3.70 20.00 33.33 37.50 22.22 73.33 51.85 25.00 48.15 0.00 0.00 0.00 0.00 0.00 14.81 37.50 25.93 0.00 0.00 0.00 0.00

K

2

4

6

0.00 0.00 0.00 8.33 20.00 34.62 48.00 25.00 80.00 46.15 24.00 41.60 0.00 15.38 0.00 19.00 0.00 3.85 28.00 25.00 0.00 0.00 0.00 0.00

Inspection of Table 10 shows none of the children to introduce the dog with a pronoun. Some of the older children in particular like to introduce the dog in both Papiamento and Dutch with a possessive pronoun plus NP. An example is given below. Gianni (12 years): Bobby en zijn hond Fik hadden een kikker gevonden. ‘Bobby and his dog Fik found a frog.’

In Table 11, the correlations between the use of a noun phrase versus a pronoun to introduce the main character in the children’s narratives in Papiamento and Dutch are presented. It can be seen that the correlation between the use of a NP in the other language is signiﬁcantly positive along with the use of a pronoun in the one language and a pronoun in the other language.

Maintenance of reference to main character in Papiamento and Dutch Table 12 presents an overview of the linguistic forms children use to maintain

 Ria Severing and Ludo Verhoeven

reference to the boy in their Papiamento and Dutch stories. It can be seen that the use of pronouns gradually increase with age. Table 11. Correlations between use of noun phrase and use of pronoun for topic introduction in Papiamento and Dutch

Noun phrase Dutch Pronoun Dutch Noun phrase Papiamento Pronoun Papiamento a

Dutch

Papiamento

Noun phrase Pronoun

Noun phrase Pronoun

1.00 −0.17 0.33 −0.48

1.00 −0.72b

1.00 −0.09 0.23a

1.00

p < .05; b p < .01

Table 12. Percentages of linguistic forms used to maintain reference to boy in Papiamento and Dutch per grade Papiamento 2

Dutch

Maintenance

K

4

6

Full NP (name) Indef det+NP Def det+NP Dem det+NP Poss pro+NP Pron e/hij (he) Pron e/zij (she) Pron dem esei/die (that) Pron plur nan/zij (they)

0.00 0.00 0.74 4.56 0.98 0.00 0.00 0.00 18.14 16.30 15.00 14.70 0.00 0.00 0.00 0.00 0.00 0.32 0.00 0.00 80.88 83.30 84.20 80.70 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

K

2

4

6

0.00 0.00 3.81 0.79 0.00 0.00 0.00 0.00 35.27 18.05 13.12 12.31 0.00 8.31 1.28 0.77 0.00 0.00 0.00 0.00 49.72 30.00 58.10 79.60 15.00 37.70 21.60 5.28 0.00 5.78 1.92 1.19 0.00 0.00 0.00 0.00

In Figure 4 the mean percentages maintenance of reference to the boy using a pronoun is presented graphically. As can be seen from Table 12 and Figure 4, older children nevertheless tend to use a NP to maintain reference to a main character at times and thereby avoid misunderstandings, the protagonist, or introduce a new episode (see example below). Stephanie (12 years): Het jongetje sprong. ‘The boy jumped.’ Toen zei hij: laten we hem zoeken ‘Then he said: Let’s search for him.’ Het jongetje ging kijken in een hol. ‘The boy went and had a look in a hole.’

Table 13 presents the percentages diﬀerent linguistic devices used to maintain reference to the dog in Papiamento and Dutch by the children in the diﬀerent grades.

Percentage

Narrative development in Papiamento and Dutch 

100 95 90 85 80 75 70 65 60

Papiamentu

Dutch

K

G2 G4 G6 Grade Figure 4. Maintenance of reference to the boy with a pronoun in Papiamento and Dutch per grade

Table 13. Mean percentages of linguistic forms used to maintain to the dog in Papiamento and Dutch per grade Papiamento 2

Dutch

Maintenance

K

Full NP (name) Def det+NP Dem det+NP Poss pro+NP Pron e/hij (he) Pron e/zij (she) Pron dem esei/die (that) Pron plur nan/zij (they)

0.00 0.00 0.83 3.79 49.44 46.30 40.60 44.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 50.56 53.60 58.50 51.70 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

4

6

K

2

0.00 55.00 0.00 0.00 45.00 0.00 0.00 0.00

0.00 0.00 1.67 36.10 19.44 49.30 17.70 3.33 5.83 0.00 3.33 5.00 18.80 58.33 33.10 20.50 12.22 0.00 6.67 3.33 5.00 0.00 0.00 0.00

4

6

The pattern of maintaining reference to the dog does not resemble the pattern of maintaining reference to the boy. All of the children show a general prefererence for the use of nominal forms to refer to the dog. This can be explained by the fact that the dog is not so prominent as the boy. It is also possible that the children want to stress that their continued reference to the dog and not the boy. Table 14 presents the correlations between the use of a noun phrase versus a pronoun to maintain reference to a character (i.e., both the boy and the dog) in the children’s Papiamento and Dutch stories. While the correlation between the use of a pronoun in the two languages was found to be signiﬁcantly positive, only a nonsigniﬁcant positive correlation was found for the use of a NP in the two languages.

 Ria Severing and Ludo Verhoeven

Table 14. Correlations between linguistic forms (NP, pro) used for topic maintenance by the children in Papiamento and Dutch Dutch Maint. NP Dutch Maint. Pronoun Dutch Maint. NP Papiamento Maint. Pronoun Papiamento a

Papiamento

NP

Pronoun

NP

Pronoun

1.00 −0.36b 0.14 −0.07

1.00 −0.15 49b

1.00 −0.41b

1.00

b

p < .05; p < .01

Switch of main characters in Papiamento and Dutch In Table 15, an overview of the linguistic forms children used to switch reference back to the boy in the Papiamento and Dutch stories is presented. It can be seen that the percentages noun phrases generally increases with age. Table 15. Mean percentage of linguistic forms used to switch reference back to boy in Papiamento and Dutch per grade Papiamento 2

Dutch

Switch

K

Full NP Indef det+NP Def det+NP Dem det+NP Poss pro+NP Pron e/hij (‘he’) Pron e/zij (‘she’) Pron dem sei/die (‘that’) Pron plur nan/zij (‘they’)

0.00 0.00 7.59 29.90 0.00 0.00 0.00 0.00 30.64 70.60 73.00 44.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 69.36 28.90 19.30 25.30 0.00 0.00 0.00 0.00 0.00 0.42 0.00 0.00 0.00 0.00 0.00 0.00

4

6

K

2

4

6

0.00 0.00 16.57 13.20 0.00 0.00 0.00 0.00 32.94 62.20 68.69 45.35 4.76 10.30 .70 8.82 0.00 0.00 0.00 0.00 51.98 9.72 13.05 31.60 8.33 15.70 .99 0.89 0.00 0.00 0.00 0.00 0.00 1.94 0.00 0.00

In Figure 5, the mean percentage NPs used to switch reference back to the boy in Papiamento and Dutch is presented graphically. Although the majority of the older children (i.e., the fourth and sixth graders) show a preference for the use of nominal forms to mark the switch of reference back to the boy in the story these children nevertheless switch more often with a pronoun than the second-grade children. One possible explanation for this is that the children intentionally use a pronoun in those cases where suﬃcient context is available to the listener to follow the switch of reference. In the example below, for instance, clear semantic constraints indicate that the pronoun hij (‘he’) can only refer to the boy.

Percentage

Narrative development in Papiamento and Dutch 

100 90 80 70 60 50 40 30 20

Dutch

Papiamentu

K

G2 G4 G6 Grade Figure 5. Mean percentages Switch of reference back to boy using NP in Papiamento and Dutch per grade

Saskia (12 years): De jongen stond op om naar de kikker te kijken. De kikker was ’s nachts weggegaan. Hij zocht in de hele kamer.

‘The boy got up to watch the frog.’ ‘The frog left at night’ ‘He looked in the entire room’

Another explanation for the increased use of pronouns among the older children may lie in their increased understanding of how to mark and maintain reference to the thematic subject in a narrative (Karmiloﬀ-Smith 1981). According to Karmiloﬀ-Smith, language-learning children may gradually allow non-thematic subjects to occasionally occupy initial utterance slots, but appropriately refer to these using full deﬁnite expressions, nominal-pronominal reduplications, and other clear linguistic markers. These same children thus continue to pronominalize the thematic subject appropriately at the discourse level. The child in the following example uses a nominal form for both the dog and the bees but appropriately switch to a pronoun for the dog. Leeroy (8 years) De hond ging bij de bijen. Toen gingen de bijen prikken. Toen was hij gevallen. Er was nog een dier. Ging hij op zijn hoofd klimmen.

‘The dog went to the bees.’ ‘Then the bees were pricking.’ ‘Then he fell.’ ‘There was another animal.’ ‘He climbed up his head.’

Inspection of Table 15 and Figure 5 clearly shows the number of switches of

 Ria Severing and Ludo Verhoeven

reference back to the boy using a NP to increase at ﬁrst and then decrease slightly. That is, the oldest children use a pronoun whenever possible and thus use a NP to mark such a switch of reference less than the younger children. The strategies used to switch reference back to the dog are almost the same in Papiamento and Dutch. In Table 16, the percentages for the diﬀerent switching devices in Papiamento and Dutch are presented per grade. As can be seen, all of the children have a clear preference for marking a switch of reference back to the dog with a NP. Table 16. Mean percentage linguistic forms used to switch reference back to the dog per grade Papiamento 2

Dutch

Switch

K

4

6

Full NP Indef det+NP Def det+NP Dem det+NP Poss pro+NP Pron e/hij (‘he’) Pron e/zij (‘she’) Pron dem esei/die (‘that’) Pron plur nan/zij (‘they’)

0.00 0.00 4.40 6.67 0.00 0.00 0.00 0.00 87.50 96.00 88.00 93.3 0.00 0.00 0.00 0.00 0.00 2.90 5.90 0.00 12.50 0.00 1.71 0.00 0.00 1.09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

K

2

4

6

20.00 4.76 13.64 7.69 0.00 0.00 0.00 0.00 80.00 75.50 77.73 70.96 0.00 10.10 0.00 17.60 0.00 4.76 7.73 3.65 0.00 0.00 .91 0.00 0.00 4.76 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Table 17 presents the correlations between the use of a noun phrase versus a pronoun to mark a switch of reference in the children’s stories in Papiamento and Dutch. It can be seen that the correlation between the use of a NP in the two languages is signiﬁcantly positive. No substantial correlation was evidenced for the use of a pronoun in the two languages. Table 17. Correlations between linguistic forms (NP, PRO) used for topic switch by children in Papiamento and Dutch Dutch Noun phrase Noun phrase Dutch 1.00 Pronoun Dutch −0.07 Noun phrase Papiamento 0.50a Pronoun Papiamento −0.21a a

b

p < .05; p < .01

Papiamento Pronoun

Noun phrase

Pronoun

1.00 0.10 −.06

1.00 −0.51b

1.00

Narrative development in Papiamento and Dutch 

Conclusions and discussion Several conclusions can be drawn on the basis of the present results. First, it can be concluded that the global structure of the children’s narratives develops with age. As the children get older, their narratives tend to get longer and more elaborated. The narratives of the older children generally contain a greater number of propositions and coherence relations than the narratives of the younger children. In addition, a striking diﬀerence in the length of the narratives for all of the age groups was found for the two languages. The Papiamento narratives were consistently longer than the Dutch narratives. It is clear that the children in the ﬁrst four grades of elementary school in Curaçao describe the events in The Frog Story more easily in their mother tongue than in Dutch. The occurrence of monitoring indicators, moreover, suggests the same. In Dutch, the children produce considerably more indicators of monitoring, because they need additional planning time. A rather striking result is that the measures of text length and text coherence in the two languages are clearly correlated. This indicates that children who tell long and elaborated stories tend to do so in both L1 and L2. With respect to the use of various syntactic devices, a number of developmental eﬀects were found. A clear developmental eﬀect was found for mean clause length. The use of clause linking devices also increased with age and particularly between grades 2 and 4. All of the children tend to use many more coordinating devices than subordinating devices. For both mean clause length and the numbers of conjunctions, signiﬁcant positive correlations occurred between the two languages, which points to an underlying growth of syntactic abilities (cf. Verhoeven 1994). With regard to the making and maintenance of narrative references tracking, some clear developmental patterns emerged. In both Papiamento and Dutch, the abilities of the children to avoid narrative ambiguities clearly develop. The children learn to use a full noun phrase to introduce the main protagonist in a narrative, a pronoun to maintain reference to the main protagonist, and a full noun phrase to mark a switch to another character in the narrative. In maintaining reference to a particular character, they also learn to use a noun phrase to avoid misunderstandings and/or, to mark episode boundaries. In switching references, they show an increased understanding of the notion of thematic subject (cf. Karmiloﬀ-Smith 1981). Finally, the children are generally more proﬁcient reference trackers in Papiamento than in Dutch. As far as the use of various unmarked devices for

 Ria Severing and Ludo Verhoeven

introducing, maintaining, and switching characters, a positive correlation between the two languages was nevertheless evidenced. Thus it is clear that universal strategies underlie the development of narrative cohesion. In whatever language children must learn to elaborate intersentential devices so that independently represented entries in memory form a system (Hickman 1995). It is by creating a control process that bilingual children, gradually learn how to con strain the production of cohesive markers across related clauses in their ﬁrst and second language narratives (Karmiloﬀ-Smith 1992; Verhoeven 1993).

References Berman, R. A. and D. I. Slobin (1994). Relating events in narrative: A cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Ferrol, O. (1982). La cuestion del origen y la formacion del Papiamento. Den Haag: UNA. Flores d’Arcais, G. B. (1978). The acquisition of the subordinating constructions in child language. In: R. N. Campbell and P. T. Smith (eds.), Language development and motherchild interaction. New York: Plenum Press, 349–93. Hickman, M. (1995). Discourse organization and reference. In: P. Fletcher and B. MacWinney (eds.), Handbook of child language. Cambridge, MA: Basil Blackwell, 194–218. 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 Klein W. (1986). Second language acquisition. Oxford, MA: Cambridge University Press. MacWhinney, B. (1991). The CHILDES project: Tools for analysing talk. Hillsdale, NJ: Lawrence Erlbaum. Maurer, P. (1988). Les modiﬁcations temporelles et modales du verbe dans le Papiamento de Curaçao. Hamburg: Buske Verlag Severing, R. (1996). Geletterdheid en onderwijssucces op Curaçao. Doctoral dissertation, University of Nijmegen. Severing, R. and L. Verhoeven (1995). Taalvaardigheid Papiamento en Nederlands van leerlingen op Curaçao. Pedagogische Studiën, 72, 357–73. Trabasso, T. and P. Rodkin (1995). Knowledge of goals and plans: A conceptual basis for Frog, where are you? In: Berman and Slobin (eds.), 85–107.

Narrative development in Papiamento and Dutch 

Verhoeven, L. (1989). Monitoring in children’s second language speech. Second Language Research 5, 141–55. 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.

Chapter 11

Linguistic features of Spanish-Hebrew children’s narratives1 Judy Kupersmitt and Ruth A. Berman

The topic of bilingualism has aroused considerable interest in language acquisition research in recent decades. Linguists, psychologists, and sociologists have investigated bilingual populations from diﬀerent perspectives, to ascertain how bilingualism aﬀects cognitive abilities like memory, perception, and metalinguistic awareness; how such skills are reﬂected in each of the bilingual’s two languages; and how acquisition and command of the two languages are aﬀected by socio-cultural and communicative factors in the context of classroom communication and conversational interaction. Findings from this broad ﬁeld of research suggest that in order to understand children’s acquisition and knowledge of two languages, it is not enough to consider their command of grammatical structure, but also how they deploy their linguistic knowledge in diﬀerent contexts of language use. To this end, the present chapter addresses the question of 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 the ability to use the language, in this case, in the discourse-embedded context of telling a story. Along with the general upsurge in cognitively and linguistically motivated studies of bilingual children, a growing body of research has emerged on narrative development among bilinguals. Dart (1992) analyzed the narrative style of a preschool (and hence preliterate) bilingual child in a spontaneous story-telling context by counting the various linguistic devices used in the two languages and comparing between them. From a very diﬀerent perspective, Duran (1985) examined the production and comprehension of narrative texts among language minority bilingual schoolchildren in order to examine how literacy develops in a context where sociocultural factors play an important role. A number of studies have also extended the Berman and Slobin (1994) crosslinguistic analyses of narrative development using the wordless ‘‘frogstory’’

 Judy Kupersmitt and Ruth A. Berman

picturebook (Mayer 1969) for a variety of purposes, such as the impact of bilingualism on emergent literacy among Canadian children enrolled in francophone preschools (Herman 1996). An important, rather diﬀerent direction of this research is concerned with bilingual contexts in immigrant situations, including children speaking Turkish and Dutch (Aarssen 1996), Moroccan Arabic and Dutch (Bos 1997), and Turkish and French (Akinci 1997; Jisa and Akinci 1996), as well as Spanish and English (Pearson and Umbel 1996). These latter studies are concerned mainly with how diﬀerent grammatical systems like markers of cohesivity and reference develop over time in the ﬁrst and second languages of their subjects. While using the same elicitation materials as such studies, the research reported on below diﬀers from them in several ways. First, the Spanish– Hebrew bilingual children that are its subjects do not fall into the category generally deﬁned by ‘‘immigrant’’ or ‘‘minority’’ children, since they come from homes with a high level of literacy and education, and as such are sociologically similar to the middle-class Spanish-speaking and Hebrew-speaking monolinguals in the Berman and Slobin (1994) study (see the chapters by Sebastián and Slobin and by Berman and Neeman respectively). Second, our study compares two distinct facets of narrative ability — command of overall narrative structure, on the one hand, and command of linguistic forms, on the other. Third, the linguistic features examined here were selected from a contrastive perspective, to focus on grammatical systems which reveal diﬀerent degrees of similarity or distinctiveness in the two target languages, Spanish and Hebrew. In a recent study on bilingual proﬁciency, Berman (1998) proposes that bilingual proﬁciency can be ranked on a continuum from non-native to nativelike, ranging from command of core morphosyntactic structures at the level of the isolated sentence to a discourse-motivated use of expressive options. She argues that diﬀerent bilingual groups may show the same level of command of a narrative schema or of narrative action structure, and yet diﬀer in their use of the linguistic and rhetorical devices of each language. In this context, narrative structure refers to the overall organization of a text in terms of an initiating event which creates a problem or determines a goal for the protagonist — in the case of the picturebook in question here, the fact that the little boy’s pet frog escapes; attempts to solve this problem — the boy’s searching for his lost frog in diﬀerent places; and an eventual resolution — the boy ﬁnds his frog or a substitute frog. Levels of narrative organization below that of overall global structure are local relations of temporality and causality

Spanish–Hebrew children’s narratives 

(Giora and Shen 1994; Shen 1988). Studies undertaken from various perspectives have shown that children’s narratives give expression to linearly local relations earlier than command of global action-structure, which emerges relatively late in diﬀerent narrative settings (Berman 1995). In monolingual development, children display command of a wide range of morpho-syntactic structures in their native language from a very early age, particularly at the level of single-clause, isolated utterances and in highly scaﬀolded conversational interaction. Lexical mastery and the ability to deploy a large range of vocabulary items appropriately and with a high level of semantic speciﬁcity are aspects of linguistic knowledge which take longer to develop. An even more complex task, at a still higher level on the continuum of proﬁciency in both language structure and language use, is the ability to make skilled, ﬂexible use of a range of rhetorical devices and expressive options to meet narrative functions such as connectivity, reference, and the expression of temporal relations and background/foreground distinctions. Against this background, the present study seeks to examine the linguistic and narrative abilities of Spanish–Hebrew bilingual children, in order to show that ‘‘proﬁciency’’ and ‘‘nativeness’’ need to be evaluated in terms of both types of evidence: linguistic forms as representing knowledge of the grammar of a given target language and narrative organization as representing knowledge of the structure of narrative discourse. Appropriate use of languagespeciﬁc forms and discourse-motivated means for meeting narrative functions like marking of temporal relations and textual connectivity reﬂects the interaction between these two types of knowledge. Underlying the study were the following three questions: 1. To what extent do bilingual children in the age range from 4–12 years succeed in maintaining narrative organization in Spanish/Hebrew storytelling? 2. How are selected linguistic devices used in the two languages to express various functions in narrative and to what extent does the increasing exposure of the subjects to the typology of Hebrew aﬀect the production of linguistic forms in Spanish? 3. What grammatical errors are made by the children as a function of their relative command of the two languages? To examine the ﬁrst question, we analyze narrative structure in terms of command of a global ‘‘action structure’’ principle of narrative organization, deﬁned by a narrative text which consists of an initiating event, attemps of the

 Judy Kupersmitt and Ruth A. Berman

protagonist(s) to solve the problem or meet the goal instigated by this event, and the outcome or resolution of the problem. The second question is addressed by examining how linguistic forms interact with narrative functions in texts produced by Spanish–Hebrew bilingual preschoolers and schoolage children, focusing on the texts produced in Spanish, as the language more likely to be aﬀected in this particular bilingual situation. Analysis concerns how the linguistic categories of tense/aspect, subject ellipsis, and relative clause constructions serve to express diﬀerent narrative functions in the two languages at diﬀerent developmental stages. These three categories were selected as potentially diagnostic for command of the two target languages, since they range from systems which are highly diﬀerentiated between the two (aspect marking is highly elaborated in Spanish, whereas Hebrew has no grammatical aspect) to one which is partially similar (null subjects are grammatically licensed in both Spanish and Hebrew, but Spanish requires subject ellipsis in more environments than Hebrew), down to a system which is highly similar in the two languages (relative clause construction). To answer the third question, an error analysis was undertaken of the texts produced by subjects. Grammatical errors include violations of wellformedness in word order, case-marking, agreement, and other morphosyntactic features at the level of simple-clause construction and in clause-combining. In terms of narrative organization, we assumed that overall narrative structure would show similar development in the Spanish/Hebrew texts as in the monolingual (Spanish and Hebrew) texts and that children would be able to maintain the structure of their narratives even under the demands of language switching. In the domain of form-function relations, we predicted that diﬀerences in the use of linguistic forms to meet relevant narrative functions between the two languages would be proportional to the degree to which the systems in question are similar across the two languages. Thus, we expected to ﬁnd considerable diﬀerences in use of tense/aspect marking of temporality, some diﬀerence in use of null subjects for connectivity, and little diﬀerence in use of relative clauses for referential and other narrative functions. In fact, however, it turns out that tense/aspect, as a very central system in the Spanish language, seems considerably reduced under the impact of a very diﬀerent typology for encoding temporal/aspectual distinctions in Hebrew. We expected the error analysis to reveal many more errors in Spanish than in Hebrew and that the number of errors would increase with age. The reason is that Hebrew turns into the dominant language for the subjects as early as by

Spanish–Hebrew children’s narratives 

the time they enter kindergarten, around age 4 to 5 years, similarly to what was found for a comparable group of English/Hebrew Israeli bilingual children (Rabinowitch 1985). We expected to ﬁnd three types of errors: a. Errors of ‘‘fossilization’’ (see Ravid 1988 for L1 and Selinker 1974 for L2) analogous to ‘‘stagnation’’ errors (Aarssen 1996; Bos 1997): these are forms which manifest a kind of ‘‘developmental deﬁcit’’ (Kupersmitt 1999), in the sense that they occur early in the acquisition process and fossilize as such; that is, they do not change with development, and so never conform to target language forms. b. ‘‘Transfer errors’’: errors deriving from what is sometimes called ‘‘interference’’ (Berman 1985) from another language, in situations of languages in contact. These are cases where structural patterns or lexical distinctions are transferred from the dominant language to the weaker language. c. Developmental errors: errors which reﬂect strategies adopted by children in attempting to deal with structural diﬃculties in the process of language acquisition (Berman 1984a; Ravid 1988). The ﬁrst and second type of errors should occur mainly in the Spanish texts, while developmental errors may occur in both Spanish and Hebrew, particularly among the younger children.

Method Participants Nine Spanish–Hebrew bilingual subjects aged 4 to 12 participated in the study. Bilingual children are deﬁned in this context as children raised by Spanishspeaking parents in a Hebrew-speaking country (Israel). These children ﬁrst acquire Hebrew through television and occasional peer contact in the neighborhood, extended from age two years to intensive peer contact through daily attendance at nursery school or daycare, from age 5 through attending Hebrew-medium kindergartens and subsequently schools six days a week. To characterize the nature of their bilingualism, we rely on the distinction made by Polinsky (1997), between the ﬁrst and second language (L1, L2) as distinguished by temporal order of acquisition, and between primary and secondary language, distinguished by prevalence of usage. The L1 of our subjects is Spanish, but we are assuming, based on personal observation and empirical research

 Judy Kupersmitt and Ruth A. Berman

(Rabinowitch 1985), that by late preschool (ages 5–6), Hebrew has become their primary language. That is, their L1 Spanish can be deﬁned by Romaine’s characterization as a ‘‘non-dominant home language without community support’’ (1995: 183–5). Subjects were divided into four groups as detailed in Table 1. Table 1. Breakdown of population by age and level of schooling Schooling

N

Age range

Mean age

Preschool

2 3 2 2

3;6–4;4 5;11–6;2 7;4–7;6 10;1–12;4

4;0 6;0 7;5 11;2

Grade II Grade VI–VII

The ﬁrst two groups of children are preschoolers, attending nursery school and kindergarten respectively; the two older groups are schoolchildren, in early grade school and junior high respectively. As such, the two main subgroups reﬂect a state of preliteracy (ages 4 to 6) compared with diﬀerent levels of literacy (ages 7 versus 11). All subjects are from urban middle-class background, and both of their parents are educated native speakers of (South American) Spanish. They all come from homes where Spanish is the dominant language, and all attend monolingual, Hebrew-medium preschools or schools. Given the small number of participants and the large age range covered by our sample, this study is exploratory in nature. Nonetheless, the patterns it reveals appear indicative of quite general trends in bilingual situations with a population such as ours where (a) subjects are of middle-class, well-educated backgrounds and (b) they acquire their home language somewhat earlier than the host language of the society. (In this case, the home language is Spanish, the host language, Hebrew).2 Thus the context and nature of bilingualism of a study such as ours is necessarily diﬀerent than many of the bilingual studies in immigrant situations cited in our introduction.

Materials Narratives were elicited by showing subjects the booklet by Mercer Mayer Frog, Where are you?, a picturebook with 24 pictures describing the adventures of a boy and his dog looking for their frog which has disappeared. This basis for text production was chosen for several reasons. First, use of a single stimulus makes it possible to control for content of the texts elicited, since all

Spanish–Hebrew children’s narratives 

subjects were relating to the same booklet. Second, contrary to other kinds of elicitation settings like script narration or personal-experience narratives, picturebook stories can be told using either the present or past tense as a temporal anchor (Berman and Slobin 1994). Third, storytelling elicited by means of a picturebook does not demand that the subject recall or imagine events, and so may reduce the cognitive load imposed by the task (Shatz 1985). And, critically from our point of view, this book has been used successfully for elicitation of narrative texts from monolingual children at the same age-range as those included in our study, with diﬀerent ﬁrst languages: English (Berman and Slobin 1994; Renner 1988; Wigglesworth 1992); French (Kail and Hickmann 1991; Kern 1997); German (Bamberg 1987; 1994); and, most importantly for our purposes, Hebrew (Berman 1988; Berman and Neeman 1994) and Spanish (Bocaz 1991, 1992; Sebastián and Slobin 1994). Also, as noted, the same booklet has recently been used in a number of studies of bilingual children, including speakers of Spanish in the United States.

Procedures Subjects were ﬁrst instructed to go through the entire booklet and to carefully look at the pictures. They were then asked to tell the story with the book open, changing the language in the middle of the story, the ﬁrst half in Spanish and the second half in Hebrew. Instructions were given in the language in which subjects had to tell the story, Spanish for the ﬁrst half of the story and Hebrew for the second half. The instructions were as follows, the same as in the Berman and Slobin (1994) study: ‘‘Here’s a picturebook about a boy (pointing to picture on cover), a dog (pointing), and a frog (pointing). First, I want you to look carefully at all the pictures. Pay attention to each picture and then you tell me the story while you look at the pictures again’’. After the subject had ﬁnished telling the ﬁrst half of the story, instructions were given in order to continue in Hebrew, as follows: ‘‘Well, from now on, go on with your story in Hebrew’’. The switch was made after the ﬁrst twelve pictures out of the total 24, selected as a point that provided an equal amount of pictorial (though not necessarily thematic or episodic) stimuli for the two languages. Interestingly enough, the subjects all appeared to make the switch without any eﬀort, with only one exception, a 3-year-old child who needed help when changing language from Spanish to Hebrew. Sessions were held in the subject’s home, often in the presence of the mother and lasted around half an hour. They were recorded by audio taperecorder and transcribed soon after the time of record-

 Judy Kupersmitt and Ruth A. Berman

ing, following the transcription conventions adopted for both Spanish and Hebrew in the Berman and Slobin study. This methodology means that the texts produced in the two languages were not based on the same scenes from the picturebook, and hence did not relate to the same story constituents or thematic content. And indeed, this procedure yielded certain incompatibilities between our ﬁndings for the two languages (Spanish in the ﬁrst part and Hebrew in the second). Nonetheless, we feel that it was worth making this attempt to vary the accepted methodology of having the same subjects tell the same story in their two languages, particularly because of the problem of order eﬀect and the inevitable eﬀect of experience on the second rendering.

Data analysis As noted, the texts were analyzed along two dimensions: narrative organization and linguistic expression.

Narrative organization Following Berman and Slobin (1994: 46–50), the content of the picturebook was analyzed into three core components specifying the narrative action structure of the story as set out below: Action structure of the picturebook story Frog, Where are you?: Initiating event The frog escapes from the jar while the boy and dog are sleeping. The boy and dog wake up to ﬁnd that the frog has disappeared. Attempts to solve the problem The boy and dog go out in search of the frog, leading them into a series of adventures and mishaps in the forest. Resolution The boy lands up in a pool of water, where he ﬁnds the frog that he has lost (or gets another frog in its place). Subjects were credited with having command of the story’s action structure if they gave explicit linguistic expression to each of these three components:

Spanish–Hebrew children’s narratives 

discovery of the frog’s disappearance, sustained search for the missing frog, and eventual recovery of the frog.

Units of analysis The basic unit of a ‘‘clause’’ has been shown to be a highly relevant element for analysis of narrative texts. In our corpus, texts were divided into clauses in order to measure their length and the relative complexity of events and situations they include. Following Berman and Slobin (1994: 660–3) a clause was deﬁned as a unit of predication which expresses a uniﬁed situation, in the sense of a single activity, event, or state. Note that a predication here cannot invariably be identiﬁed with the category of (single) verb. In Hebrew, copula sentences in the present tense may contain no overt verb form, and yet constitute independent predications or clauses (e.g., ha-yeled Ø al ha-mita ve ha-cfardea Ø btox ha-cincenet ‘The-boy (is) on the-bed and the-frog (is) inside the-jar’). And in both languages, more than one verb might be included in a single clause, speciﬁcally: (a) in the case of direct or indirect speech, where the matrix verb is treated as a single clause together with its complement, i.e., the statement or question following the matrix verb; (b) aspectual and modal verbs are included together with the nonﬁnite verb which follows as a single predication; and (c) when the predicate consists of two verbs indicating a repeated action, it is also counted as a single clause, as in example (1), where the square bracket ] indicates clause boundary. (1) Daniel se ﬁjó y se ﬁjó ] y no encontró Daniel looked and looked ] and didn’t ﬁnd . . . [Nicole, girl, 7;6].

Linguistic forms Three linguistic categories were analyzed: one which is very similar in Spanish and Hebrew (relative clauses), one which is partially similar in the two languages (null subjects), and one which diﬀers completely (grammatical aspect). T/M/A distinctions are much more elaborately marked in the grammatical system of Spanish than Hebrew: verbs are inﬂected for perfective versus imperfective in past tense, as well as for progressive and perfect aspect in present, past and future tense, and for conditional and subjunctive mood. Hebrew verbs, in contrast, are inﬂected only for inﬁnitive and imperative mood and for tense (present which is also participial, and represents both

 Judy Kupersmitt and Ruth A. Berman

immediate and generic present; past tense — used for both durative and telic anteriority and also for irrealis conditionals; and future tense — which also serves for a variety of irrealis moods).3 For present purposes, all markings of tense/aspect were counted, but analysis focuses on those categories that serve to express the narrative functions of durativity and foreground-background distinctions. Durativity implies that the action described by the verb is extended in time, so that attention is not given to the beginning or the end of the activity. Foreground-background distinctions are essential in constructing a narrative text, since they mark oﬀ plotline events from their associated circumstances (Berman and Slobin 1994: 6–9; Hopper 1979; Reinhart 1984). The foreground of a narrative is provided by the ordered sequence of events that constitute the main plot line, typically expressed by verbs of achievement and accomplishment in perfective aspect (in the past), while background is provided by collateral, descriptive material, typically expressed by states or, in a language like Spanish, by activity verbs in imperfective aspect (Fleischman 1990). The example in (2) is taken from our data base, and illustrates how in Spanish, imperfective aspect is used to indicate durative events. (2) Todas las noches Daniel se ﬁjaba y miraba si la rana estaba en su lugar. Every night, Daniel looked-ifv and looked-ifv if the frog was-ifv in its place. [Nicole, 7;6]

Imperfective aspect is also used to indicate background as opposed to foreground material, typically with verbs which describe physical states or which express cognitive or aﬀective inner states and states of modality, as illustrated in (3). (3) Después se subieron a una casita, no sabían de quién era. Then they climbed-pfv to a little house, they did not know-ifv whose it was-ifv [Michal, girl, 5;11].

Since Hebrew lacks this rich system of aspect, narrators might be expected to use other syntactic and discourse devices in order to mark these functions, for example, tense shifting, lexical repetition, or expression of ‘‘extended aspect’’ by means of aspectual verbs. The next two examples are typical of Hebrew texts in which repetition and tense shifting are used as a means of indicating durativity and foreground-background distinctions respectively: (4) Ve hu rac, rac, rac, rac, ve hu maca sham exad, katan shelo. And he ran, ran, ran, ran, and he found there one, a little one of his. [Roni, boy, 4;4]

Spanish–Hebrew children’s narratives 

(5) Hu tipes al ha-sela, lir’ot me’al ha-sela mi-raxok im ha-cfardea nimcet sham. He climbed on the-rock, to-see from-above the-rock from-afar if the frog is there. [Eyal, boy, 6;2]

Note, however, that it is misleading to identify these devices with the use of the grammatically marked aspectual distinctions illustrated for Spanish. In (4) the lexical repetition of an inherently durative verb serves to express protracted aspect, while in (5) the switch from past tense in the main clause to present tense in the complement clause is the way in which Hebrew expresses the sequence-of-tense or ‘‘relative tense’’ function of aspect, in this case denoting the two events as simultaneous. Subject ellipsis is deﬁned as the licensing of the omission of surface subject pronouns in all types of clauses. Both Spanish and Hebrew allow the omission of surface subjects. However, while Spanish is a ‘‘pure’’ pro-drop language in that it allows the omission of surface subject pronouns in all types of clauses (Borer 1984; Hyams 1989; Weissenborn 1992), Hebrew does not behave in the same fully consistent fashion (Armon-Lotem 1996; Berman 1990, 1997). Surface subject pronoun omission is grammatically licensed in isolated clauses in past and future tense, but not in the present, and in ﬁrst and second person, but not in third person. When constructing a text, bilingual children are assumed to exploit the use of null subjects according to what is permitted in the grammar of the language they are using. In Spanish, this means obligatory subject ellipsis in both coordinated and subordinated clauses with the same subject, while in Hebrew it is optional in both cases. This contrast is illustrated in examples (6) and (7), taken from a 5-year old Spanish and Hebrew text respectively (with null subjects indicated by Ø): (6) Y después el niño lo agarró y Ø le hizo’’nu, nu, nu’’. Después Ø se fueron a buscar. And then the boy caught it and scolded+3-sg ‘‘nu, nu,nu.’’ Then went+3pl to look. [Michal, girl, 5;11] (7) axarkax ha-yeled ve ha-kelev hodu la cfarde’im ve hem xazru im hacfardea ha-haktana shelahem ha-bayta. Then the boy and the dog thanked+3pl the frogs and they returned+3pl home with their little frog to their home. [Michal, girl, 5;11]

Examples (6) and (7) illustrate coordinated clause structures, where the use of a null subject is obligatory in Spanish but optional for Hebrew. This is reﬂected

 Judy Kupersmitt and Ruth A. Berman

in (6) and (7) respectively, since the same narrator uses a null subject in Spanish (as is obligatory in cases of coreference), but opts for an optional overt subject pronoun in Hebrew with ‘they’ coreferential to ‘the boy and the dog’. Use of subject ellipsis is analysed in the form/function perspective of this study as a means for expressing the narrative functions of discourse connectivity and topic maintenance. Discourse connectivity is achieved by omission of the subject pronoun in coordinated and subordinated clauses with the same subject as their related main clause. Topic maintenance is characterized by the subject pronoun being omitted across clauses, adjacent or distant. Relative clauses, the third type of construction analysed here, are largely similar in Spanish and Hebrew. They are invariably postnominal, and they are marked in both languages by an obligatory, invariant complementizer (que in Spanish and še in Hebrew), which is not marked for categories such as gender, number, or animacy (Dasinger and Toupin 1994; Slobin 1990). Moreover, as in other languages, they serve primarily to express several main functions in narrative discourse. These include, ﬁrstly, presentation or characterization of a referent (typically the main protagonist), as in example (8): (8) Había un nene que tenía una rana y un perro. There was a boy that had a frog and a dog. [ Yael, girl, 7;4]

A second function is setting up expectations about narrative entities and events, as in (9): (9) Y el nene sabe lo que hay ahí. And the boy knows what (that) is there. [Roni, boy, 4;4]

Third, relative clauses are used in contexts of temporal retrospection, referring back to earlier episodes or events in the story, as in (10). (10) hem amru ‘‘atem yexolim lehaš’ir lanu et ha-cfardea še macanu?’’ They said ‘‘Can you leave-pl us the frog that found-1pl = we found ?’’ [Michal, girl, 5;11]

Finally, continuative relatives can be used for advancing the narrative, moving from one event to a sequentially subsequent event, as in the following example from a monolingual adult Spanish text, taken from Dasinger and Toupin (1994: 471). El niño cayó en la cabeza del alce, que, aterrorizado, corrió ‘The boy fell on the head of a deer, that, terriﬁed, ran’. As in their crosslinguistic study of monolingual ‘‘frogstory’’ narratives, we found that relative clauses used for a continuative function are restricted to more mature adult texts.

Spanish–Hebrew children’s narratives 

Results Findings are presented for analyses of overall narrative structure, formfunction relations and occurrence of errors.

Narrative structure Text length Length of texts, as noted, was analysed by number of clauses. We expected text-length to increase with age, as was found for the diﬀerent languages in the crosslinguistic monolingual samples, especially comparing the Hebrew and Spanish texts produced by preschoolers aged 3 to 6 years with those of schoolchildren aged 9 to 12 years (Berman and Slobin 1994: 31; Berman 1988). We further expected the Hebrew texts to be equal to, and maybe even longer than the Spanish texts mainly among the older subjects, for whom the L2 Hebrew has become the dominant language during their school years. Results of the breakdown into clauses is shown in Table 2. Table 2. Mean number of clauses per text in Spanish and in Hebrew, by age Age

Range of clauses Mean Spanish

Hebrew

4;0 6;0 7;5 11;2

12–20 16 24–33 28.5 24–37 30.5 35–7 35

11–17 13.5 17–28 22.5 14–27 20.5 23–31 27

Table 2 corroborates the ﬁrst prediction: with age, text length increases gradually in both languages, and both the Spanish and Hebrew parts of the narratives are on the average longer for older than for younger children. The latter tend to produce less complex syntactic structures at the local level, and texts which are less elaborated at the global level, containing fewer descriptive details and evaluative comments than the older, schoolage children. This ﬁnding may be attributed to several factors. In general, number of clauses per text is aﬀected by two main variables: the structure of the clauses at a local level, and overall content of the text at a more global level. As suggested earlier, these two factors appear to be interrelated: younger children produce both simpler syntactic structures and less thematically and descriptively elaborated (monologic) narratives than more mature and proﬁcient narrators.

 Judy Kupersmitt and Ruth A. Berman

And ﬁndings of prior research suggest that these variables are languageindependent, being determined by a combination of linguistic, cognitive, and communicative maturation. On the other hand, the second prediction is not conﬁrmed: contrary to what was expected, the Spanish texts are longer than the Hebrew. Moreover, this diﬀerence between average length of text in the two languages also increases with age; that is, the schoolchildren’s Spanish texts are relatively longer than their Hebrew texts than are those of the younger children. This result is evidently due to the particular task and the nature of the two diﬀerent parts of the story elicited in each of the two languages in this study: the ﬁrst 12 scenes (the Spanish part) with background setting or orientation, initiating event, and three ‘‘search episodes’’ and the remaining 12 scenes and resolution in the second, Hebrew part. Although both halves have the same number of pictures, the ﬁrst half covers more events. In terms of content, the second half of the story can be characterized as having fewer elements rating high on semantic salience of the kind associated with a hierachy of narrative ‘‘importance’’ (Bamberg and Marchman 1991; Marchman 1989).

Narrative organization Analysis of our data reveals a parallel development of ‘‘action structure’’ in Spanish and Hebrew storytelling. This is illustrated by the opening (11a) and closing (11b) segments of the texts produced by a 4- and a 10-year-old bilingual in examples (11–1) and (11–2) respectively, translated into English from the original Spanish and Hebrew. Note that these excerpts relate to the beginning and ending parts of the story, its exposition, orientation, or ‘‘scenesetting’’, on the one hand, and its resolution, closure, or coda, on the other, which are recognized by all research on narrative structure, whether in more formal ‘‘story-grammar’’ analyses or the ‘‘content-oriented’’ approach associated with Labov and his followers, as crucial elements of narrative-discourse structure. (11) Excerpts from two bilingual texts translated into English describing the story-beginning (setting) in Spanish and story-ending (resolution) in Hebrew 1. Preschooler — Roni, boy, aged 4;4 [nursery-school, prekindergarten]: a. There was a boy that had a dog and a frog, and at night there was a moon. And here is a little bit night, and the dog woke up, and the boy . . . and they woke up.

Spanish–Hebrew children’s narratives 

b. And they climbed up to this, until they climbed up to this. And, the frogs laughed: ‘‘ What is this boy doing here?’’ 2. Schoolchild — Shirli, girl, aged 10;1 [grade-school, 6th grade]: a. Once upon a time, there were a boy, a dog and a frog, and the three of them were very good friends. One day, the dog and the boy were sleeping, and the frog, very slowly, escaped. Nobody knew why, and nobody knew where. Then, in the morning, the boy and the dog woke up, looked into the frog’s house, and looked, and it was not there! b. Suddenly, the boy heard something. Silently, he carried the dog and said to him: ‘‘Come, I can hear something’’. They looked at the other side of the log and saw their frog together with one, two, three, four, ﬁve, six, seven, eight, nine other frogs. They took a little one and said: ‘‘Bye, see you tomorrow!’’

These two examples clearly show that narrative structure has universal, or at least language-independent properties in cultures with shared narrative conventions such as the two backgrounds and socio-cultural contexts characteristic of the children in our population. As has been demonstrated for monolingual ‘‘frogstory’’ as well as other types of (picture-based) narrative texts in a variety of languages (e.g., Katzenberger 1994; Kern 1997), here, too, younger children tended to treat each scene as an isolated event, describing the contents of each picture at a local level. In terms of interclausal connectivity, younger, preschoolage children will at most sometimes chain events linearly by means of connectors like (and) then (Hebrew (ve) az, Spanish (y) después). They do not provide relevant temporal, spatial, or motivational scene-setting, nor do they make explicit reference to the initiating event which sets in motion the chain of events which make up the story (the frog’s disappearance, and the boy’s cognizance thereof). They also fail to provide a resolution which encodes an explicit causal connection with the rest of the story, but instead give a separate, unconnected description of the contents of the ﬁnal pictures in the storybook. Around the age of six, children begin to link between events sequentially to create a more explicit and tighter temporal organization of the text. Subsequently, narrative texts are causally structured in a way which expresses a hierarchical organization at the level of overall, global action structure. As shown in example (11–2), older children combine their clauses into a wellconstructed ‘‘action structure’’ including explicit marking of temporal and causal connections and evaluative comments. Their opening clauses provide

 Judy Kupersmitt and Ruth A. Berman

all relevant information about the characters, the temporal and spatial setting, motivational background, and the event initiating the complicating action. And not only is their resolution clearly connected to the complicating action which constitutes the backbone of the story, it typically also contains overt reference to the event immediately preceding the resolution, i.e., the ‘‘high point’’ of the story (in this case, reached when the boy and the dog climbed up the log and see a family of frogs behind it) and often also an explicit ‘‘coda’’ expressing closure, and relating the entire series of events of the story to the present situation and the narrator’s point of view. In sum, across our small corpus of texts, younger preschool children were clearly diﬀerentiated from the older, schoolage subjects in terms of the way in which they gave expression to a relatively less or more well-developed command of a hierarchical level of global narrative organization.

Linguistic forms and narrative functions The discussion in the preceding section indicates that no diﬀerences were detected in the organization of the texts as a function of language change. A markedly diﬀerent picture emerges, however, when it comes to the use of linguistic forms, in this case, the categories of tense/aspect, subject ellipsis and relative clauses. These forms are analyzed in terms of quantitative and qualitative distribution and how they function as rhetorical devices in each of the two languages, from a developmental point of view. In this context, the term rhetorical devices refers to ‘‘the way in which speakers organize the information which they present in ongoing construction of a text by means of the selection of appropriate options out of the grammatical and lexical devices available in the target language’’ (Berman 1999). The notion thus critically refers to speakers’ choice of relevant linguistic devices (prosodic, morphological, syntactic, and lexical) to meet such functions as narrative cohesion — by lexico-syntactic marking of interclause and intersentential relations; narrative perspective and selection of agent versus patient focus — through alternations of voice and valency; and background/foreground distinctions — inter alia by deployment of tense-aspect switches to alternate between informative, descriptive, or evaluative background material and the sequentially ordered foregrounded narrative or referential elements.

Tense/Aspect We predicted that the Tense/Aspect system would reveal marked diﬀerence

Spanish–Hebrew children’s narratives 

between the two target languages, with Spanish texts displaying a wide range of aspectual forms in both present and past tense and Hebrew texts having an impoverished repertoire of tense choice only, as determined by typology of each language. Tables 3 and 4 summarize the distribution of tense/aspect forms used in the Spanish and Hebrew renderings of the frog story (rather, the ﬁrst and second part thereof respectively), deﬁned across ‘‘lexical verbs’’, that is, all verbs other than the equivalents of copular be or have and other auxiliaries in the two languages. Tables 3 and 4 show that the bulk of lexical verbs in both languages are in some form of Past Tense (around 90 per cent across age-groups), but in Spanish this is divided between Perfective and Imperfective aspect, yielding three favoured forms compared with two in Hebrew. Comparing the Spanish bilingual texts in our sample with Spanish texts produced by monolingual children from Spain and South America (Sebastián and Slobin 1994; Bocaz 1989) reveals marked diﬀerences that can be attributed to the impact of Hebrew typology in two related domains: morphology and narrative temporality. At the morphological level, Spanish bilingual texts contrast sharply with the Spanish monolingual samples in which narrators from as young as age three years made use of almost every Tense/Aspect combination available in Table 3. Percentage of Tense/Aspect inﬂections (excluding Future) in Spanish by age T/A Categories

Pres. (Simple)

Pres. Prog.

Ifv. Prog.

Pfv. Prog.

Ifv.

Pfv.

Past Perf.

4;0 6;0 7;5 11;2

18 26 5 5

0 0 0 2

0 0 0 0

0 0 0 0

25 11 31 14

57 63 64 79

0 0 0 0

Total

14

0

0

0

20

66

0

Pres.=Present, Pres. Prog=Present Progressive, Ifv. Prog.=Imperfective Progressive, Ifv.=Imperfective [Past], Pfv.=Past Perfective, Past. Perf.=Past Perfect

Table 4. Percentage of Tense/Aspect inﬂections (excluding future) in Hebrew, by age T/A Categories

Present

Past

4;0 6;0 7;5 11;2

8 4 16 4

92 96 84 96

8

92

Total

Present includes participial, immediate and generic

 Judy Kupersmitt and Ruth A. Berman

their language.4 From the point of view of narrative temporality, the diﬀerence lies in the choice of dominant or anchor tense for the narrative. As opposed to Spanish monolingual narrators, who quite often anchor their (picturebookbased) stories in the present tense, the dominant tense among Spanish–Hebrew bilingual narrators is clearly the Perfective, corresponding to the (simple) past tense in Hebrew, the prototypical form for Modern Hebrew narrations (Berman and Neeman 1994). Only one subject (a 6-year-old) changed the tense in which he anchored his story as a function of language change: his Spanish text is anchored in the present (except for the typically punctual verb cayó ‘fall’ in past tense), while the Hebrew part is in the past tense. This is shown in example (12), where ## indicates change in language. (12) Todas las abejas lo siguen al perro y el nene lo busca, lo busca en un pozo que está en un árbol, y todas las abejas lo siguen todavía al perro, y el nene se cayó del árbol, y el nene se cayó por un búho ## az ha-yeled hitxabe me axorey ha-sela, hu tipes al ha-sela, lir’ot me al ha-sela miraxok im ha-cfarde’a nimcet sham. Az ha-kelev hitxabe me axorey ha-sela, ve ha-yeled be ta’ut ala al ayal. ‘All the bees go-pl after-pres the dog and the boy looks-pres for it, looks-pres for it in a hole that is-pres in a tree, and all the bees still go-pl after-pres the dog, and the boy fell-pfv from the tree, and the boy fell-pfv because of an owl ## then the boy hid-past behind the stone, he climbed-past up the stone, to see from above the stone in the distance if the frog is-pres there. Then the dog hid-past behind the stone, and the boy accidentally got-past on a deer’ [Eyal, boy, 6;2]

Note that the 4-year-old children, the youngest in our sample, used Present Tense relatively more in their Spanish texts than any of the other age-groups. Though our sample is too small to draw any ﬁrm conclusion, this may be due to the fact that these children are still in the ‘‘picture-description’’ mode of storytelling, and because at this age, children still do not rely on a single dominant tense for anchoring their narrations.5 Rather, they switch between past and present from clause to clause, with these switches not motivated by either syntactic or discourse factors, as illustrated in (13). (13) Y el nene se fué, y el nene sabe lo que hay ahí. Ahí había un b ú ho! Y aca lo busca, y el perro se escapó.6 And the boy left-pfv, and the boy knows-pres what is-pres there. There, as-ifv an owl! And here (he) looks-pres for it, and the dog escaped-pfv’ [Rony, boy, 4;4]

Spanish–Hebrew children’s narratives 

Especially interesting is the use of the Imperfective (in Spanish). Table 5 shows the forms used in the Imperfective, by type of predicate (atelic statives versus dynamic activities, telic and atelic). As shown in the table, the children in our sample use the Imperfective overwhelmingly with stative predicates, which refer to physical and internal aﬀective or cognitive states and conditions of modality, as opposed to events that build the main plotline — which are typically in the Perfective in monolingual Spanish narratives. Table 5. Distribution of Spanish imperfective forms in percentages, by type of predicate and age Age-group

Stative

Dynamic

4;0 6;0 7;5 11;2

100 85 77 70

0 15 23 30

83

17

Total

The ﬁgures in Table 5 reﬂect a general tendency in Spanish for stative verbs, which are inherently durative, to take Imperfective Aspect. However, this identiﬁcation of statives with imperfective marking is not grammatically obligatory, since some stative predicates, particularly modals, can occur in the Perfective. Besides, the tendency for children in our bilingual sample to use Imperfective marking only for stative predicates is clearly age-related, occurring exclusively among the youngest preschoolers, going down to threequarters to two-thirds (70 per cent) among the older children. The most frequently used verbs in this form in our sample are estar ‘be’, poder ‘be able to’, querer ‘want’ and saber ‘know’, as illustrated in (14). (14) El nene estaba afuera, y el perrito quería subir y no sabe. Y el nene subió. The boy was-ifv outside, and the doggie wanted-ifv to-climb, but not knows-pres. And the boy climbed-pfv [Nili, girl, 3;6]

This ﬁnding appears to support the claim of Slobin and Bocaz (1988) that ‘‘preschool children seem to be guided by Aktionsart (lexical aspect) in their choice of aspectual form’’. This would predict that young children use the Imperfective with verbs that are inherently durative, as are stative verbs. Other studies, however, have demonstrated that children as young as age 3;6 do not choose aspectual form only by Aktionsart but also by a well-motivated need to

 Judy Kupersmitt and Ruth A. Berman

contrast between ongoing versus punctual activities in a narrative context (Sebastián and Slobin 1994; Weist, Wysocka and Lyytinen 1991). As noted, use of Imperfective aspect with dynamic predicates increases with age, but still remains much lower than with verbs of state (30 per cent vs. 70 per cent with verbs of state among the 11-year-olds). When Imperfectives are used with nonstatives, the narrator clearly indicates a contrast between ongoing, background states or activities and plot-advancing events, as illustrated in examples (15) and (16) from a 7- and a 10-year-old respectively. (15) Todas las noches Daniel se ﬁjaba y miraba si la rana estaba en su lugar. Después se iba a dormir. Hasta que un día pasó, que el nenito se levantó y no vió la rana en su lugar. Every night Daniel looked-ifv and looked-ifv if the frog was-ifv in its place. Then he would-go-ifv to sleep. Until it happened-pfv that one day, the boy woke up-pfv and not saw-3-pfv [=he didn’t see] the frog in its place. [Nicole, girl, 7;6] (16)

Un día, el perro y el nene dormían, y la rana, muy despacito, escapó. One day, the dog and the boy slept-ifv, and the frog, very quietly, escaped-pfv. [Shirly, girl, 10;1]

Examples (15) and (16) are typical instances of aspectual contrasts in the past tense, where the narrator sets up the scene-setting background before moving forward to the next event, and in so doing sets oﬀ sequentially ordered plotline events (encoded by what Labov termed ‘‘narrative clauses’’) from their associated circumstances and the temporal, spatial, and motivational states surrounding them. Overall, use of the Imperfective diﬀers from the Spanish monolingual sample (Sebastián and Slobin 1994). Spanish–Hebrew bilinguals appear to show little ﬂexibility in deciding to employ Imperfective aspect for the purpose of backgrounding as a more global, narrative device, and this nonalternation of grammatical aspect is particularly marked after age 5. This lack of variability and ﬂexibility makes the bilingual children’s Spanish-language texts more linearly constructed, since they show less in the way of narrative foreground-background distinctions, as illustrated in example (17) from a 5-year-old girl: (17) Y después las abejas que salieron de ahí lo picaron. Después buscaron en todos lados y después buscaron en otro lado y no encontraron. Después el perro miró en cada árbol. Después el niño miró adentro.

Spanish–Hebrew children’s narratives 

And then the bees that came out-pfv from there stung-pfv him. Then looked-3-pl- pfv in every place and then looked-3-pl-pfv in another place and not found-3-pl-pfv. Then the dog looked-pfv in every tree. Then the boy looked-pfv inside. [Michal, girl, 5;11]

More evidence of non-nativelike use of the Imperfective is shown in (18): (18) Y la buscaban y no la encontraban adentro del zapato y el perro no la encontraba adentro de donde estaba. And looked-3-pl-ifv for-it and not found-3-pl-ifv it inside the shoes and the dog not found — pl-ifv it in the place where it was-ifv [Dror, boy, 6;2]

In the example above, the imperfective aspect is inappropriately used with the event of ‘ﬁnding’ encontrar, where the adverbial expressions adentro del zapato ‘inside the shoes’ and adentro de donde estaba ‘in the place where it was’ force a punctual interpretation of the event. In contrast, Spanish monolingual children as young as age 4 show proﬁcient and well-motivated use of the Imperfective. The following example from a 4-year-old, describing the same scene as in (17), shows how the Imperfective/ Perfective contrast is used to diﬀerentiate between an ongoing situation salían ‘went out-ifv’ and a situation which is completed salieron ‘went out-pfv’: (19) Y acá salían muchas abejas y ahí salieron todas. And here went out-ifv many bees and there went out-pfv all of them. [as04;02b]7

Example (19) shows that this Spanish-speaking 4-year-old is already able to take two diﬀerent perspectives on two events described by the same verb, by using an aspectual contrast to encode an event either as a process or as completed. The relatively low use of Imperfective aspect and other varieties of Tense/ Aspect marking in the Spanish bilingual texts, coupled with the decline in use of the Imperfective with age, indicates a deviation from the discourse devices typical of monolingual Spanish narrators. This may be due to increasing avoidance of devices which are markedly diﬀerent in linguistic typology from the dominant language, Hebrew. As for the Hebrew parts of their texts, the question arises whether bilingual speakers, in this case, speakers of Spanish and Hebrew, will try to ‘‘compensate’’ for the lack of certain grammatical distinctions in one of their

 Judy Kupersmitt and Ruth A. Berman

languages. A similar question was addressed in the crosslinguistic study by Berman and Slobin (1994). They concluded that monolingual speakers of diﬀerent languages do not use ‘‘compensatory’’ linguistic devices to mark certain distinctions, like Tense and Aspect, that are grammatically overt in other languages (Berman and Neeman 1994; Slobin 1995). Nonetheless, in the case of bilingual children, it is tempting to assume that since they are exposed to two diﬀerent ways of talking about events (if not necessarily of experiencing them), they might seek alternative linguistic devices to express certain notions already framed in their minds by the categories of their other language. This would mean, for example, that the Spanish–Hebrew bilinguals in our sample might ﬁnd other ways, outside of verb-internal grammatical inﬂections, in order to express notions such as durativity and background-foreground distinctions in their Hebrew narrations. But this prediction is not clearly corroborated by our data. Overall, the bilinguals’ use of other discursive and syntactic devices to express aspectual contrasts does not diﬀer from that of monolingual speakers of Hebrew (Berman 1988; Berman and Neeman 1994). The most common devices we found in our bilingual Hebrew texts are closely parallel to those in the monolingual Hebrew corpus: tense shifts between past and present, expression of ‘‘extended aspect’’ by means of aspectual verbs like those meaning start, go on, keep on, and other more typically Hebrew devices such as lexical repetition, and occasional morphological marking of aspect by verb-pattern alternation (e.g., rac ‘run’ versus hitrocec ‘run around’ in two diﬀerent verb-pattern conjugations). Tense changes were mostly motivated by syntactic ‘‘sequence of tense’’ requirements between matrix and complement clause to express the temporal relation of simultaneity between two events, as shown in example (20). (20) Daniel samax nora še yeš lo mišehu leyado, Daniel ra’a še hu, yeš lo kala. Daniel rejoiced-past that it has-pres company with-it, Daniel saw-past that it, it has-pres a bride. [Nicole, girl, 7;6]

In example (20), the tense shift is motivated by an obligatory sequence of tenses, in this case, expressing simultaneity between the state of having in the present tense and the prior activity of rejoicing and of seeing in the past tense. Interestingly, this kind of syntactically motivated tense-shifts from past to present were also found in the Spanish part of some of our bilingual texts, resulting in non-nativelike uses of sequences of tenses, as in (21) and (22).

Spanish–Hebrew children’s narratives 

(21) Y después se fueron a dormir y la rana sin que vean, se fué. Y vieron que la rana no está y la llamaron. [Dror, boy, 6;2] And then (they) went-pfv to-sleep and the frog without (them) seeing it, left-pfv. And (they) saw- pfv that the frog is-pres not (there) and (they) called-pfv her. (22) A la mañana el chico se despertó y vió que la rana no está.[Anat, girl, 12;4] In the morning the boy awoke-pfv and saw-pfv that the frog is-pres not (there)

This kind of sequence of tenses was not found at all in the Spanish monolingual corpus. To express simultaneity, monolingual speakers of Spanish would use the Imperfective after the past. The unacceptable tense combinations in (21) and (22) provide clear evidence for a kind of ‘‘merge’’ between Spanish and Hebrew discourse devices.

Subject ellipsis We assumed that the Spanish texts would evince more subject ellipsis than the Hebrew ones and, further, since Spanish allows the use of null subjects in a wider range of contexts, we assumed that null subjects would show a diﬀerent pattern of distribution in the two languages. Table 6 indicates the use of subject ellipsis in Spanish and Hebrew for purposes of discourse connectivity and topic maintenance. Table 6. Percentage of clauses with null subjects, by type of narrative function and by age Spanish

Hebrew

Age

Discourse connectivity

Topic maintenance

Discourse connectivity

Topic maintenance

4;0 6;0 7;5 11;2

15 19 22 24

8 17 12 16

10 27 14 26

0 3 7 0

% out of Total Nulls

60

40

88

12

The ﬁgures in Table 6 corroborate our expectations: the total proportion of subject ellipsis per number of clauses is greater in Spanish than in Hebrew, ranging from 23 per cent in the Spanish compared with 10 per cent in the

 Judy Kupersmitt and Ruth A. Berman

Hebrew 3-year-old texts to 40 per cent in the Spanish and 26 per cent in the Hebrew 11-year-olds. Moreover, also as expected, the functions of subject ellipsis are diﬀerently distributed across the two languages. While in both languages, children use subject ellipsis more for what we have termed ‘‘discourse connectivity’’ (in same-subject coordinated and subordinate clauses) than for ‘‘topic maintenance’’ (across independent clauses), the disparity between the two functions is far less marked in Spanish than in Hebrew. Thus, rather more and rather less than half of all null subjects (60 per cent versus 40 per cent) serve for each function respectively in Spanish; in contrast, an overwhelming proportion of the Hebrew null subjects (88 per cent) serve for the more local, grammatical function of discourse connectivity. This contrast between the Spanish and Hebrew texts is illustrated in (23) and (24) by the two excerpts from the narrative of the same child (Shirli, aged 10;1) in each of her two languages. (23) El nene salió muy muy furioso y el perro subió y lo chupó. Después Ø fueron al bosque y Ø lo buscaron a la rana. De repente Ø miraron la casa de las abejas, puede ser que Ø lo comieron!. Ø Preguntaron a un animalito que estaba ahí, Ø le preguntaron si Ø vió una ranita linda, roja. The boy came out very very upset and the dog got on and licked him. Then Ø went-pl to the forest and Ø looked-pl for the frog. Suddenly, Ø looked-pl at the bees’ house, maybe Ø ate-pl it! Ø asked-pl a little animal that was there, Ø asked — pl it if Ø saw-3sg a nice, red frog. (24) hem nitkeu aﬁlu betox ha-karnayim shel eyal ve ha-kelev himshix lexapes. ha-ayal daxaf et ha-kelev ve mamash Ø tas bimhirut raba, Ø dahar ve mamash kim’at Ø hipil et ha-yeled. hu hipil et ha-yeled ve basof hu hipil et ha-yeled ve et ha-kelev. They got-stuck-pl even in the horns of a deer and the dog kept on looking. The deer pushed the dog and really Ø ﬂew-3sg very quickly, Ø galloped-3sg and really nearly Ø threw-3sg the boy. He threw-3sg the boy and ﬁnally he threw-3sg the boy and dog.

Table 6 further shows that only 4 year-olds use subject ellision relatively very little (less than 10 per cent of their clauses) for the more global discursive function of topic maintenance in Spanish. This could be because young preschool children are still describing the episodes in the story in picture-bypicture fashion, and so tend to repeat the names of the protagonists at the beginning of each clause. This interpretation is conﬁrmed by the fact that in

Spanish–Hebrew children’s narratives 

Hebrew, the younger children tend to over-use surface subjects in an unacceptable fashion, as illustrated by the following example from our bilingual sample (and see, too, Berman’s (1990) analysis of a similar phenomenon in 4and 5-year-old monolingual Hebrew narratives). (25) ve ha-shinayim ha-ele, hem hayu shel karish. And those teeth, they were of a shark [Roni, boy, 4;4]

From a contrastive perspective, the bilingual and the monolingual Spanish data show a strikingly similarity in the use of subject ellipsis for a range of various narrative functions. Thus, developmentally, Spanish-speaking bilingual and monolingual children seem to follow a very similar trend when using null subjects in a narrative context. 5-year-olds already use null subjects across nonadjacent clauses with diﬀerent subjects (referents introduced in a previous clause). That is, they freely omit the subject pronouns for previously introduced characters, as shown in the following examples from our bilingual sample and from the monolingual corpus: (26) Y la rana sin que vean se fué, y Ø vieron que la rana no está, y Ø la llamaron. And the frog without (them) seeing it went away and Ø saw-pl that the frog is not (there), and Ø called-3pl it. [Dror, boy, 6;02] (27) Y cuando Ø se estaban durmiendo la rana salió y luego Ø se despertaron. And when Ø were-3pl sleeping, the frog left and afterwards Ø woke up3pl [S5e–5;4]8

Spanish-speaking bilingual children also seem to share with their monolingual counterparts the knowledge that overt pronouns are required in particular syntactic and discourse contexts usually for contrast or disambiguation as in (28) from a 12-year-old bilingual girl and in (29) from an 11-year-old monolingual: (28) Ø puso la rana en el frasco y el perro también se quedó en la pieza y él se fué a dormir. Ø [=the boy] put the frog in the jar and the dog also stayed in the room and he [=the boy] went to sleep. [Anat, bilingual girl, 12;4] (29) Mientras el perro olía el suelo para buscar rastros de la rana él pensó que era un árbol. While the dog was smelling the ground to look for any trace of the frog he [=the boy] thought that (it) was a tree.9 [as 11;05g]

 Judy Kupersmitt and Ruth A. Berman

In sum, with regard to subject ellipsis, bilingual children tend to behave much like monolinguals in both Spanish and Hebrew. Thus, in Spanish, subject pronouns are dropped across the board, except when deliberately mentioned for contrastive purposes by older children. In Hebrew, the use of subject pronouns reﬂects their optionality as a syntactic requirement. Thus, in the monolingual Hebrew sample, ‘‘narrators of all ages use subjects pronouns in one-third to one-half of all clauses’’ (Berman and Slobin 1994: 540).

Relative clauses Of the three linguistic systems analysed, we expected relative clauses to show the greatest degree of similarity in the two languages. Dasinger and Toupin’s (1994) crosslinguistic comparison of relative clause usage in frog-story narratives notes that Spanish and Hebrew speakers produce a higher rate of relative clauses than speakers of the other languages in their sample (English, Turkish, and German). Surprisingly, however, our results do not conﬁrm this ﬁnding, nor were they as we expected. Table 7 shows that the number of relative clauses used in our sample is clearly larger in Spanish than in Hebrew. Moreover, they show up earlier in the Spanish than in the Hebrew texts, although there is no marked rise with age in the number of relative clauses in either language. Table 7. Mean number of relative clauses and of total clauses in Spanish and Hebrew, by age French

Total

Age

4;0 6;0

7;5

11;2

RCs Mean # clauses

1 2 2 2 16 28.5 30.5 35

Hebrew 4;0

7 27.5

6;0

Total 7;5

11;2

0 1 0 0.5 13.5 22.5 20.5 27

1.5 21.5

Thus, the Spanish texts show a proportion of 25 per cent relative clauses out of overall mean number of clauses per text, compared with only around 7 per cent in Hebrew. These results may in part be due to a diﬀerence in content between the ﬁrst and the second part of the story, pointing to one of the disadvantages of this sort of elicitation procedure, as noted earlier, since use of certain linguistic devices might be content driven and content aﬀected. Yet it is clear that while children use relatively few relative clauses in both languages, there are consistently more both in absolute and in proportional terms in the Spanish than in the Hebrew texts.

Spanish–Hebrew children’s narratives 

The main function of relative clauses in the Spanish texts was to present or describe a new referent just introduced into the narration, as in (30). (30) El perro quiere saltar a la rama que está en el árbol. The dog wants to jump over the branch that is on the tree [Eyal, boy, 6;2]

In developmental terms, children ﬁrst use relative clauses in a narrative context for presentation rather than for other functions in Spanish as in other languages (Dasinger &Toupin 1994; Sebastián and Slobin 1994). Here, context dependency could be playing a role in the distribution of relative clauses in the Spanish compared with the Hebrew texts. Recall that the children all began the story in Spanish, so they might be relying more on relative clauses at this initial point of the narrative, where characters and objects are ﬁrst being introduced. On the other hand, in keeping with our prediction of similarity between relative clause usage in the two languages, both in Spanish and Hebrew, children as young as age 6 produced relative clauses that served several diﬀerent narrative functions. This is shown in the examples in (31) and (32) from the texts in both languages produced by Michal, aged 5;11. (31) Y después, las abejas que salieron de ahí lo picaron. And then, the bees that came out from there stung him [= the boy] (32) hem amru ‘‘atem yexolim lehaš’ir lanu et ha-cfardea še macanu?’’ They said ‘‘Can you leave us the frog that we found ?’’ [Michal, girl, 5;11]

In example (31), the relative clause has the ‘‘continuative’’ function of moving the narrative forward, linking the two sequential events of the bees coming out of the hive and then stinging the boy, while in (32) the relative clause serves the function of retrospection, referring back to earlier episodes or events in the story, in this case to the ﬁnding of the frog. Thus, overall, ﬁndings for the bilingual Spanish–Hebrew sample is largely consistent, as expected, with what was found for narratives produced by monolingual children in these two languages, with respect to the construction and functioning, although not the overall distribution, of relative clauses.

Error analysis The hypothesis was that the grammaticality of the linguistic forms used by our subjects would be aﬀected by their relative command of the two languages. As

 Judy Kupersmitt and Ruth A. Berman

expected, we found far more grammatical errors (deﬁned along the lines detailed in Section 1 above) in Spanish than in Hebrew: more than 4 times the number of errors in Spanish compared with Hebrew (22 vs. 5). Their distribution also seems to support the hypothesis that the number of errors in Spanish would increase with age: we found three errors among the 4-year-olds versus eleven errors among the 11-year-olds. This could be because, as Hebrew dominance becomes more established, along with a developmental push for more complex narrative structure and expression, the older bilinguals produce what appears to be less proﬁcient Spanish, when proﬁciency is deﬁned as amount of error in ‘‘core’’ grammar. Further quantitative data are not presented owing to the small number of subjects in the sample. The analysis instead relates to trends observable in our database which will be examined in more depth on a larger group of subjects (Kupersmitt, in progress). According to our prediction, analysis of errors revealed three types of errors along the lines detailed in Section 1 above: errors of fossilization, transfer errors, and developmental errors. Recall that we expected to ﬁnd the ﬁrst and the second type in the Spanish texts and the third type in both texts. Our ﬁndings corroborate this expectation. Errors of fossilization occurred in the Spanish texts from age 6 on, in the following contexts: (a) in clitic constructions as in lo buscaron a la rana ‘it-acc-masc looked-pl for the frog’ or el perro lo busca en el árbol ‘the dog it-acc-masc looks for in the tree’, where Masculine lo is used instead of la to refer to rana ‘frog’ (Feminine) and (b) with the reﬂexive particle se as in y (la rana) no se podía salir ‘and (the frog) rﬂ-could not go out’ or después se salieron afuera ‘then (they) rﬂ-went out’ where the particle se is inappropriately used with the verb salir ‘go out’, analogously to the correct use of this particle with other verbs of motion like bajar ‘get down’ or subir ‘climb’, in which case it indicates completion of activity or episode boundary (Sebastián and Slobin 1994).10 Interestingly, errors of this kind occur in selected grammatical constructions: those that are characterized as ‘‘breaking points’’ in the grammar, that is, diﬃcult, opaque, or vulnerable subsystems as well as multi-functional grammatical devices (Berman 1992; Karmiloﬀ-Smith 1979; Ravid 1995; Slobin 1977). Thus, clitic pronouns are an instance of structural opacity since their morphology varies according to case and gender: the accusative forms are lo-masc, la-fem and their plurals los and las respectively, but the dative form is le as in le dijo al perro ‘iti-dat told to-the dogi’ (les in the plural). These combinations turn out to be diﬃcult even for native speakers of Spanish,

Spanish–Hebrew children’s narratives 

leading to what is known as le-lo-ísmo ‘le-lo-ism’ (Alarcos Llorach 1995; López Ornat 1994). One reason is the diﬃculty involved in ‘‘backwards’’ agreement, where the lexical noun is discontinuous with respect to the preceding (‘‘antecedent’’) pronoun. The use of the particle se is complex as well. It serves as a reﬂexive pronoun, in which case it agrees in number with the subject and is used with the following functions: (a) as a reﬂexive marker, to indicate a direct or indirect object co-referential with the grammatical subject (e.g. se lava las manos ‘Tohim/herself washes the hands’); (b) as a marker of reciprocality (e.g., los dos amigos se separaron ‘the two friends from-each-other parted’); (c) to indicate inception in change-of-state verbs such as sentar-se ‘sit down’ or levantar-se ‘wake up’; (d) as a marker (optional) of completion with verbs of motion as in se bajó ‘got down’ and se subió ‘climbed up’. In double clitic constructions, it agrees in person and number with the object, e.g., Se lo dió al niño ‘to-himdat it-acc gave to-the boy’; Te lo dió a tí ‘To-you-dat it-acc gave to-you’. This same multifunctional particle is invariant in number when used in impersonal passives (e.g., se espera un aumento de precios ‘(it) is expected a rise in prices’, se busca secretaria ‘(it) is wanted a secretary’; as a marker of unaccusative-type verbs (e.g., se cayó ‘fell (down)’, se rompió ‘broke’). In sum, the grammar of our subjects seems to get ‘‘stuck’’ at certain breaking points in the language where errors in selected grammatical categories become fossilized as such. These same type of errors can be found in texts produced by young monolingual speakers of Spanish as well as in naturalistic data from young Spanish acquirers (López Ornat 1992). However, more data is needed in both Spanish monolingual and bilingual usage, in order to conﬁrm the hypothesis that these errors are equivalent to those found in monolingual early language development or, conversely, in order to claim that we are dealing with errors which are exclusive to bilingual language development. The second type of error found in the texts were errors of transfer. As we predicted, these occurred in the Spanish excerpts from age six on. Errors of this kind involved the following two processes: (a) the replacement of a form in language X with a form from language Y and (b) loss of a form in language X which does not exist in the system of Y — the latter in line with Weinreich’s notion of ‘‘neglect’’ or ‘‘elimination’’ of obligatory categories (1974: 30). Almost all cases of (a) included wrong use of prepositions, e.g., se ﬁjaron/ salieron de la ventana corresponding to Hebrew histaklu/yacu me ha-xalon ‘(they) looked/went out from [=through] the window’ for se ﬁjaron por/a través de la ventana; se subieron adentro de árboles corresponding to alu le-tox

 Judy Kupersmitt and Ruth A. Berman

ecim ‘(they) climbed into trees’ for the Spanish se subieron a los árboles. In the last case, Hebrew is similar to English in that the preposition expresses two dimensions in the movement trajectory (the path and the end of the path), but diﬀerent from Spanish where the preposition a expresses only one dimension of this trajectory (Slobin 1996). The likelihood that prepositions will be susceptible to inﬂuence from the dominant language derives from the fact that these are highly language-speciﬁc lexical elements, which express diﬀerent perspectives in the organization of concepts such as movement and space (Bowerman 1985; Slobin 1996). Another error of transfer was redundant use of the relative clause marker, e.g., salieron afuera adonde que estuvieron el otro día (cf. adonde estuvieron), corresponding to Hebrew yacu haxuca eyfo she hayu beyom axer ‘(they) went-outside where that they were the other day = (at the place) where they had been’. Cases of ‘‘elimination’’ of obligatory markers were scarce and are best illustrated by the omission of plural deﬁnite markers as in el niño buscó en botas following the Hebrew ha-yeled xipes ba-magafayim ‘the-boy looked inthe-boots’, or subieron adentro de árboles corresponding to alu la-ecim ‘climbed-pl to-the-trees’, where the Spanish nouns botas ‘boots’ and árboles ‘trees’ should be preceded by las ‘the-fem-pl’ and los ‘the-masc-pl’ respectively. In Hebrew, the deﬁnite marker ha ‘the’ diﬀers from Spanish in two ways: (i) it incorporates into the bound prepositions be ‘in’ and le ‘to’ as ba and la respectively, and (ii) it is invariant with respect to number and gender (Berman 1985). These last examples could be instances of what is known in the literature as ‘‘language mixing’’ (Pfaﬀ 1979) or ‘‘interlanguage’’ usage (Selinker 1974) at the level of lexical-semantic, morphological and syntactic processing, in which case, for example, the syntax of the dominant language may be applied when using the weaker language. Finally, we predicted that we would ﬁnd developmental errors mainly among the younger children in both languages. This prediction was partially conﬁrmed. In Spanish, developmental errors included one regularization in forming the simple past tense: pudía for podía ‘could’ from the present tense form puede ‘can’11 and one error in the use of ser/estar (a dual-form copula corresponding to the verb ‘be’ in English): y acá está un poquito de noche ‘and here (it) is a little bit night’, using está from estar, in place of es from ser. The use of ser versus estar is said to be constrained by semantic oppositions (e.g., ser is used to denote permanence while estar is used to denote transient situations) as well as by syntactic contexts (e.g., the choice between ser and estar depends on whether the predicate is nominal, adjectival, or locative).

Spanish–Hebrew children’s narratives 

These interrelated factors may result in a burden for children when having to make the correct choice in discourse. In a study on the developmental mastery of the ser/estar contrast, Sera (1992) found that children sometimes tended to overuse the forms of estar, particularly in contexts where the syntactic cues were not salient enough to make the right distinction.12 In Hebrew, as noted, there were very few errors overall. One was the use of the deﬁnite marker ha ‘the’ in complex NPs with the accusative marker et as in pitom rau et štey cfardeim meohavim for pitom rau et štey ha-cfardeim meohavim ‘suddenly (they) saw acc-two frogs in love’ or ani ekax et ha-cfardea exad for ani ekax et ha-cfardea ha-exad ‘I will take acc the frog (the one)’. Contrary to what we expected, these errors were found in the text of a 7-yearold girl. This is evidently due to the diﬃcult conditions governing use of the deﬁnite marker in complex NPs, known to be a very late acquisition in child Hebrew (Zur 1983). We also found other type of errors in the Hebrew texts which could not be classiﬁed as any of the categories described above. These errors resembled violations of normative or prescriptive usage known in the literature as ‘‘standard deviations’’ (Berman 1987; Ravid 1995). They include omission of syntactically required resumptive pronouns in complex relative clauses, e.g., ve higiu le makom še aleyhem lehagia ‘and (they) got to (a) place that were supposed to-arrive’ for makom še aleyhem lehagia elav ‘place that were supposed to-arrive at-it’;13 lack of agreement between subject and verb in VS word order as in ve haya šam cfarde’im ‘and (there) was there frogs’ for normative ve hayu šam cfarde’im ‘and (there) were there frogs’; and morphophonological errors of vowel alternation in opaque context verbs, e.g., hepil for normative hipil ‘dropped’. These same type of ‘‘native-like’’ errors were documented by Berman (1985) for preschoolers and by Ravid (1995) for schoolchildren and adolescent monolingual native speakers of Hebrew. Besides the errors detailed in this section, we found that subjects had diﬃculty in the retrieval of highly speciﬁc lexical terms in Spanish. Generally, they opted to cope with this diﬃculty by adopting one of the following strategies: (a) selecting a word or phrase referring to some of the semantic attributes of the inaccessible word, e.g., la casa de la rana ‘the house of the frog’ for frasco ‘jar’ or la casa de las abejas ‘the house of the bees’ for colmena ‘beehive’; (b) replacing the word with another word which retains some of its semantic features, e.g., mosquitos ‘mosquitoes’ for abejas ‘bees’ or animalito ‘little animal’ for ardilla ‘gopher’; (c) replacing the word by a very general term, e.g., el perro con el lugar de la rana ‘the dog with the place of the frog’,

 Judy Kupersmitt and Ruth A. Berman

where the general term lugar ‘place’ is used instead of a word like frasco ‘jar’; (d) asking the investigator to supply them with the word; (e) avoiding the word, e.g., y el perro buscó. . y el perro también ‘and the dog looked. . and the dog also’. Lexical substitution as detailed in (a) and (b) is also found in texts of monolingual Spanish-speakers, and so seems clearly developmental in character. On the other hand, it is also reported for bilingual adult frogstories (Olshtain and Barzilay 1991). In sum, results of the error analysis show three diﬀerent types of errors which seem to be aﬀected by diﬀerent variables such as age of the subjects, typology of the languages in contact, structural complexity of grammatical paradigms in those languages, and directionality of inﬂuence from one language on the other.

Conclusions and discussion This study provides further evidence for the proposal that the development of narrative structure is largely language-independent.14 It clearly demonstrates that command of global-level narrative action structure develops as a function of age rather than according to language in the Spanish and Hebrew texts alike. Overall narrative construction appears to be aﬀected by level of overall development and maturation, less so by level of language mastery. This suggests that the ability to produce a well-constructed narrative text relates to a more general domain of cognitive structures and processes, unrelated to a speciﬁc language. In terms of linguistic expression, our ﬁndings do not conﬁrm the prediction that diﬀerences in structural command and overall use of linguistic forms in the two languages are proportional to the degree of similarity between the particular linguistic categories they encode in each language, with tense/aspect being furthest apart, null subjects quite similar, and relative clauses almost identical. Rather the degree of correspondence between how particular linguistic categories are used in a narrative context is determined by the accessibility of such linguistic forms to the speaker. Thus, the system of T/A in Spanish is not easily accessible to bilingual subjects who are speakers of Hebrew, a language in which temporal qualities of events are not categorized by a rich system of grammaticized aspect. As a result, the Spanish narratives of these bilingual children are characterized by a rather ﬂat temporal texture, lacking in rich and varied contrasts between the temporal contours of the

Spanish–Hebrew children’s narratives 

events in the story. This was clearly shown in the children’s deployment of tense/aspect distinctions, which showed almost no diﬀerences between the two languages, with the single exception of relatively widespread use of Imperfective in the Spanish texts.15 Two psycholinguistically motivated explanations could account for this ﬁnding. First, Spanish imperfectives obey the criterion of formal simplicity, since they consist of a quite transparent combination of stem plus suﬃx, whereas the progressive and perfect forms require a structurally complex and morphonologically opaque combination of auxiliary plus participle (a type of construction which, moreover, occurs in only one very restricted context, the habitual past, in Hebrew). A second factor is that the clear binary contrast between perfective/imperfective might facilitate acquisition: initially, at the ﬁrst phase in development of T/A forms, children will use the imperfective for predicates which are inherently durative, like statives, and subsequently, at a later phase in development, they will be able to transfer this semantic distinction to other types of predicates to denote the contrast between telicity and atelicity, e.g., in dormía (Imperfective) ‘was asleep, was sleeping’ versus durmió (Perfective) ‘slept’. Interestingly, the subjects in this study seem to get stuck at the ﬁrst phase, where they use the imperfective mainly with stative verbs. These ﬁndings suggest that further research is needed on the bilingual use of tense/aspect systems, to yield indepth analysis of the expression of temporal relations in extended discourse such as narrative texts. Such ﬁndings might have important implications for our understanding of the interrelation between language and cognition in the encoding of temporal distinctions. The use of subject ellipsis was as expected. The results show that this device was accessible to narrators in both languages, and that it was exploited for the narrative functions of discourse connectivity and topic maintenance respectively in accordance with the typology of the target language, Spanish or Hebrew. Thus, the Spanish texts have a larger proportion of clauses with null subjects, and these appear across longer stretches in the text, while Hebrew narrations generally use null subjects more locally, in coreferential coordinated and subordinated clauses. Relative clauses were used relatively more in Spanish than in Hebrew. This result may have been content-related, since the most favored function of relative clause constructions was a presentative one, and the story characters were largely introduced in the ﬁrst, Spanish-language, part of the story. This ﬁnding may, however, reveal a genuine diﬀerence between the narrative rhetoric in the two languages which for some reason was obscured in comparing relative clause

 Judy Kupersmitt and Ruth A. Berman

distribution in the monolingual Spanish and Hebrew frog story texts (Dasinger and Toupin 1994). Beyond this diﬀerence in amount of use, relative clauses served similar discursive functions, as was expected, in both languages. The error analysis performed in this study suggests that grammatical errors in the ‘‘weaker’’ language of bilingual subjects should be analyzed from a psycholinguistic perspective, on the one hand, and from a typological perspective, on the other. This kind of analysis, in a more extended study, could shed light on the characteristics of two interlinked processes: the process of language loss and the process of language acquisition in a bilingual situation. Analysis of errors in core grammatical morpho-syntactic usage (Berman 1999) revealed that across the sample, subjects produced more errors in Spanish than in Hebrew. The categorization of errors according to three diﬀerent types — fossilization, transfer, and developmental errors — seems indicative of quite general trends in the development of linguistic devices in bilinguals. These may take one of the following forms: (a) the development of grammar is identical in monolinguals and bilinguals; (b) the development of grammar in bilinguals is like that of monolinguals, but it gets ‘‘stuck’’ at a certain point in development; and/or (c) the construction of grammar in bilinguals and monolinguals follow diﬀerent developmental paths (Kupersmitt, in progress). The question of what determines which pattern will predominate under what circumstance is of considerable interest, but lies well beyond the scope of the present chapter. In order to compare these ﬁndings with previous studies in the general domain of bilingualism and in bilingual narratives speciﬁcally, it must be borne in mind that the notion of bilingualism has many diﬀerent interpretations. A bilingual population can in principle be deﬁned, and often is deﬁned in practice, by a variety of criteria, including: amount of use of L1 and L2, time of acquisition of both L1 and L2, whether L1 or L2 is a minority or a majority language in a certain society (see, further, Romaine 1995). This classiﬁcation is critical when analysing the results of a study such as this one, since a variety of other background variables, including age and overall exposure, also have an eﬀect on the process. For example, Berman’s (1999) comparison of the English (L1) and Hebrew (L2) narratives of North-American immigrants to Israel and of the English (L2) and Hebrew (L1) narratives of Israelis who had spent time in the United States revealed clear diﬀerences in their use of not only of morpho-syntax and the lexicon, but also of rhetorical devices and register in L1 compared with L2, relative to the nature of the bilingual population in question. The Hebrew L2 speakers showed excellent command of Hebrew syntax and lexicon, but their use of Hebrew was often

Spanish–Hebrew children’s narratives

communicatively inappropriate in the juvenile type of oral narrative task which they were asked to perform. The English L2 speakers, in contrast, revealed a rather diﬀerent linguistic proﬁle, with numerous errors at the more basic level of morpho-syntax and lexical usage. In terms of methodology, the elicitation procedure which was adopted, of having children tell one half of the picturebook-based story in Spanish, and the other half in Hebrew, proved eﬀective in some ways. It solved the problem of rehearsal, which is inherent in studies which have subjects tell the same story twice, in a diﬀerent language each time. And it revealed, for example, that most children were well able to cope with the demands of having to switch the language in midstream. On the other hand, the ﬁndings are biased by the diﬀerent content of the two parts of the story: the ﬁrst half includes the background setting or exposition and introduction of the characters, mention of the plot-initiating event and motivation, while the second half continues with a further set of adventures and ends with a resolution, hence evoking more evaluative elements. One solution would have been to counterbalance the elicitation procedures, such that one half of the population tells the ﬁrst part of the story in Spanish and then continues in Hebrew, with the other half — matched for such variables age and school-level — starting in Hebrew and continuing in Spanish.16 The small number of children involved in this study, 2–3 at each agegroup, did not allow for well-motivated quantitative analyses. In order to reveal clear developmental trends, a larger, more balanced group of subjects is required for each age-group and for diﬀerent levels of schooling (preschool, gradeschool, junior high school) and these should be compared with adults (on condition that a comparable population is available in the country in which the research is conducted). The present analysis thus needs to be considered as a preliminary pilot study, or along the lines of indepth casestudy reports, rather than as a representative, cross-sectional piece of research. As such, it is suggestive of how carefully detailed, qualitative analyses of speciﬁc linguistic categories can and should be related to particular narrative functions and compared across languages as well as across age-groups. A larger, more carefully controlled population is required to yield more signiﬁcant quantitative results. Nonetheless, smaller populations may be legitimate, and possibly even better suited, for indepth qualitative investigation relating linguistic forms to narrative functions. Further extension of the present study, which is presently under way by the ﬁrst author, is detailed comparison of the same categories of form/function



 Judy Kupersmitt and Ruth A. Berman

relations between a bilingual population and monolingual speakers of the ‘‘weaker’’ language (in this case, Israeli children who are also Spanish-speaking). What already seems clear from the present study is that the children’s Hebrew-language narratives correspond in all respects to those of their monolingual counterparts. Finally, results of such an analysis as we have attempted should be related to other factors, particularly to such independent variables as literacy development and type and amount of exposure to each of the two languages. Such a requirement is particularly necessary within the conceptual framework motivating the present study: the view that proﬁciency goes beyond the knowledge of the morpho-syntactic structural properties of a given language. It is the ability to use the forms of the language(s) one knows in performing a communicative task such as telling a story that determines one’s proﬁciency in a given target language.

Notes . This chapter is a revised and extended version of a paper presented by the ﬁrst author at the International Symposium on Bilingualism, Newcastle Upon Tyne, April 1997. The authors are indebted to Dan I. Slobin for his careful reading and insights on an earlier draft and to the editors of this volume for their very helpful comments. Some of the examples are taken from data gathered by Aura Bocaz in Santiago, Chile, and Buenos Aires, Argentina. We are grateful to her for sharing het texts with us. Responsibility for inadequacies rests with the authors. . See the large cross-sectional studies by Rabinowitch (1985), where ‘‘home’’ language is English and ‘‘host’’ language is Hebrew and by Polinsky (1997), where ‘‘home’’ language is Russian and ‘‘host’’ language is English. . In fact, Hebrew verbs are rich in bound morphology, being inﬂected for person, as well as for number and gender. These categories of agreement (subject-verb and noun-adjective) are disregarded here. . Exceptions were the Past Perfect, a later acquisition, and the Past Perfective Progressive, which is rare even in adults (Sebastián and Slobin 1994). The Present Perfect is not considered here, since it has largely disappeared from Latin American Spanish usage, the input dialect of the mainly Argentinian-origin children in our sample (Slobin 1997). . As noted by Silva-Corvalán (1983), a narration can be carried out in the past or in the present tense, or in an alternation of the two tenses. All three possibilities are available in this picture-story type of narration. . The ﬁrst tense-switch here, from past to present (se fue ‘left’ to sabe ‘know’) can probably be accounted for by the so-called ‘‘defective tense’’ hypothesis, which claims that

Spanish–Hebrew children’s narratives

initially children identify certain verbs with a ﬁxed value for semantic aspect (Aktionsarten). True, the ‘‘aspect before tense’’ analysis is associated in the literature with younger children, at an earlier stage of acquisition. But this ﬁnding for children as late as 4 years old may be due to the exigencies of extended, monologic online text production compared with the more highly scaﬀolded conversational interaction. Thus, Spanish-speaking children typically identify a stative, partially modal verb like saber with durative-type present-tense marking (compare with Michal, aged 5;11 who does use sabía ‘knew-ifv’). This corresponds to what has been found for Hebrew-speaking children’s tendency to use the verb rotse ‘want’ purely in present tense and never in past until as late as age 4 or 5 years, whereas the opposite is true for the telic verb nafal ‘fall’, which they invariably use in past and not in present tense for a long period of time. This type of locally-motivated tense-shifting is thus typical of younger monolingual children as well. . This example was taken from the monolingual data gathered in Santiago, Chile and Buenos Aires, Argentina, by Aura Bocaz. The letters as indicate Argentinian Spanish. . Sebastián and Slobin 1994, p. 29. . This example is taken from the corpus gathered by Aura Bocaz. . The use of the reﬂexive particle se in this particular example would be acceptable in Peninsular Spanish, but not in Latin American Spanish, which is the variety under consideration here. . Regularization errors in the acquisition of Spanish and other Romance languages are reported by Clark (1985). . Sera (1992) reports on four studies on the acquisition of ser and estar. Overall, results indicated that children manage to make the right choices between ser and estar early on in development, except with locations, where they overuse the forms of estar and they treat location of events as location of objects. Moreover, she concluded that children seem to rely more on syntactic than on semantic cues in this respect. . These constructions also occur in adult relative-clause formation in everyday spoken Hebrew, and are identical to some of the relative clauses produced in monolingual Hebrew frogstory texts, including from children aged 9 to 10 years (Berman and Neeman 1994). . This is not to say that narrative abilities are also culture-independent (see, for example, Invernezzi and Abouzeid [1995], Minami and McCabe [1995]). But the children in this study share very much the same socio-cultural and educational background in both the languages which they know. . Along with the use of Perfective forms, parallel to the Past Tense in Hebrew. . This was not feasible in the present study because of its small population.

References Aarssen, J. (1996). Relating events in two languages: Acquisition of cohesive devices by Turkish–Dutch school-age bilingual children. Tilburg: Tilburg University Press.



 Judy Kupersmitt and Ruth A. Berman

Akinçi, M. A. (1999). Développement des compétences narratives des enfants bilingues turc-français en France, âgés de 5 à 10 ans. Thèse de doctorat nouveau régime en sciences du language. Université Lumière, Lyon 2. Alarcos Llorach, E. (1995). Gramática de la Lengua Española. Madrid: Esposa Calpe. Armon-Lotem, S. (1996). The Minimalist Child: Parameters and Functional Heads in the Acquisition of Hebrew. Tel Aviv University doctoral dissertation. Bamberg, M. (1987). The acquisition of narratives: Learning to use language. Berlin: Mouton. Bamberg, M. (1994). Development of linguistic forms: German. In: Berman and Slobin (eds.), 189–238. Bamberg, M. and V. Marchman (1990). What holds a narrative together? The linguistic encoding of episode boundaries. Papers in Pragmatics, 4, 58–121. Berman, R. A. (1984a). From nonanalysis to productivity: interim schemata in child language. Working Paper # 31. Tel Aviv University Unit for Human Development and Education. Berman, R. A. (1985). Interference in FLT. In: T. Husen and T. N. Hustlethwaite (eds.), International Encyclopaedia of Education, Oxford: Pergamon Press, 1943–5 Berman, R. A. (1985). The acquisition of Hebrew. In: D. I. Slobin (ed.), The crosslinguistic study of language acquisition. Vol. 1: The data. Hillsdale: Lawrence Erlbaum, 255–371. Berman, R. A. (1988). The ability to relate events in narrative. Discourse Processes, 11, 469–97. Berman, R. A. (1990). Acquiring an (S)VO language: Subjectless sentences in children’s Hebrew. Linguistics, 28, 1135–66. Berman, R. A. (1992). Child language and language change. [in Hebrew]. In: U. Ornan, R. Ben-Shachar, and G. Toury (eds.), Hebrew as a Living Language. Haifa: Haifa University Press, 9–22. Berman, R. A. (1995). Narrative competence and Storytelling performance. Journal of Narrative and Life History, 5, 285–313. Berman, R. A. (1997). Preschool knowledge of language. In: C. Pontecorvo (ed.), Writing development: An interdisciplinary view. Amsterdam: John Benjamins, 61–76. Berman, R. A. (1997). Developing form/function relations in narrative texts. Lenguas Modernas, 24, 45–60. Berman, R. A. (1998). Bilingual proﬁciency/proﬁcient bilingualism: Insights from narrative texts. In: G. Extra and L. Verhoeven (eds.), Bilingualism and Migration. Studies on Language Acquisition. Berlin: Mouton de Gruyter Berman, R. A. (1999). Bilingual proﬁciency/proﬁcient bilingualism in Hebrew-English narrative texts. In: G. Extra and L.Verhoeven (eds.), Language Change in Migration Contexts. Berlin: Mouton de Gruyter, 187–218. Berman, R. A. and E. Dromi (1984). On marking time without aspect in child language. Papers and Reports on Child Language Development, 23, 21–32. 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 (1994). Development of linguistic forms: English. In: Berman and Slobin (eds.), 127–88. Berman, R. A. and D. I. Slobin (eds.) (1994). Relating events in narrative: A cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum.

Spanish–Hebrew children’s narratives

Bocaz, A. (1989). Estudio evolutivo de la marcación aspectual de los sucesos en la producción del discurso narrativo. Actas del III Congreso Internacional sobre el Español de América. Salamanca, Spain: Gráﬁcas Varona, 1311–20. Bocaz, A. (1991). Marcación de la organización secuencial de los sucesos en la producción de discurso narrativo. In Actas del VIII Seminario Nacional de Investigación y Enseñanza de la Lingüística. Santiago, Chile: Universidad de Santiago de Chile, 249–55. Bocaz, A. (1992). Procesos inferenciales abuductivos en la interpretación de escenas narrativas complejas. Lenguas Modernas, 19, 99–106. Borer, H. (1984). Parametric syntax: Case studies in Semitic and Romance languages. Dordrecht: Foris. Bos, P. (1997). Development of bilingualism: A study of Moroccan children in the Netherlands. Tilburg: Tilburg University Press. Bowerman, M. (1985). What shapes children’s grammar? In: D. I. Slobin (ed.), The crosslinguistic study of language acquisition: Vol. 2. The data. Hillsdale: Lawrence Erlbaum, 1257–320. Clark, E. V. (1985). The acquisition of Romance, with special reference to French. In: D. I. Slobin (ed.), The crosslinguistic study of language acquisition. Vol. 1: The data. Hillsdale: Lawrence Erlbaum, 687–782. Dart, S. N. (1992). Narrative style in the two languages of a bilingual child. Journal of Child Language, 19, 367–87. Dasinger, L and C. Toupin (1994). The development of relative clause functions in narrative. In: Berman and Slobin (eds.), 457–514. Duran, R. P. (1985). Discourse skills of bilingual children: precursors of literacy. In: A. Aguirre and J. A. Fishman (eds.), International Journal of the Sociology of Language: Lamguage in the Chicano speech community, 53. New York: Mouton, 99–114. Fleischman, S. (1990). Tense and narrativity. London.: Routledge. Giora, R. and Y. Shen (1994). Degrees of narrativity and strategies of semantic reduction. Poetics, 22, 447–58. Herman, J. (1996). ‘‘Grenouille, where are you?’’ Crosslinguistic transfer in bilingual kindergartners learning to read. Harvard Graduate School of Education, doctoral dissertation. Hopper, P. J. (1979). Aspect and foregrounding in discourse. In: T. Givón.(ed.) Syntax and semantics, Vol. 12: Discourse and syntax. New York: Academic Press. Hyams, N. (1989). The null subject parameter in language acquisition. In: O. Jaeggli and K. J. Saﬁr (eds.), The null subject parameter. Dordrecht: Kluwer. Invernezzi, M. A. and M. P. Abouzeid (1995). One story map does not ﬁt all: A crosscultural analysis of children’s written story retellings. Journal of Narrative and Life History, 5, 1–20. Jisa, H. and M. A. Akinci (1996). Inﬂuences of L1 Turkish on L2 French. Paper presented at VIIth International Congress for the Study of Child Language, Istanbul. Kail, M. and M. Hickmann (1991). French children’s ability to introduce referents as a function of mutual knowledge. First Language, 12, 73–94. Karmiloﬀ-Smith, A. (1979). A functional approach to child language: A study of determiners and reference. Cambridge: Cambridge University Press. Katzenberger, I. (1994). Cognitive, linguistic, and developmental factors in picture-series narration. Tel Aviv University unpublished doctoral dissertation [in Hebrew]



 Judy Kupersmitt and Ruth A. Berman

Kern, S. (1997). Comment les enfants jonglent avec les contreaintes communicationnenelles, discursives et linquistiques dans la production d’une narration. Doctoral dissertation, Université Lumière, Lyon 2. Kupersmitt, J. (1999). Development of form/function relations in the narratives of Spanish– Hebrew bilinguals: What Stays and what Goes?, M.A Thesis, Tel Aviv University. López Ornat, S. (1992). Sobre la gramaticalización. Prototipos para la adquisición de la concordancia verbo-sujeto: datos de la lengua española en niños de 1,6 a 3,6. Cognitiva, 4, 1, 49–74. López Ornat, S. (1994). La adquisición de la lengua española. Madrid: Siglo Veintiuno. Marchman, V. (1989). Episodic structure and the linguistic encoding of events in narrative. Doctoral dissertation. University of California, Berkeley. Mayer, M. (1969). Frog, Where are you? New York: Dial Press. Minami, M. and A. McCabe (1995). Rice balls and bear hunts: Japanese and North American family narrative patterns. Journal of Narrative and Life History, 22, 423–46. Olshtain, E. and M. Barzilay (1991). Lexical retrieval diﬃculties in adult language attrition. In: H. W. Seliger and R. M. Vago (eds.), First language attrition. Cambridge: Cambridge University Press, 139–50. Pearson, B. Z. and V. M. Umbel (1996). The relationship between narrative and linguistic skills in English and Spanish among Hispanic schoolchildren in Miami. Paper presented at VIIth International Congress for the Study of Child Language, Istanbul. Pfaﬀ, C. (1979). Constraints on language mixing: intrasentential code-switching and borrowing in Spanish/English. Language, 55, 291–318. Polinsky, M. (1997). American Russian: Language loss meets language acquisition. In: W. Browne (ed.), Formal approaches to Slavic linguistics. Ithaca: Cornell University Press. Rabinowitch, S. (1985). Hebrew proﬁciency of English-Hebrew bilingual preschoolers compared with their monolingual Hebrew-speaking peers. Tel Aviv University Ph.D. dissertation Ravid, D. (1988). Transient and fossilized phenomena in inﬂectional morphology: varieties of spoken Hebrew. Doctoral dissertation, Tel Aviv University. Ravid, D. D. (1995). Language Change in Child and Adult Hebrew: A Psycholinguistic Perspective. New York, Oxford: Oxford University Press. Reinhart, T. (1984). Principles of gestalt perception in the temporal organization of narrative texts. Linguistics, 22, 779–809. Reinhart, T. (1988). Development of temporality in children’s narratives. Unpublished doctoral dissertation, University of California, Berkeley. Romaine, S. (1995) Bilingualism (2nd edition). Oxford: Blackwell. Sebastián, E. and D. I. Slobin (1994). Development of linguistic forms: Spanish. In: Berman and Slobin (eds.), 239–84. Selinker, L. (1974). Interlanguage. In: J. Richards (ed.), Error Analysis. London: Longman. Sera, M. D. (1992). To Be or to Be: Use and Acquisition of the Spanish Copulas. Journal of Memory and Language, 31, 408–27. Shatz, M. (1985) A song without music and other stories: How cognitive process constraints inﬂuence children’s oral and written narratives. In: D. Schiﬀrin (ed.), Meaning, Form and Use in Context: Linguistic Applications. Gurt, 35.

Spanish–Hebrew children’s narratives 

Shen, Y. (1988). Schema theory and the processing of narrative texts. Journal of Pragmatics, 12, 415–67. Silva-Corvalán, C. (1983). Tense and aspect in oral Spanish narrative: Context and meaning. Language, 59, 760–80. Slobin, D. I (1990) The development from child speaker to native speaker. In: J. W. Stigler, G. Herdt, and R. A. Shwereder (eds.), Cultural psychology: Essays on comparative human development. Cambridge: Cambridge University Press, 235–56. Slobin, D. I (1995). Special issues in the acquisition of Spanish: Contributions to theory. Keynote address presented to Primer Encuentro Internacional sobre Adquicisión de las Lenguas del Estado, Universidad de Santiago de Compostela, Spain. Slobin, D. I (1996). Two Ways to Travel: Verbs of Motion in English and Spanish. In: M. Shibatani and S. A. Thompson (eds.), Grammatical constructions: Their form and meaning. Oxford: Clarendon Press, 195–219. Slobin, D. I (1997). Talking perfectly: Discourse origins of the present perfect. In: W. Pagliuca and G. Davis (eds.), Perspectives on grammaticalization. Amsterdam: John Benjamins. 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) Weinreich, U. (1974). Languages in Contact. The Hague: Mouton. Weissenborn, J. (1992). Null subjects in early grammars: Implications for parameter setting. In: J. Weissenborn, H. Goodluck, and T. Roeper (eds.), Theoretical issues in language acquisition: continuity and change in development. Hillsdale, NJ: Lawrence Erlbaum, 269–99. Weist, R., H. Wysocka, and P. Lyytinen (1991). A crosslinguistic perspective on the development of temporal systems. Journal of Child Language, 18, 67–92. Wigglesworth, G. (1992). Investigating cognitive and linguistic development through narrative. Doctoral dissertation, La Trobe University. Zur, B. (1983). On the acquisition of deﬁniteness by Hebrew-speaking children. Tel-Aviv University master’s thesis [In Hebrew].

Chapter 12

Narrative development in Hebrew and English Dorit Kaufman

Children’s narratives are shaped by the social, cultural, and linguistic milieu within which they are situated. As children mature, their narratives reﬂect their cognitive and linguistic development and their unique sociolinguistic and cultural context (Applebee 1978; Berman 1995, 1996; Berman and Slobin 1994; Bamberg 1987, 1997; Champion, Seymour and Camarata 1995; McCabe and Peterson 1991; Wigglesworth 1997). Narratives recounted in the native language (L1) by children whose families have relocated to another country where a second language (L2) is dominant and where exposure to L1 has been dramatically reduced open a window to the integration of the two sociolinguistic and cultural contexts within which the children are immersed and concurrently to the conﬂict for dominance between them. Socialization into the sociocultural and linguistic milieu of the host environment and increasing L2 dominance co-occur with a transition away from the socio-cultural and linguistic domains of the native language and decreasing L1 proﬁciency. This study explores linguistic features of Hebrew narratives recounted by Israeli children, native speakers of Hebrew, whose families relocated to the United States temporarily or permanently. Narrative development and ﬁrst language attrition are explored among these children whose L2 acquisition cooccurs with attrition in L1 production. The children’s L1 oral narration is compared with narrative data from monolingual speakers of Hebrew and English (Berman and Slobin 1994). The monolingual data serve as a baseline for the investigation of ﬁrst language attrition among the bilingual children across age groups and across proﬁciency levels. Detailed analyses of the narratives provide insight to the ways in which the children’s narratives are constrained by diminishing productive knowledge of L1, the compensatory strategies that the narrators employ to counteract dwindling linguistic and communicative L1 resources, and the autonomous and integrative processes

 Dorit Kaufman

that contemporaneously impact on their production. These issues are addressed in the context of the narrators’ references to the protagonists and inanimate objects, their description of action, L1 compounding trends, and selection of temporal anchoring. An overview of relevant sociolinguistic features that characterize the Israeli community in the United States provides the sociolinguistic backdrop.

First language attrition among Hebrew speakers in the United States First language attrition is the progressive decline of language proﬁciency once exposure to the native language has ceased or has been dramatically reduced. It occurs in young children (Burling 1978; Kaufman and Aronoﬀ 1989, 1991), in the elderly who have had little or no exposure to the language since childhood (Levine 1997), and in whole communities, as in instances of language death (Dorian 1981; Schmidt 1985). First language attrition among children in immigrant communities is triggered by reduced exposure to the ﬁrst language and contemporaneous acquisition of the dominant language of the host environment. Children’s integrative orientation and assimilatory socialization patterns are strong predictors of fast acquisition of L2 and consequent attrition of L1 and, in fact, the predominance of L1 attrition among pre-puberty children in immigrant communities is widely documented (Extra and Verhoeven 1993a, 1993b; Ferguson and Heath 1981; McKay and Wong 1988; Romaine 1991; Silva-Corvalan 1994). L1 attrition among children at this age is attributed to sociocultural, linguistic, and aﬀective variables that collectively contribute to the children’s waning motivation to speak their native language. At pre-puberty, children’s integrative orientation, socialization, and communication patterns sharply contrast with those of their parents and older siblings whose ties to the homeland and to the native language remain strong. Children who were born in the L2 environment or have entered it at a young age lack their parents’ and siblings’ aﬀective aﬃnity to the native language and to their historical, cultural, and ethnic roots and consequently, they are quick to assimilate and to acquire the language and culture of their peer group. Israelis residing in the United States have a high level of L2 proﬁciency across age groups and a high level of education (Bureau of the Census 1990). Both of these greatly enhance employment opportunities, expedite assimilation into L2 settings, and contribute to Israelis’ successful integration in the United States (Kaufman 2000). Voluntary immigration and a secular way

Narrative development in Hebrew and English 

of life further reduce dependence on local ethnic community organizations and facilitate integration into L2 sociocultural and linguistic contexts. In contrast to immigrant groups whose religious rituals and dietary restrictions encourage social isolation and L1 maintenance (Huﬃnes 1989, 1991; Sridhar 1997), Israeli immigrants in the United States have not formed their own ethnic organizations or community centers and have not socially isolated themselves (Shokeid 1989). Furthermore, the sociocultural similarity between Israeli and American way of life has required little readjustment to the host environment and has stimulated fast integration into L2 settings. The high prestige of L2 among Israelis has likewise motivated its increased use in multiple domains. At the same time, acquisition and maintenance of the native language has been very important for Israelis residing in the United States (Kaufman 2000). Israeli children’s early linguistic socialization, including those born in the United States is highly L1-centered. Parents provide a literacy-rich L1 environment, communicate with their young children in Hebrew, and expect siblings to do the same. The children’s fast acquisition of L2 and sociocultural integration into the host environment and peer group begins upon starting school, where their linguistic and sociocultural locus is increasingly L2-dominant. Subsequently, their exposure to L1 and opportunities for its productive use are further reduced and are almost exclusively conﬁned to the home. In an attempt to cultivate and maintain their children’s native language skills, parents adhere to L1 use when engaging in discourse with them. The children who initially continue to speak the native language, over time increasingly shift to L2 in response to their parents’ L1 discourse (Kaufman 1991, in press). Such unreciprocal communication (Gal 1979; McConvell 1991) is pervasive among Israeli and other immigrant groups (Haugen 1969; Silva-Corvalan 1994) and mirrors parents’ and children’s contrasting patterns of socialization. Although unreciprocal communication endorses the children’s choice of L2 for communication in the home environment and contributes to L1 attrition, it also ensures continued exposure to L1 and development of the children’s L1 receptive skills. Such exposure however fails to inhibit fast L1 attrition which has been documented to occur within a single generation, while the shift in language use among the community as a group takes several generations (Extra and Verhoeven 1993a; Fase, Koen and Kroon 1992; Ferguson and Brice Heath 1981; Fishman 1989; Kaufman 1995; Kaufman and Aronoﬀ 1989, 1991; McKay and Wong 1988; Romaine 1991; Van Els 1986; Waas 1996).

 Dorit Kaufman

Renewed immersion in L1 settings cultivates and revitalizes native language use, reverses the attrition process, and interrupts the attrition continuum. The discontinuity in L1 attrition is inherent to the Israeli experience in the United States and to the demographic composition of this immigrant community that consists of two groups that intermingle socially and form a single linguistic community. The ﬁrst group includes representatives of Israeli organizations whose residence in the United States is temporary and the second includes individuals and families whose residence in the United States is prolonged or becomes permanent. Both groups are comparable in their socioeconomic status, cultural, and sociolinguistic orientation. The composition of the former group is constantly in ﬂux since families returning to Israel are replaced by new arrivals who form social ties with the more permanent residents of the local Israeli community, revitalize their linguistic and cultural links with the homeland, and provide opportunities for rejuvenation and updating of the native language among children and adults. Additional ways that have helped to reverse the progression of L1 attrition have included the establishment of Israeli supplementary schools to increase the children’s exposure to L1 in academic, social, and cultural contexts, prolonged visits to Israel during the summer months, increased contact with newly arrived immigrants, and visits of Israeli relatives and friends (Kaufman 2000).

The present study Subjects The data for this study are drawn from a larger corpus elicited to investigate attrition processes among Israeli immigrants in the United States. The focus of this paper is on narrative data that were collected from 30 children, 15 boys and 15 girls native speakers of Hebrew ranging in age from 6;2–13;11 whose parents are all ﬁrst generation immigrants, native speakers of Hebrew (see Table 1). The children have resided in the United States for varying periods of time ranging from birth to a minimum of two years. In all the homes, maintenance of the Hebrew language is highly valued and promoted by the parents who all communicate with their children in the native language. L1 literacy practices are pervasive and children’s books, Israeli newspapers, and magazines are widely available. In this context, all the children, even the ones who were born in the United States have mastered L1 proﬁciency, prior to their entering

Narrative development in Hebrew and English 

L2-dominant schools, at least at the level of proﬁciency of three year old monolingual native speakers who by this age have mastered the nominal and verbal paradigms and the required morphological and inﬂectional operations (Berman 1985, 1994; Berman and Dromi 1984). Table 1. Distribution of the narrators Grade

Age

Gender

Years in the U.S.

Preferred Language

1 1 1 2 2 3 3 3 3 3 3 4 4 5 5 5 5 5 6 6 6 6 6 6 7 8 8 8 8 9

6,2 6;9 6;9 7;0 7;11 8;1 8;3 8;3 8;8 8;10 8;11 9;1 9;3 10;5 10;8 10;8 10;9 11;1 10;10 11;0 11;3 11;2 11;5 12;0 12;7 13;0 13;2 13;3 13;4 13;11

F M F F M F M F F M F F M F M F M M M M M M F M F F F M M F

2 3 Since birth 5 Since birth Since birth Since birth 5 Since birth 3 6 8 8 7 7 4 8 3 2 4 10 6 9 Since birth 6 Since birth 2 7 8 Since birth

Hebrew Hebrew + Englisha Hebrew English Hebrew Hebrew Hebrew English Hebrew + English English English English English English English English + Hebrew English English English Hebrew English English English English + Hebrew English English Hebrew English Hebrew + English English

a

Hebrew+English or English+Hebrew represent the narrator’s responses

Data collection The narrative production data were elicited with the wordless picture book Frog, Where are you? (‘Mayer 1969’) which describes the adventures of a boy

 Dorit Kaufman

and his dog as they search for their frog that had escaped (Bamberg 1987; Berman and Slobin 1994). The use of this instrument has kept the content constant throughout thereby allowing comparison among children across age groups and proﬁciency levels. This picture-book instrument was selected for elicitation of oral and written narrative discourse for a variety of reasons. First, picture-based narrative data provide particularly rich context for the study of ﬁrst language attrition among bilingual children. In contrast to conversationbased data that are collaboratively shaped by interlocutors’ reciprocal input that facilitates informants’ lexical accessibility and masks their restrictive competence, picture-based narrative data are exclusively the children’s own monologic rendition of the story depicted in the pictures. These provide robust evidence for the compensatory strategies that the children employ to oﬀset their dwindling linguistic knowledge and communicative skills in the native language. Second, narrative conventions and structure are familiar to the children from L1 storytelling and other literacy-related activities that are widespread in Israeli homes (Kaufman 2000) and from L2 literacy activities at school. Third, this instrument has been used successfully in language acquisition studies among monolingual speakers of both Hebrew and English and it has been found to be linguistically and culturally appropriate for all ages (Berman and Slobin 1994). Fourth, the data available from monolingual speakers of L1 and L2 provide a baseline for the investigation of attrition in bilingual children and allow comparison across age groups and across proﬁciency levels. Finally, this picture-book has also been used successfully in attrition and other crosslinguistic studies and relevant data are therefore available (Cohen 1989; Olshtain and Barzilay 1991; Yoshitomi 1999; Yukawa 1997). The children’s L1 skills were evaluated through individual interviews that ranged in duration from 35–45 minutes each. These were conducted in Hebrew by a native speaker and were recorded and transcribed. The children were initially asked to look through the wordless picture book to become acquainted with the story and then to tell the story in Hebrew to the interviewer as they turn the pages. The storytelling activity served as a warm-up and provided a measure of the children’s L1 oral proﬁciency. All the children were able to perform this oral L1 narration task. This was followed by additional tasks that included comprehension and production of innovative compounds and pragmatic appropriateness of speech act production. For the ﬁnal task the children were asked to produce a written version of the wordless picture book. Only 67 per cent produced the written version in L1. The other 33 per cent were unable to write in Hebrew and requested to use English for

Narrative development in Hebrew and English 

their written narratives. The discussion in this paper addresses ﬁndings from the oral narratives. The children’s successful production of the oral narration task is attributed in part to prolonged summer visits to Israel and to continued exposure to L1 at home and among peers in the academic setting of the supplementary schools. Over half the children had spent an extended two month-summer vacation in Israel a mere three months prior to the data collection. The other summer visits took place between 15 months to 52 months prior to the study (see Table 2). Table 2. Percentages of visits to Israel (3–8 weeks duration) prior to data collection 3 months

15 months

27 months

39 months

52 months

54

23

13

3

7

Language cultivation endeavors by the parents and the community scaﬀold native language use and enhance receptive and, to some extent, production skills. Despite these eﬀorts, however, two thirds of the children declare English as their preferred language of communication (see Table 3). Only two thirds of the children communicate with their parents in Hebrew, 20 per cent do so with siblings, and a mere 3 per cent speak Hebrew with peers. The children’s choice of L2 as their preferred language of communication represents a shift in their sociolinguistic loyalty away from their native language and towards their incipient L2-centered sociocultural and sociolinguistic identity. Table 3. Percentages of language selected by children for communication Language preferred Hebrew English Hebrew + English

23 60 17

Language actually spoken With parents With siblings

With friends

60 33 7

3 94 3

20 43 37

Data analysis Stories are important agents of socialization (Applebee 1978) and children are able to produce narrative rendition of real or ﬁctional events as early as two years of age (Hicks 1991). With age, children develop multiple ways of representing their knowledge through language and their narratives reﬂect the way in which they structure events both cognitively and linguistically (Bruner

 Dorit Kaufman

1986). L1 literacy activities are introduced very early in Israeli homes and form an integral part in the children’s early socialization to their culture. Through storytelling and poetry, children learn to recognize sociocultural roles and relationships, develop knowledge of narrative schema, and become aware of linguistic structures that pertain to narratives. With the onset of attrition, narrators are challenged to frame their stories in a genre-appropriate manner as well as to reformulate the narrative within the boundaries of their dwindling L1 proﬁciency. As attrition progresses, dwindling knowledge of L1 constrains the narrator’s linguistic structuring of events and results in narratives that lack linguistic sophistication and inaccurately reﬂect cognitive immaturity. The narratives are increasingly characterized by lexical substitutions to compensate for loss of lexical speciﬁcity, repetition of lexical items and phrases, simpliﬁcation, and frequency of pauses, hesitation, false starts, and repairs. Narrators’ linguistic constraints trigger autonomous and integrative processes which contemporaneously impact on their L1 storytelling. Autonomous processes generate linguistic structures that are ascribed to intralanguage interplay among linguistic elements within the L1, while integrative processes involve crosslinguistic interplay among L1 and L2 grammatical elements. In both types of interactions, robust and unmarked linguistic features cause the restructuring of more marked and permeable features and result in innovative L1 forms that attest to gaps in the L1 system and provide important insights into the narrators’ reformulation of L1 paradigms (Kaufman 1995). The narrative data provide robust evidence for ongoing autonomous and integrative processes as narrators draw upon their dwindling L1 resources to compensate for lexical and structural gaps.

Results Reference to entities The narratives of monolingual L1 and L2 speakers demonstrate that they have easily recognized, retrieved, and named target lexical referents in their description of animate protagonists, such as cfardea (‘frog’), dvorim (‘bees’), yanshuf (‘owl’), or cvi (‘deer’), inanimate objects, such as cincenet (‘jar’), and setting, for example, yaar (‘forest’) that are depicted in the pictures. In contrast, the less proﬁcient bilingual narrators whose command of the L1 lexicon has

Narrative development in Hebrew and English 

dwindled were unable to do so and they used, rather creatively, circumlocutory strategies and introduced diverse lexical alternatives that include nouns, demonstratives, or descriptive phrases to compensate for their lexical gaps when the target L1 words were not accessible (see Table 4). Drawing upon autonomous strategies, these narrators described the deer as: xaya (‘animal’), ha-xaya ha-zot (‘this animal’), xaya gdola (‘large animal’), mashu (‘something’), and by the innovative circumlocutory phrase: ma she-hakelev haya kore sus im karnaim (‘what the dog would call a horse with horns’), an idiosyncratic attempt that draws upon resources available in L1 while assigning responsibility for this descriptive reference to the dog. Other innova-

Table 4. Narrators’ description of protagonists Frog cfardea *cardea *cfardak hadavar haze mashu frog

(correct form) (phonetic approximation) (phonetic approximation) (this thing) (something) (L2 code mixed)

Owl yanshuf cipor gdola cipor gdola gdola cipor meod yafa shaul

(correct form) (big bird) (big big bird) (very beautiful bird) (metathesized shual, ‘fox’) or (idiosyncratic phonetic blending shual ‘fox’ + owl) (L2 code mixed)

owl Deer cvi ayil xaya ha-xaya ha-zot eize xaya xaya gdola ma she-ha-kelev haya kore xaya im karnaim mashu ze ani lo yodaat eix korim la-ze moose deer reindeer

(correct form) (correct form) (animal) (this animal (some animal) (large animal) (what the dog would call an animal with horns) (something) (this) (I don’t know how you call this) (L2 code mixed) (L2 code mixed) (L2 code mixed)

 Dorit Kaufman

tive descriptions include references to the owl (correct form yanshuf), for example, cipor gdola (‘big bird’), cipor meod yafa (‘very beautiful bird’), repetition for emphasis cipor gdola gdola (‘big big bird’), and a particularly innovative shaul which could be attributed to integrative phonetic codeblending of L1 shual (‘fox’) and L2 ‘owl’, or to autonomously metathesized form of L1 shual (‘fox’). Code mixing of L2 words is rare even in the narratives of the least proﬁcient narrators. It is only occasionally used as a strategy and is generally limited to nouns and to animate beings. Using autonomous processes, narrators generally draw upon their L1 resources, most creatively at times, rather than integrate code-mixed L2 words. When target L1 words are not available, the narrators select from a variety of available L1 alternatives that are semantically related. When naming speciﬁc animate beings such as the frog, deer, and owl, the linguistic choices are lexically too speciﬁc and alternatives are limited, so narrators may resort to code-mixed L2 words, demonstratives, or circumlocutory means of describing the animate beings. On the other hand, the narrators’ description of inanimate objects, such as, cincenet, (the jar that at the outset contains the frog), present a rich array of L1 lexical alternatives to Table 5. Narrators’ description of objects and setting Jar

cincenet bakbuk kufsa kluv kumkum kaze kad kli zxuxit keara kerax mashu hamakom shehayeled sam oto

Forest yaar xhursa makom sheyesh kama ecim leyad ecim leyad hagan haxuca badeshe baecim

(correct form) (bottle) (box) (cage) (water pot) (some sort of jug) (container) (glass) (bowl) (ice, phonetically similar to keara [bowl]) (something) (the place that the boy placed it) forest (correct form) wood (correct form) (a place that has some trees) (next to trees) (next to the garden) (outside) (on the grass) (in the trees)

Narrative development in Hebrew and English 

replace the L1 target word. None of the bilingual narrators including the least proﬁcient ones use L2 code-mixed words to describe the jar (see Table 5). The words selected by the bilingual narrators as substitutes are all commonly used in the home domain and have ranged from words for containers, such as bakbuk (‘bottle’), kufsa (‘box’), kumkum (‘water pot’), kli (‘container’), kad (‘jug’), keara (‘bowl’), to enclosures as in kluv (‘cage’) which was used by two of the narrators. This variety of L1 lexical alternatives selected to describe inanimate objects contrasts with the more lexically-speciﬁc choices evident in the monolingual data. Additional variations that are evident in the bilingual data and that diverge from the monolingual data are examples of existing or non-existing words that phonetically approximate the target L1 words or other words. This phenomenon is prevalent in attrition (Kaufman 1991, 1995). In one of the narratives, for instance, the word kerax (‘ice’) which is phonetically similar to keara (‘bowl’) is initially used to describe the jar. In subsequent references to the jar, the narrator uses the word keara (‘bowl’).

Reference to actions Monolingual speakers of Hebrew master the process of verb formation and the required morphological and inﬂectional operations by age three (Berman 1985, 1994). Longitudinal and cross-sectional attrition data from pre-puberty native speakers of Hebrew who are immersed in an English speaking environment have shown restructuring in the nominal and verbal systems of the native language as proﬁciency in L2 increases (Kaufman 1991; Kaufman and Aronoﬀ 1989, 1991). The verbal data in particular are intriguing in that the new forms that emerge do not resemble developmental forms that appear in monolingual acquisition, but represent the operation of autonomous and integrative processes, a synthesis of the two languages in the narrators’ environment, and the children’s reformulation of the L1 verbal paradigm. A detailed investigation of the narrative data and the narrators’ description of action provide evidence for restructuring, reformulation, and synthesis processes that conﬁrm previous ﬁndings and augment current knowledge. An analysis of the narrators’ verb formation across the narratives reveal a systematic transition away from lexically-speciﬁed verbs that are most prevalent in L1 towards using verb+particle constructions that are widespread among English speakers. This transition is attributed in part to the typological diﬀerences between L1 and L2. Hebrew has been described as a verb-framed language

 Dorit Kaufman

where lexical information is encoded in the verb (Berman and Slobin 1994). Lexically-speciﬁed monolexemic verbs such as tipes (‘went up’ or ‘climbed’) xipes (‘looked for’ or ‘searched’) represent informal and colloquial speech in modern Hebrew and are widely used by monolingual Hebrew speakers from a young age. In contrast, English encodes movement, direction, and manner by

Table 6. Shift from lexically-speciﬁed verbs to verb + particle combinations Narrators’ Normative L1 lexically- verb + particle speciﬁed verb constructions

Examples from the bilingual narratives

barax ‘escaped’

yaca me‘went out of ’

Ha-cfardea bore..ha-cfardea yoce me-ha-kufsa ‘the-frog esc.. the-frog goes-out of-the-box’

rac me‘ran from’

va-ha-kelev rac me-hadvorim ki hem holxim axarav ‘and — the-dog runs from-the-bees because they go.pl after.him

halax me‘went from’

holex me-ha-cincenet ‘goes from — the-jar’

halax axrey ‘went after’

veha-owl holex axar axrey ha-yeled ‘and-the-owl go after (repair) after the boy’

halax la ‘went to’

ha-zvuvim holxim la-kelev ‘the ﬂies go to the dog’

rac axrey ‘ran after’

ha-xaya rac axrey ha-kelev ‘the animal runs after the dog’

raxav axrey ‘rode after’

va-ha-bees roxvim axrey ha-kelev ‘and-the-bees ride after the-dog’ (phonetically similar but inappropriate word)

ala al ‘went.up on’

la’alot al ha-ec ‘to.go.up on the-tree’

halax al ‘went on’

az hu holex ha-yeled holex al even gadol ‘then he goes the-boy goes on stone big’

histakel ba‘looked in’

ha-yeled ve-ha-kelev shelo mistaklim ba-xeder shelo lo lo roim oto ‘the-boy and-the-dog his look in-the-room his no no see him’

radaf ‘chased’

tipes ‘climbed’

xipes ‘searched’

histakel *bishvil ha-yeled ada’in mistakel bishvil ha-cfardea ‘looked for’ ‘the boy still look for the-frog’ xipes *bishvil ‘searched for’

hayeled mexapes ada’in bishvil ha-cfardea ‘the boy search for still for the-frog’

Narrative development in Hebrew and English

the pervasive use of the verb+particle construction and monolingual English speaking school-aged children frequently use such verb+particle combinations throughout their narratives (Berman and Slobin 1994). The data show that declining proﬁciency leads to a dramatic shift from use of lexically-speciﬁed verbs to use of verb+particle combinations. The narratives of the least proﬁcient bilingual children are characterized by almost exclusive use of verb + particle structures. Furthermore, even in instances where a lexically-speciﬁed L1 verb is used, it is recast in a verb + particle mold and a redundant particle is added resulting in a non-native form, such as, mexapes+bishvil (‘search + for’) instead of the correct form mexapes et ha-yeled (‘looks-for OBJ the-boy’) as illustrated in Table 6. Another linguistic feature that systematically occurs across the narrative data of the less proﬁcient bilingual narrators is the use of the unmarked general-purpose verb halax ‘go’ with a particle (see Table 7). The proliﬁc use of this construction in the narratives is analogous to the developmentallydetermined widespread use of this form among the youngest monolingual Table 7. Loss of speciﬁcity and use of the generic verb holex (‘go’) + particle L1 Verb

Examples form the children’s narratives

yaca ‘exited’

ha-yeled ve-hakelev holxim baxuc ‘the-boy and the-dog go-3mpl outside’

nixnas ‘entered’

hu holex betox makom she-yesh kama ecim ‘he goes inside a place that-there (are) some trees’

barax ‘escaped’

hacfardea holex me-ha-cincenet ’the frog ’goes from-the-jar’

radaf ‘chased’

ha-zvuvim hem holxim la-kelev ve-holxim lisho ..linshox oto ‘the-ﬂies they go to the-dog and go to bi…to bite him’ ha-bees holex axrey ha-kelev ‘the bees go.3s after the-dog’

tipes ‘climbed’

ha-yeled holex al ec ‘the-boy goes on (a) tree’ ve-hayeled holex al deer ‘and-the-boy goes on deer’ hu halax al ha-hill ‘he went on the-hill’

raxav ‘rode’

holex al xaya ‘goes on animal’



 Dorit Kaufman

Hebrew-speaking narrators. However, unlike the bilingual narrators, monolingual speakers use these structures in conjunction with many other verbs that are lexically-speciﬁc, for example, ala (‘ascend, went up’), tipes (‘climb’) yaca (‘exit, went out’). The Germanic bilexemic construction of the verb ‘go’ plus particle, as in ‘go up’, ‘go in’, ‘go out’ is ubiquitous is English discourse and it often replaces the Latinate monolexemic construction of verbs like ‘climb’, ‘enter’, ‘exit’ that represent more formal usage in English. Monolingual English speakers of all ages pervasively use this bilexemic construction in their narratives. The verb holex (‘go’)+particle is used by the bilingual narrators as a generic verb to substitute for a host of lexically-speciﬁed verbs (see Table 7). Both the verb+particle and the general-purpose go+particle constructions replace common L1 lexically-speciﬁed verbs that are used pervasively in the narratives of the youngest preschoolers monolingual native speakers of Hebrew. Use of this construction in the bilingual children’s L1 narratives provide evidence for paucity in expressive predicates, loss of speciﬁcity in L1 verb selection, and operation of autonomous and integration processes as L1 lexical items are superimposed on L2 structures generating forms that range from developmentally juvenile to non-native.

Compounds and related genitive constructions The two types of compounds that are most commonly used in Hebrew for expressing relations in colloquial speech and in written language are the bound genitive (smixut or construct-state) (N+N) and its alternative analytic, or periphrastic form with the free genitive particle shel (N shel N). These two compounding types are most accessible to children native speakers of Hebrew, and are most pervasive in their speech. Bound-genitive compounding in Hebrew is initially acquired by monolingual Hebrew speakers between the ages of 1;6–2;0 as rote-learned structurally-unanalyzed compounds of familiar words in the children’s environment (Berman 1987; Ravid and Shlezinger 1995). Examples include: aruxat-boker, (breakfast), beged-yam (‘bathing suit’), bet-sefer (‘school’). The free genitive (N shel N) phrase along with other nounpreposition-noun phrases are acquired around the age of 2;4 to 2;6. Speakers can choose to use the bound genitive construction bakbuk maim (‘water bottle’) or its periphrastic alternative bakbuk shel maim (‘bottle of water’). Productive use of bound genitives that require morphological and phonological changes in the head noun is acquired as late as 4–7 years old (Berman

Narrative development in Hebrew and English 

1985; Clark and Berman 1987). This late acquisition sharply contrasts with the early acquisition of English compounds which are mastered by children as young as two years old (Clark, Gelman and Lane 1985). The Hebrew monolingual narrative data (Berman 1987; Berman and Slobin 1994) show that preschoolers’ use of bound genitive compounding is restricted to highly lexicalized items that include: gan xayot (‘animal garden, zoo’), kaveret-dvorim (‘bee-hive, geza-ec (tree-trunk, log’). The school-aged monolingual narrative data attest that compounding has been acquired as a productive device. Highly specialized non-lexicalized compounds such as bulec, (log), mexilat axbar (‘mouse-burrow’), anfei ha-ec, (tree branches) nexildvorim (‘a swarm of bees’) are pervasive in the data and average 1–2 bound genitive compounds per story (Berman 1987). In contrast, the bilingual children’s narrative data show little evidence for productive use of boundgenitive compounding. The few examples in the bilingual data are restricted to the juvenile periphrastic or analytic compounding using the free genitive ‘shel’ (N shel N). References to the beehive, for example (see Table 8) include gender errors, loss of speciﬁcity, diﬃculty in lexical accessibility and codemixed L2 ‘beehive’. Table 8. Examples of analytic compounding in the bilingual data Beehive

a

kaveret-dvorima ha-bayit shel ha-dvorim ha-bayit shel ha-dvorot ha-bait shel ha-zvuvim ha-. . . ken shel ha-dvorim ha-davar shel ha-dvorim

(the-house of the-bees-masc.) (the-house of the-bees.fem) (the-house of the-ﬂies.masc) (the-nest of the-bees) (the-thing of the bees)

No evidence of the correct lexicalized bound-genitive compound is found in the data

Lexical gaps and lexical retrieval Examples of lexical substitutions when target L1 words are unavailable are pervasive throughout the bilingual data and are attributed to several possibilities. First, the target L1 word has never been encountered, second, a once familiar L1 word has been lost, and third, temporary diﬃculty in lexical retrieval. The ﬁrst possibility seems unlikely in this case, since preschool monolingual speakers of English and Hebrew showed no diﬃculty in naming the animate protagonists, inanimate objects, and setting. The bilingual narrators had all reached at least preschoolers’ native language proﬁciency and

 Dorit Kaufman

their exposure to L1 and its production have not ceased since then. Determining whether a lexical gap is due to loss or temporary inaccessibility is more problematic. Attrition studies have traditionally treated non-use of a lexical item as evidence for loss. For example, if several versions of a story narrated several months or years apart were compared and target L1 words were absent in subsequent retelling, it was determined that a word had been lost. The bilingual data show that such conclusions may be tentative and premature. Words that are seemingly inaccessible at the beginning of the story surface in descriptions of later scenes in the narrative. While narrators’ initial attempts to retrieve words may result in false starts, repairs, and circumlocution, they show little hesitation in producing the supposedly ‘lost’ words in later scenes. This has also been exempliﬁed with description of action. The verb tipes (climb) was temporarily inaccessible for one narrator (8;1, female, born in the United States) in the owl scene (picture 11), but was readily available in the log scene (picture 21) at the end of the story. The following sequence shows the narrator’s attempts to describe climbing scenes in the story. She uses verb+particle constructions ala-al (‘go up on’), holex al (‘walks on’) and ﬁnally the lexically speciﬁed, metapes (‘climbs’). The verb tipes is widespread in children’s stories, poems, and songs and every Hebrew speaking child is highly likely to be exposed to it. The example illustrated above underscores the need for further research to explore why semantic associations that seem to trigger production of speciﬁc lexical items at some scenes fail to trigger these words at other scenes in the story. Another related question that begs further investigation is how seemingly ‘lost’ words resurface with or without an apparent stimulus. Diﬃculty in lexical accessibility is compounded when narrators describe scenes that are more complex and require greater cognitive processing. This is the case with the middle scenes of the story that are depicted in pictures 8–17 (Wigglesworth 1997) and describe the adventures of the boy and the dog from the time they go into the forest in search of the frog, their action-ﬁlled encounters with the gopher, the bees, the owl, and the deer, and their fall into the water. The data show that even some of the more proﬁcient narrators have opted to bypass or to summarize these scenes by describing the boy as having encountered harbe xayot (‘many animals’), baalei xaim (‘living things’), as having gone to mekomot axerim (‘other places’), kol miney mekomot (‘all sorts of places’), and as having seen od xayot (‘more animals’). Post-task discussion has shown that when narrators ﬁnd it linguistically diﬃcult to provide precise portrayal of protagonists and events, they avoid the detailed account and

Narrative development in Hebrew and English 

summarize several scenes by using phrases such as ‘the boy encountered many animals’. Table 9. Lexical accessibility — from verb + particle to a lexically-speciﬁed verb Picture Narrator’s description 9–10 11 13–14 20–21

ve-ha-kelev menase la’alot al ha’ec the dog tries to go up on the tree hakelev od menase la’alot al ha’ec ve-ha-yeled yoshev al ec the dog still tries to go up on the tree and the boy sits on tree ha-yeled um holex al even ad halemala, hu ala lemala, hu lemala the boy um goes on (a) stone up to the top, he went to the top, he (is) at the top ha-yeled omed al yad ec shavur, ha-yeled metapes al ec shavur the boy stands next to a broken tree, the boy climbs on a broken tree

Tense marking The past tense is the unmarked tense for temporal anchoring of narratives. The temporally removed past tense is the narrative convention in children’s books and its use distinguishes storytelling from picture description that is characterized by the deictically anchored present tense. Selection of the past tense for storytelling is developmentally determined. Monolingual Israeli school-aged children overwhelmingly select the past tense for their narratives (94 per cent). While younger preschool children fail to maintain a consistent storytelling mode and tend to shift between the present and past tenses. Preschool data show that 50 per cent selected the past tense as their anchoring tense. The English monolingual data are less deﬁnitive. While the past tense is still the preferred tense for the English speaking monolingual school-aged children, only two thirds of the narratives are temporally anchored in the past tense (see Table 10). Table 10. Tense anchoring percentages among monolingual and bilingual childrena Monolingual children Hebrew Preschool School aged Past 50% Present 15% Mixed 35% a

Israeli immigrant English

94% 0% 6%

children

Preschool School aged

Grade 3 Grade 6

33.3% 33.3% 33.3%

0% 83% 17%

67% 33% 0%

Native speakers’ data is based on Berman and Slobin (1994) Preschool: 3–5 year olds. School-aged: 9 year olds

67% 33% 0

 Dorit Kaufman

Analysis of the bilingual data show that tense anchoring is developmentally determined not only by age but also by level of L1 proﬁciency. Older and more proﬁcient narrators show a preference for the past tense for their temporal anchoring and exhibit fewer instances of tense shifting. Table 10 illustrates a comparison between the bilingual third graders (8;1–8;11; N=6) and sixth graders (10;10–12;00; N=6) and the monolingual English and Hebrew preschool and school-aged data. Although there is a developmental shift from use of the present tense to the past tense across narrators, none of the bilingual third grade, eight-year-old group use the past tense for their temporal anchoring. By age eight, monolingual English and Hebrew speakers are developmentally cognizant of tense anchoring norms for narratives, yet the bilingual 8-year-old group overwhelmingly select the present tense. Although conclusions can only be tentatively discussed until more data are available, it is plausible that the bilingual children’s present-tense temporal anchoring is due to their daily linguistic experience with L1 discourse that occurs primarily in oral and conversational domains that are deictically and temporally anchored in the present. The bilingual sixth graders exhibit mixed temporal anchoring patterns that are strikingly similar to those of the monolingual English speakers, where two thirds anchor their narratives in the past and one third in the present. The number of children is small in this case and further research is required to determine whether the bilingual children’s non-deﬁnitive preference for past tense temporal anchoring is due to autonomous processes that result in narratives that follow more juvenile developmental patterns typical of younger monolingual speakers of Hebrew, or whether the bilingual narrators’ tense anchoring preference is attributed to the same processes that determine the monolingual English school-aged narrators’ patterns of mixed temporal anchoring.

Conclusions and discussion Bilingual children’s L1 narratives are rooted in their sociolinguistic experiences as they move between the two cultures in their environment and simultaneously engage in the discourse of both. The bilingual narrative data mirror the narrators’ integrative orientation and assimilatory socialization patterns in L2 domains and conﬁrm the progressive decline in their native language skills. The narratives are sociolinguistic texts that reﬂect the children’s shifting

Narrative development in Hebrew and English 

identities, their transition away from their ethnic roots, and their progression towards the L2-centered realm of their peer group. The linguistic features evident in the L1 narratives discussed in this study represent the bilingual narrators’ own formulation of L1 word formation rules and their synthesis of the two languages in their environment. Languages frame narratives in uniquely language-speciﬁc ways and oﬀer narrators formal linguistic devices and expressive options that provide linguistic and rhetorical scaﬀolding. Bilingual narrators are cognizant of formal devices available in each of their languages and in their narratives appropriately draw upon linguistic and expressive resources in their respective languages. With the onset of attrition, progressive decline in the native language and fragmentation in its linguistic structures and word formation paradigms constrain the narrators’ lexical and structural choices. Autonomous and integrative processes contemporaneously impact on narrative structure, complexity, and rhetorical style. The resulting narratives appear rhetorically and linguistically immature and are incompatible with the narrators’ cognitive ability and age. These ﬁndings warrant further investigation to augment emerging understanding of narrative development in the context of L1 attrition. Further analysis is needed to probe the growing disparity between narrators’ cognitive ability, their knowledge of narrative discourse conventions, and their L1 production, to investigate the linguistic features that characterize narrators’ individualized recounting of events, and to examine the systematic trends that appear across age groups and proﬁciency levels.

References Applebee, A. N. (1978). The child’s concept of a story. Chicago: University of Chicago Press. Bamberg, M. (1987). The acquisition of narratives: Learning to use language. Berlin: Mouton de Gruyter. Bamberg, M. (ed.). (1997). Journal of Narrative and Life History. Special issue: Oral versions of personal experience: Three decades of narrative analysis, 7. Mahwah, NJ: Lawrence Erlbaum, 1–4 Berman, R. (1985). The acquisition of Hebrew. In: D. I. Slobin (ed.), The crosslinguistic study of language acquisition: Vol 1. The data. Hillsdale, New Jersey: Lawrence Erlbaum, 255–371. Berman, R. (1987). A developmental route: learning about the form and use of complex nominals in Hebrew. Linguistics, 25. Amsterdam: Mouton de Gruyter, 1057–1085. Berman, R. (1994). Word formation as evidence. Paper presented at the 19th annual Boston University Conference on Child Language Development, Boston.

 Dorit Kaufman

Berman, R. (1995). Narrative competence and storytelling performance: How children tell stories in diﬀerent contexts. Journal of Narrative and Life History, 5, 4. Hillsdale, NJ: Lawrence Erlbaum, 285–313. Berman, R. (1996). Form and function in developing narrative abilities. In: D. Slobin, J. Gerhardt, A. Kyratzis and J. Guo (eds.), Social interaction, social context, and language. Mahwah, NJ: Lawrence Erlbaum , 343–367. Berman, R. and E. Dromi (1984). On marking time without aspect in child language. Papers and Reports on Child Language Development, 23, 23–32. Berman, R. and D. Slobin (1994). Relating events in narrative: A cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum Associates. Bruner, J. (1986). Actual minds, possible worlds. Cambridge, MA: Harvard University Press. Bureau of the Census. U. S. Department of Commerce. 1990. 1990 Census of population: Social and economic characteristics of selected language groups for U.S. and States. vol. CP-2–1-B and CPH-L 159. Also Ancestry of the population in the U.S. vol. CP-3–2. Burling, R. (1978). Language development of a Garo and English speaking child. In: E. Hatch (ed.), Second language acquisition. Rowley, MA: Newbury House, 54–75. Champion, T., H. Seymour and S. Camarata (1995). Narrative discourse of African American children. Journal of Narrative and Life History, 5, 333–352. Clark, E. V. and R. A. Berman (1987). Types of linguistic knowledge: interpreting and producing compound nouns. Journal of Child Language, 14, 3, 547–568. Clark, E. V., S. A. Gelman and N. M. Lane (1985). Compound nouns and category structure in young children. Child Development, 56, 84–94. Cohen, A. (1989). Attrition in the productive lexicon of two Portuguese third language speakers. In: Studies in Second Language Acquisition, 11, 2. Cambridge: Cambridge University Press, 135–150. Dorian, N. C. (1981). Language death: The life cycle of a Scottish Gaelic dialect, Philadelphia: University of Pennsylvania. Extra, G. and L. Verhoeven (eds.) (1993a). Immigrant languages in Europe. Clevedon: Multilingual Matters, 67–100. Extra, G. and L. Verhoeven (1993b). A bilingual perspective on Turkish and Moroccan children and adults. In: G. Extra and L. Verhoeven (eds.), Immigrant languages in Europe. Clevedon: Multilingual Matters. Fase, W., J. K. Koen and S. Kroon (eds.) (1992). Maintenance and loss of minority languages. Philadelphia: John Benjamins. Ferguson, C. A. and S. B. Heath (eds.) (1981). Language in the USA. Cambridge: Cambridge University Press. Fishman, J. (1989). Language and ethnicity in minority sociolinguistic perspective. Clevedon: Multilingual Matters. Gal, S. (1979). Language shift. Social determinants of linguistic change in bilingual Austria. New York: Academic Press. Haugen, E. (1969). The Norwegian language in America. Vol. 1 and 2. Bloomington: Indiana University Press. Hicks, D. (1991). Kinds of narrative: genre skills among ﬁrst graders from two communities. In: A. McCabe and C. Peterson (eds.), Developing narrative structure. Hillsdale, NJ: Lawrence Erlbaum, 55–88.

Narrative development in Hebrew and English 

Huﬃnes, M. L. (1989). Case usage among the Pennsylvania German sectarian and nonsectarian. In: N. C. Dorian (ed.), Investigating obsolescence: studies in language contraction and death. Cambridge: Cambridge University Press, 211–226. Huﬃnes, M. L. (1991). Pennsylvania German: convergence and change as strategies of discourse. In: H. Seliger and R. Vago (eds.), First language attrtion. Cambridge: Cambridge University Press, 125–137. Kaufman, D. (1991). First language attrition as a creative interplay between two languages. Unpublished doctoral dissertation. New York: State University of New York at Stony Brook. Kaufman, D. (1995). Where have all the verbs gone: Autonomy and interaction in attrition. Southwest Journal of Linguistics, Special Issue on Language Loss and Public Policy, I. Vol. 14, 1–2, 43–66. Kaufman, D. (2000). Attrition of Hebrew in the United States: Sociolinguistic Perspectives. In: E. Olshtain (ed.), Immigration, identity and language. Jerusalem: Magnes Publications, Hebrew University, 173–96. Kaufman, D. and M. Aronoﬀ (1989). Morphological interaction between L1 and L2 in language attrition. In: S. Gass, C. Madden, D. Preston and L. Selinker (eds.), Variations in Second Language Acquisition: Psycholinguistic Issues. Avon, England: Multilingual Matters, 202–215. Kaufman, D. and M. Aronoﬀ (1991). Morphological disintegration and reconstruction in ﬁrst language attrition. In: H. W. Seliger and R. M. Vago (eds.), First language attrition. Cambridge: Cambridge University Press, 175–188. Levine, G. (1997). Incomplete L1 acquisition in the immigrant situation: The case of Yiddish in the United States. Unpublished doctoral dissertation. The University of Texas at Austin. McCabe, A. and C. Peterson (eds.). (1991). Developing narrative structure. Hillsdale, NJ: Lawrence Erlbaum. McConvell, P. (1991). Understanding Language shift: a step towards language maintenance. In: S. Romaine (ed.), Language in Australia. Cambridge: Cambridge University Press, 143–156. McKay S. and S. C. Wong (1988). Language Diversity Problem or Resource? A Social and Educational Perspective on Language Minorities in the United States. New York: Newbury House. Mayer, M. (1969). Frog, Where are you? New York: Dial Press. Olshtain, E. and M. Barzilay (1991). Lexical retrieval diﬃculties in adult language attrition. In: H. W. Seliger and R. M. Vago (eds.), First language attrition. Cambridge: Cambridge University Press, 139–150. Ravid, D. and Y. Shlesinger (1995). Factors in the Selection of compound types in spoken and written Hebrew. Language Sciences, 17, 2. Elsevier Sciences, 147–179. Romaine, S. (ed.). (1991). Language in Australia. Cambridge: Cambridge University Press. Schmidt, A. (1985). Young people’s Dyirbal: An example of language death from Australia, Cambridge: Cambridge University Press. Shokeid, M. (1989). Children of Circumstances: Israeli emigrants in New York. Ithaca, NY: Cornell University Press.

 Dorit Kaufman

Silva-Corvalan, C. (1994). Language contact and change: Spanish in Los Angeles. Oxford: Clarendon Press. Sridhar, K. (1997). Aspects of language and culture of Asian Indians in New York. In: O. Garcia and J. Fishman (eds.), The Multilingual Apple: Languages in New York City. Amsterdam: Mouton de Gruyter, 257–280. van Els, T. (1986). An overview of European research on language attrition. In: B. Weltens, K. de Bot and T. van Els (eds.), Language attrition in progress. Dordrecht, Holland: Foris, 3–18. Waas, M.. (1996). Language attrition downunder: German speakers in Australia. Frankfurt: Peter Lang. Wigglesworth, G. (1997). Children’s individual approaches to the organization of narrative. Journal of Child Language, 24, 2, 279–309. Yoshitomi, A. (1998). On the loss of English as a Second Language of Japanese returnee children. In: L. Hansen (ed.), Second language attrition: Perspectives from Japanese contexts. Oxford: Oxford University Press, 80–111. Yukawa, E. (1997). L1 Japanese attrition and regaining: Three case studies of early bilingual children. Stockholm: Center for research on bilingualism, Stockholm University.

Chapter 13

Trilingual children narrating in Hebrew, English and Spanish Anat Stavans

The ability to relate events in a coherent and cohesive manner has been the focus of recent research in child language acquisition. The most recent and among the pioneering cross-linguistic studies to date has been a study conducted by Berman and Slobin (1994). The primary focus of their study was to trace the developmental pattern of form-function relations in the language used for relating events and situations. This sort of mapping sheds light not only on the linguistic knowledge as it develops but also on the formation of conceptual knowledge. This dynamic interaction of language and cognition has shown similarities among diﬀerent age monolinguals across diﬀerent languages. However, little is known about the representation and processing of narratives in the bilingual person (Verhoeven 1993; Lanza 1994; Minami and McCabe 1995; Stavans 1997; and more recent reports in this volume). The bilingual faces a reconciliation task between the linguistic worlds available to him/her. Such reconciliation entails ‘juggling’ not only conceptual information but also linguistic and cultural information as well. The manipulation of a multi-communication system implies constant and very controlled monitoring of all the cognitive, linguistic and cultural channels. Attending to all these channels of information, in turn, implies a certain degree of negotiation where trades and concessions are made. The mature and ‘balanced’ bilingual attempts to make the least number of trades and concessions when dealing with such multi-communication systems. Hence, the bilingual must conceptualize, encode and decode information in accordance with the formal parts of language (syntax, morphology and phonology) and the functional roles these forms take in the language use (meaning, discourse, intentions, interpretations, interactions, etc.). This task is far from being trivial and impermeable to mistakes as has been noted in Berman (1999).

 Anat Stavans

To date, bilingual research has focused on comparisons of each of the bilingual’s languages to a monolingual cohort. In this way, the bilingual’s ‘languages’ are viewed as two separate and almost independent linguistic and cognitive entities that have been put together as in a simple arithmetical procedure. Yet, the bilingual’s processing of the two languages is far more complex. Bilinguals live, by deﬁnition, in two linguistic worlds. Some researchers claim that the bilinguals do not behave like two monolingual speakers/hearers contained in a single brain (Grosjean 1992; Paradis 1987) and that the language processing is a result of the interaction between these linguistic worlds (Hernanadez, Bates and Avila 1994; Selinker 1972; Bates and Mac Whinney 1982, 1987, 1989). Yet there are other researchers who have claimed the opposite (Volterra and Taechner 1978). Recent studies supporting the former view on bilingual processing have shown that bilinguals, especially simultaneous bilingual children, develop two separate linguistic systems. These systems are described in these studies as being in some ways similar and in other ways diﬀerent from the linguistic system of the monolinguals cohorts (Genessee 1989; Meisel 1989; De Houwer 1990, 1994; Lanza 1992; Stavans 1990). Both these perspectives are concerned with bilingual processing. A point of interest would be to see to what extent these views apply to trilingual processing. Studying the narrative production of bilinguals provides an opportunity to evaluate these two perspectives. For instance, it is important to see not only how a 9 years old handles the narratives in Hebrew, Spanish or English, but also how a trilingual 9 years old ‘juggles’ the languages in similar or diﬀerent ways when asked to tell the same story in each language. A comparison of a multilingual and a monolingual production of a narrative tells us to what extent these two narratives are similar. But it is the contrast, which brings out the diﬀerences that allow for a study of the interaction between the languages. Such contrast is highlighted in the bilingual’s ‘binary’ linguistic options as claimed by Lanza (in this volume) in focusing our attention not only on the how languages are used in the bilingual narrative but also on the what-why relations in those narratives. To that end, Lanza poses questions such as: What are the diﬀerent forms used to render two accounts of the same story? What are the similarities and diﬀerences in the functions these forms have? When and under what circumstances does the form-function mapping of one language interfere with that of the other? These questions will serve as basis for analysis in these two trilingual subjects as a way of establishing similarities or diﬀerences between mono/bi/trilingual processing.

Trilingual children narrating in Hebrew, English and Spanish 

The present study analyzes the narratives produced by two trilingual school-aged children. The children were raised trilingually from birth in Hebrew, English and Spanish. The children were M (female), aged 7;10, and E (male), aged 9;8 at the time of data collection. E was born in Israel and arrived in the United States when he was three weeks old, and M was born in the United States. Three years prior to data collection the children moved to Israel. The children were both developing normally. E was attending sixth grade during the time of data collection, and M was attending third grade. Both children received input in Hebrew, Spanish, and English from birth. They were raised following the one-person, one-language strategy, according to which they were spoken to exclusively in English by their mother and exclusively in Spanish by their father. The parents spoke to one another in Spanish, using Yiddish in situations when they did not want the children to understand. In situations where other adults were present, the children might hear their parents speaking one of the three languages to others, but the rule with regard to input to the children was carefully adhered to. The children were exposed to English and Hebrew through interaction with adult caregivers, with children of their age in a social setting, and at school. Their Spanish-speaking relatives addressed the children in Spanish, and their Hebrew-speaking relatives addressed them in Hebrew. As noted earlier, the parents consistently spoke to the children in the appropriate language, while the children responded mostly in Hebrew, or if not, in the language used by the parent to the child. The stories were collected and coded following the procedures described by Berman and Slobin (1994). Each child was asked to tell the Frog, Where are you? story by Mercer Meyer, which is a booklet based on a 24- colorless picture sequence. Each of the three narratives, produced in one of the three languages by each child (a total of 6 narratives), was collected in two weeks intervals. Each child was allowed to chose the order in which the languages were used. Both children rendered the three versions starting with the Hebrew rendition, followed by the English version and last was the Spanish narrative. The narratives were recorded and transcribed in broad transcriptions. Coding followed the Berman and Slobin (1994) ‘‘clause’’ unit. The results outline the structural and functional relationships in each of the narratives as produced by these trilingual children. It is not the aim of this chapter to establish comparisons between monolingual and trilingual narratives. However, reference will be made to the monolingual data of Berman and Slobin (1994) as means to establish what is developmentally and linguistically characteristic to these

 Anat Stavans

children and these languages. Although references to monolingual data will be cited as Berman and Slobin (1994), speciﬁc chapters were used for monolingual information. For the monolingual productions in Hebrew all references will be drawn from Berman and Neeman (1994); for the English productions references will be drawn from the Berman and Slobin (1994); and for the Spanish productions references will be made to the Slobin and Sebastian (1994). As for general developmental ﬁnding the references will be made to Berman and Slobin (1994) in general. Moreover, the working assumption is that trilingual processing will not diﬀer in terms of narrative ability or linguistic ability, but rather will diﬀer in that language dominance between three languages calls for a hierarchical order rather than a sequential order of dominance involved in a two-language repertoire. In other words, the working assumption is that while there will be language dominance, it will be relative to one other language at a time rather than one language versus the other two. So we could say that Hebrew is more dominant in relation to English, and English is more dominant in relation to Spanish, and by consequence Hebrew is more dominant in relation to Spanish. Another assumption is that each language may turn to be more dominant in one aspect yet less dominant in another. That may be attributable to language-related issues (such as an extensive temporal marking in the predication of the language) or to language contact issues (such as dominance, transfer and preference).

Narrative structure: A sketch Length The length of the narratives produced by each child in the three languages shows that overall the narratives of these trilingual children are within the ranges of the narratives of the monolingual cohorts as reported in Berman and Slobin (1994). Table 1 illustrates the length of the trilingual narratives. This table illustrates that both trilinguals produce narratives in Hebrew and English within the range of their monolingual cohort. In Spanish however both trilinguals produce narratives slightly shorter than their cohorts. Yet, the narrative of the older trilingual E falls on the shorter end of the range in both productions for Hebrew and English which is characteristic of 9-years-old texts as being: ‘‘highly condensed and tightly organized around the central theme, with quite complex syntax. But [the] stories are dry and businesslike,

Trilingual children narrating in Hebrew, English and Spanish 

Table 1. Total number of clauses in the narratives of monolingual and trilingual school-aged children, by language Monolingual (age 9)*

Trilingual E (9;8)

Trilingual M (7;10)

Hebrew Average # clauses Range # clauses

62 41–98

49

72

English Average # clauses Range # clauses

45 39–51

40

47

Spanish Average # clauses Range # clauses

51 38–61

35

35

a

This data was extracted from Berman and Slobin 1994: 31, Table 3

with no personal ﬂavor.’’ (Berman and Slobin 1994: 74). The younger child, who is less proﬁcient in both English and Spanish makes a rougher attempt at lengthier narratives. Moreover, an ‘order eﬀect’ is suggested by the decrease in story length by both children. On one hand, this order eﬀect may be a result of familiarity with the text. On the other hand, the order eﬀect may suggest the subjects’ choice as it relates to language dominance at the time. Length is not an empirically solid measure for narrative analysis. It is clear that some people can be more concise and precise with their discourse than others. This precision is not necessarily correlated with length but rather with quality. Perhaps a more valuable and empirically sound measure of length would be time of production (i.e. it may well be the case that the narratives produced by multilinguals are as long as those of monolingual cohorts but take twice as much time to produce). For the present study, a more in depth analysis of the narratives is necessary to study not only the length (quantity) but also the ‘text/ure’ of these narratives (quality).

Components The structure of the Frog, Where are you? story is established by the various components of the story. Berman (1988) outlined the events that take place in the story (‘plotline’) and categorized these into major components: setting, initiation, search indoors, search outdoors, adventures, and end. The trilingual narratives were analyzed according to these six categories. Figure 1 illustrates the proportion of each story component in each

 Anat Stavans

E: Hebrew

E: English

E: Spanish

M: Hebrew

M: English

M: Spanish

Sett

Init

S. in

S. out

Adve

End

Figure 1. Proportion of each story component in each rendition by each child

rendition by each child. Both children make speciﬁc mention of each narrative component in each of the three languages. Both E and M give similar emphasis to the stage setting, initiation and beginning of search components. Both children take approximately 6 per cent of the narrative to set the stage in all three languages. The initiation event takes up a larger portion in the Hebrew and the Spanish narrative as compared to the English narrative by both children (such fact has been previously observed by Stavans, forthcoming, in adult Hebrew–English bilingual narratives). The search indoors however is more emphasized in the English narratives than in the Hebrew or Spanish ones by both children. The place where both children’s productions diﬀer is in the emphasis placed on the search outdoors, the adventures and the end components. In the search outdoors, E — the older and more balanced trilingual, shows a heavy emphasis of this component in the Hebrew (dominant) language and a light emphasis on the Spanish. Whereas M, the younger and less balanced trilingual, emphasized the search outdoors component in the Spanish narrative , a language in which M is least dominant. Another diﬀerence between these two trilingual productions comes through in the adventure component. E prefers to give the adventure component a larger ‘slice’ of the Spanish narrative (a language in which he is least dominant) as compared to Hebrew and English where E seems to be in control and able to

Trilingual children narrating in Hebrew, English and Spanish 

encapsulate more of the events. M on the other hand, puts greater emphasis on the adventure component when telling the story in either Hebrew (dominant) or English. Lastly, the end component gets more heavily emphasized in English by E suggesting a certain awkwardness in the ability to come to a closure. M, on the other hand seems to be more rushed into closing the English narrative and devotes a smaller proportion to this component in the narrative. Below are some examples of the expressions used by each child to elaborate on each component in the three productions: Setting The boy and the dog are in the room at night looking at the frog in the jar. Child E Hebrew: mesupar al yeled exad sheyesh lo kelev . . . vecfardea . . . becincenet it is told about one boy that has a dog . . . and a frog . . . in a jar English: once upon a time there was a boy that caught a frog . . . Spanish: una vez habia un niño que tenia un sapo . . . once upon a time there was a boy who had a frog . . . Child M Hebrew: hayeled roe et hacfardea betox habakbuk vegam hakelev shelo . . . the boy sees the frog inside the bottle and also his dog . . . English: the frog is in the glass and the boy and the dog look at the frog . . . Spanish: el niño y el perro Ø-sentados y el sapo Ø en el tubo the boy and the dog 0–sitting and the frog 0 in the tube Initiation The frog escapes while the boy and dog are asleep. The boy and dog wake up to discover the frog has ﬂed. Child E Hebrew: hayeled vehakelev yeshenim ubentaim hacfardea boraxat . . . roim et hacincenet velo roim et hacfardea . . . the boy and the dog sleep and meanwhile the frog escapes . . . we see the jar and we don’t see the frog English: when the boy and the dog were asleep the frog went out . . . they didn’t see him . . . Spanish: cuando el perrito y el niño fueron a dormir, el sapo se salio . . . no sabian donde esta el sapo . . . when the doggie and the boy went to sleep, the frog went out (reﬂexive) . . . [they] did not know where the frog was . . .

 Anat Stavans

Child M Hebrew: hayeled yashen . . . hacfardea boreax mehaxalon . . . kshehem kamim hem mistaklim . . . vehem lo roim et hacfardea . . . the boy sleeps . . . the frog escapes through the window . . . when they wake up they look . . . and they don’t see the frog . . . English then the doggie and the boy look in the glass . . . there is no frog Spanish: el niño y el perro Ø dormidos . . . el niño y el perro no pueden a ver el sapo . . . the boy and the frog Ø sleeping . . . the boy and the dog cannot to-see the frog . . . Search indoors and outdoors The boy and the dog begin looking for the frog in the boots, under the bed, in the jar, out the window, in the forest. Child E Hebrew: hem mexapsim bexol makom baxeder . . . hem mistaklim derex haxalon vekorim lo . . . axarei ze hem yocim mehabait . . . hem holxim laxursha ... they search everywhere in the room . . . they look through the window and call [to] him . . . afterwards they go out of the house . . . they go to the woods . . . English: in the morning they didn’t see him . . . they looked for him . . . they went ﬁnding him in the woods . . . Spanish: y el niño y el perro decidieron encontrarlo . . . vieron por la ventana . . . el perrito se cayo . . . el niño salio a buscarlo . . . and the boy and the dog decided to ﬁnd him (reﬂx.) . . . they looked through the window . . . the doggie fell oﬀ (reﬂx.) they boy went out looking for him . . . Child M Hebrew: hem mexapsim . . . vehayeled mexapes betox haxulca shelo . . . hakelev nixnas labakbuk . . . hakelev nofel mehaxalon . . . hem holxim raxok . . . they look [for] . . . and the boy looks inside his shirt . . . the dog goes into the bottle . . . the dog falls from the window . . . they go far . . . English: they Ø looking for the frog . . . the doggie ran to the garden . . . the boy picked the doggie . . . the boy is shouting to the frog . . .

Trilingual children narrating in Hebrew, English and Spanish 

Spanish: el niño y el perro buscando el sapo en el gardina . . . el niño y el perro andando en el bosque. the boy and the dog looking for the frog in the garden (CS) . . . the boy and the dog walking in the forest. Adventures Boy + owl, Bees chase, Dog running away, Boy + deer. Child E Hebrew: hayeled moce bait shel oger vehaklavlav moce kaveret shel dvorim . . . they boy ﬁnds a gopher’s home and the dog ﬁnds a beehive English: they tried and tried in the trees and all the places and didn’t ﬁnd it . . . Spanish: al niño le dieron un zbeng en su nariz y el perro tuvo problemas con las abejas . . . [to] the boy [they -reﬂx.] gave him a zbeng [Hebrew] on his nose and the dog had problems with the bees. Child M Hebrew: hayeled kore baxor shebaadama . . . vehakelev menase lifgoa babait shel hadvorim . . . the boy calls in the hole in the ground and the dog tries to hit the house of the bees English: the doggie is looking on the bees . . . the boy is looking in a hole on the tree . . . Spanish: el niño buscando en el arbolito . . . el animalito andando en la arena le rasco la nariz al niño . . . the boy searching in the tree . . . the animal (dimin.) walking on the sand scratch (reﬂx.) the nose to the boy End Boys listens to sounds, Boy and Dog follow the sound, Boy and Dog see the frog couple, they see the other frogs, Boy has a frog in his hand, Boy waves goodbye. Child E Hebrew: veaz hem mocim et hacfardea shelo veod axat . . . hacfardeim notnim layeled exad and then they ﬁnd the frog [of his] and another one . . . the frogs give to the boy one English: and then they saw his frog and his wife . . . then the child took one of the frogs . . . one child of the frog that he had.

 Anat Stavans

Spanish: y vieron a los sapos . . . la sapo le dio agarar uno y con permiso and [they] saw the frogs . . . the frog (fem.) gave to him one to hold and with permission Child M Hebrew: veaz hakelev lokeax et hacfardea shehaya lo babait veholex im haklavalav habaita and then the boy takes the frog he had at home and goes home with the dog English: they see the frog and two frogs in the tree . . . and the boy and the dog Ø taking the frog to the house Spanish: y en el arbol dos sapos . . . el perro y el niño agarro el sapo. and in the tree two frogs . . . the dog and the frog took the frog

The degree and skill that each of the trilinguals has in each of the language varies and can be seen in the emphasis they place on the search outdoors, adventures and end component. These components comprise the larger part of the Frog, Where are you? story regardless of whether the teller is bilingual or monolingual. However, these components lend themselves to more variability between diﬀerent degrees of bilingualism. The components of stage setting, initiation and search indoors are more picture bound and canonical in all narratives. These comprise the beginning of discourse usually. They are very similar in all three languages both in content and structure. The means by which the events of a narrative are organized does not yet guarantee a good story. It is important not only to have a thematically coherent chaining of the event but it is also important to see how these events are ‘glued’ together and what sort of ‘glues’ are used to ensure cohesion. Berman and Slobin (1994) studied cohesion by means of connectivity as well as temporality.

Temporal frame Temporality consists of the place each event takes on the timeline and the temporal relations between these events. In this chapter temporality will be analyzed in terms of the verb tense/aspect forms used in each language. Reference will be made to the Berman and Slobin (1994) study of monolingual narratives in order to see which forms are used or avoided by the trilinguals. All three languages use grammatical means to express whether an event took place in the past or the present, or whether one event is related to the other in a particular way. Temporal marking in all three languages, Hebrew,

Trilingual children narrating in Hebrew, English and Spanish

English and Spanish is achieved by means of verb morphology. The grammaticalization of these temporal concepts is obligatory. Tense is the most common temporal expression for all three languages. It formally arranges events in the timeline. Aspect, on the other hand, is more problematic for crosslinguistic analysis. It essentially deals with the perspective the narrator chooses to cast the particular event. It therefore poses more language speciﬁc traits that are not necessarily comparable both formally and functionally. Tense/Aspect forms for Spanish, English and Hebrew have been succinctly described in Berman and Slobin (1994). The following is a summary of the possible forms in the three languages: Table 2. Tense/Aspect forms by language Hebrew

English

Spanish

Tense Aspect

+ −

+ +

+ +

Tense/Aspect Forms

Present

Present Present Progressive

Present Present Progressive

Past

Present Perfect Past Perfect

Present Perfect Past Perfect

Past

Perfective Imperfective

Past Progressive

Perfect Progressive Imperfect progressive

Adapted from Berman and Slobin (1994: 115, Table 2A)

Both English and Spanish oﬀer a wide array of expressions for temporal marking by tense as well as aspectual markings. Hebrew, however, presents the most reduced and limited system for temporal expressions by means of tense marking. The narrator may choose to deploy these forms in diﬀerent ways in order to relate events as present or past; or in order to cast events in time from a particular perspective by means of aspectual marking. The narrator may choose to relate events in one tense (present or past) or to shift from one tense to the other for speciﬁc discursive purposes. Berman and Slobin point out that tense shifting is a common practice, which plays a functional role of foregrounding and backgrounding as well as the introduction of new and old information.



 Anat Stavans

The two trilingual subjects have at their hand the entire set of tense/aspect forms in each language and the possibility to combine the diﬀerent forms. Furthermore, the trilingual may not only exploit each of the formal tense/aspect system in the language but also a combination of these forms to best express each event. The trilingual exploitation of these forms poses interesting questions about which forms dominate the narrative in which language; which forms in a particular language are more vulnerable than others to formal/functional-interference from one of the other systems; and what compensatory means are used by the narrator in the absence of one form in a language; or whether languages which have the same forms would be more prone to exchanges and switches of such forms. The similarities and diﬀerences in the narratives of these two children in each of the languages suggest some answers to these issues. Let us ﬁrst look at the distribution of tense or tense dominance in the trilingual narratives.

Tense preference Table 3 shows the percentage of verb tense used in each language by each child. For purposes of crosslinguistic comparison tenses and aspects have been collapsed into tense marking categories to establish some general comparisons. In the Hebrew version of both E and M, the preferred tense is the present. Unlike the monolingual cohorts (reported by Berman 1988) who prefer the past tense as the matrix tense. The preference of these trilingual children in Hebrew accords with other ﬁndings of Hebrew narratives by both bilinguals and monolinguals in which the preferred tense was the present tense (Stavans, forthcoming). Moreover, the reported preferred tense by monolingual English narrators is the past tense. Looking at these trilinguals’ English version. an interesting point comes up. E, the more balanced trilingual, shows an unbiased preference for the past in his English version. While M, the least balanced of the two, shows a preference for the present tense, opposed to the monolingual Table 3. Dominant verb tense Hebrew Tri-E (9;8) Tri-M (7;10) Mono (9)a a

English

Spanish

Present Past Mix

Present Past

92% 93%

− 92%

2% 6% 4% 3% +

Mix

100% − 6% 2% +

Present Past Mix 3% 75% +

This data has been extracted from the ﬁndings reported in Berman and Slobin (1994) + indicates predominance of above 75%

94% 3% 3% 22%

Trilingual children narrating in Hebrew, English and Spanish 

cohorts, in the English version. This ﬁnding indicates that the more balanced trilingual conforms to the monolingual preferred tense while the less balanced trilingual transfers the preferred tense from the dominant language. The Spanish version produced by E carries a preference for the past tense (as in the English version) while M maintains the tense preference for the present (as in the dominant language, and the E version). Berman and Slobin (1994) report a mixed preference for the Spanish monolinguals between the past and the present. Such mixed preference may be the product of the variety of means with which one may express temporality in Spanish. To sum up, the trilingual preference of a dominant tense is in accordance with the monolingual cohorts for the Hebrew language (the children’s dominant language) and for the more balanced trilingual child in the English language version. The degree of trilingualism allows the more balanced trilingual to break away from the dominant language patterns and create suitable patterns in the other languages; whereas the less balanced trilingual adheres to the dominant language preference and transfers it on to the other versions.

Tense shifting Tense forms were reported as preferred, meaning that these are the most frequently used tenses in the narratives. However, the ﬁndings indicate that none of the narratives are perfectly impermeable to tense shifts. It has been reported by Berman and Slobin (1994) that older children, as well as adults, do not make random tense shifts in their narratives and that these tense shifts are grammatically or thematically constrained to the plotline of the narrative. Moreover, they report that older monolingual Hebrew, English and Spanish speaking children do not shift tenses as much as the younger children. When older children shift tenses it is not always clear why and for what purpose this is done. However, the adult pattern of tense shifting, which seems to be the closest one to the older children’s, suggest that tense shifting may have discursive, organizational narrative purposes. In the narratives of the trilingual children tense shifts are characterized as : Hebrew narratives Child E A total of one tense shift for E from Past tense to Present tense. tafsu cfardea vesamim oto becincenet [they] caught a frog and put (pres. 3rd. pers. pl) it in a jar.

 Anat Stavans

One tense shift for E from Present tense to Future tense. hem mistaklim derex haxalon vekorim lo ulai hu yaxzor they look (pres. 3rd. pers. pl) through the window and call (pres. 3rd. pers. pl) to him perhaps he will come (fut. 2nd. pers. sing.) [back]. Child M A total of three tense shifts for M from Present tense to Past tense. vehem lo roim et hacfardea, hem yod’im shehu barax and they [do] not see the frog, they know that he escaped vehayeled megared et haaf biglal shehaaxbar axal lo et haaf and the boy scratches[to] nose because the mouse ate [to him] the nose veaz hayeled lokeax et hacfardea she-haya lo babait and then the boy takes the frog that was [to him] in the house English narratives Child E No tense shifts at all for E. Child M A total of three tense shifts for M from Present tense to Past tense. and the boy is looking on the doggie, the doggie fell the doggie look on the bees . . . the house of the beens, the bees fell the boy is looking on the deer, the boy and the doggie fell to the water Spanish narratives Child E A total of one tense shift for E from Past tense to Present tense. bueno cuando el niño y el perrito vieron eso, no sabian donde esta el sapo, y el niño y el perro decidieron encontrarlo well when the boy and the doggie saw that, [they did] not know where is the frog, and the boy and the dog decided to ﬁnd him A total of one tense shift for E from Past to Imperative tense. el niño oia unos sapos, le dijo al perro shhh, y (que) vea los dos sapos the boy was hearing some frogs, [to him] told to the dog shhh, and (that he should) look at the two frogs

Trilingual children narrating in Hebrew, English and Spanish 

Child M A total of one tense shift for M from Present tense to Past tense. el perro esta rascando el naris, y el niño llorando, el niño buscando el sapo en el arbol, y el niño y el perro buscando el sapo en el gardin, el [ . . . ] agarro el niño, el perro y el niño caen en la agua the dog is scratching the nose, and the boy crying, the boy looking for the frog in the tree, and the boy and the dog looking for the frog in the garden, the [ . . . ] caught the boy, the dog and the boy fall in the water

From these examples it is clear that both children are rather consistently using the respective preferred tenses as the matrix tense for the temporal casting of the events. E shifts tenses in Hebrew narratives from the present tense to the past or future keeping constant the present for plot advancement purposes (foregrounding devise). E uses shifting to the other tenses as a backgrounding devise describing a scene-setting (e.g., the Frog was caught prior to its placement in the jar); or an associated speculative state of mind of the protagonists which also opens the possibility for a prospective happy ending of the story (e.g., perhaps the Frog will come back). It is clear that E does not shift tenses at all in the English narrative and it is not clear whether the shift of tense in the Spanish narrative was purposeful. M’s tense shifts in Hebrew are all from the preferred tense — present — into the past. All of M’s Hebrew tense shifts occur from a main clause to a subordinate clause (i.e. beginning with she-) with the function of introducing background information. As for M’s English narrative tense shift, it seems to be the case that M does not know the verb ‘to fall’ in any other than the bare past tense form and that is the way it is inserted every time even in constructions that are meant to be present progressive such as ‘the boy is fell from the tree’. As for the tense shift in Spanish, M seems to display a similar backgrounding pattern as that of the Hebrew tense shifts, though it is not absolutely clear that this is the case. To sum, both children conform with the general developmental pattern of the same age monolingual cohorts who are consistent with a single tense throughout the narrative, but they tend to have fewer (if any) tense shifts altogether. Yet the few shifts in the Hebrew narratives of these trilingual seem to be consistent and functional in terms of the thematic and organizational needs of their narratives. This agrees with patterns found in some of the adult Hebrew monolingual narratives.

 Anat Stavans

Aspect forms The tense preference is the basic frame in which the narrative time is established. As such, it is an obligatory feature both formally and functionally in each of the three languages. Yet, each language avails diﬀerent means to express the perspective of time given to each event in terms of its durability, progression, termination or recurrence. In languages where aspect is not grammatically marked, as in Hebrew, other means of expressing aspect are available (i.e., aspectual adverbs, etc.). Given the diﬀerent means to mark aspect, the task of ordering events in a temporally cohesive frame may become more taxing for the trilingual. Decisions must be made as to the existence or lack of grammatical or lexical aspectual marking in each language. However, given the diﬀerent forms availed by each language, aspect may be a vulnerable structure for language contact. Hence, the trilingual may successfully separate, use, alternate and even over-use the aspectual system of each language. For the time being, I will focus on the grammatical aspectual markings the trilinguals exploit given the forms licensed in each language. Table 4 shows the distribution of aspect forms in each language by each child. This distribution is not at all surprising since it conforms to the patterns each language avails. The Hebrew aspect form is restricted to the simple tense (without grammatical aspect markings) while in Spanish and in English aspect forms are more varied. Let us turn to the trilinguals’ aspectual grammatical markings in the Spanish and the English narratives, which are the two languages that mark aspect grammatically with similar forms and distinct functions. Table 5 shows that both trilinguals predominantly use of the progressive aspect in English and Spanish. E exploits the English progressive in combination with the preferred tense thus producing a non-ﬁnite -ing form in: then they went ﬁnding him (the frog) in the woods, as do 9 years old monolingual cohorts in 30–50 per cent of the -ing forms they use. E’s Spanish narrative resembles the Hebrew one though Spanish provides many more aspectual grammatical

Table 4. Percentage of aspect forms by each trilingual in each of the languages

Non-ﬁnite Present Progr. Present Perfect

Trilingual E (9;8)

Trilingual M (7;10)

Hebrew English Spanish

Hebrew English Spanish

− − −

− − −

3 − −

− − −

− 57 4

− 68 −

Trilingual children narrating in Hebrew, English and Spanish 

Table 5. Use of progressive forms by each trilingual in English and Spanish

Total # of verbs Percentage of progressive forms Total progressive forms # of ‘bare’ progressive # of present progressive # of past progressive # of non-ﬁnite -ing forms

Trilingual E (9;8)

Trilingual M (7;10)

English

Spanish

English

Spanish

40 3 1 − − − 1

35 − − − − − −

47 55 26 5 21 − −

35 57 20 17 3 − −

Table adapted from Berman and Slobin 1994: 138, Table 2)

markings. M, on the other hand makes more extensive use of the -ing progressive form in combination with the present tense in both the English and the Spanish narratives. As the progressive aspect is the most exploited one in these trilingual English and Spanish narratives, a closer analysis of the progressive forms in these narratives (i.e., /-ing/ for English and /-ndo/ for Spanish) is called for. M exploits the progressive forms both in Spanish and in English. While it seems that M is fully aware that these languages avail such forms, it is not all that clear that M has a fully developed sense of their discursive function in the narrative. In M’s English narratives 55 per cent of the verbs are marked as the -ing progressive form combined with the present tense. This is not a typical pattern observed in the monolingual cohort of M’s age, but rather a pattern observed in the preschooled children. It may well be that M’s English (the less dominant language) has never undergone formal instruction and at that time M was primarily exposed to English in an informal oral way. Hence the experience which monolingual English children have with texts and narratives via formal and informal means was not one that M had at the time. Moreover, M’s use of the ‘bare’ -ing form seems to be a result of an unmarked aspect transferred from M’s dominant language rather than a ‘primitive’ use of the progressive as used by the younger monolingual children (Berman and Slobin 1994). M’s ‘bare’ progressive forms occur within a discourse tense contour of her preferred tense, as in: ‘‘the boy opens the window and the doggie looking for the frog ; the doggie runs to the garden’’.

M’s Spanish narrative is similar in its exploitation of the progressive aspect to that of her English narrative. Only the Spanish narrative contains a higher

 Anat Stavans

number of the ‘bare’ progressive forms than full progressive forms. For example: el niño y el perro durmiendo y el sapo saltando el niño y el perro no ven al sapo . . . y entonces el niño y el perro buscando el sapo . the boy and the dog sleeping and the frog jumping the boy and the dog (do) not see the frog . . . and then the boy and the dog looking for the frog

The simple present tense or any other tense/aspect forms are always ‘framed’ within a ‘bare’ progressive form. One possible explanation is that M’s Spanish narrative is an ‘Inter’language narrative. Since Hebrew and Spanish are similar in that they mark agreement of number and gender on the verb forms, M is creating a verb that will agree in number and gender (as in her dominant language — Hebrew) and also will have the progressive aspect available in Spanish (which is missing in the Hebrew tense/aspect formations). In all, the means by which each language marks temporality are partially deployed by these trilinguals. The more balanced trilingual, while having the wide array of aspectual markings in English and Spanish, ‘sticks’ to a single tense/aspect form and uses it throughout each narrative consistently. This pattern has been reported to be typical of the monolingual 9 year old narratives in all these languages. The more interesting case of the exploitation of these tense/aspect forms is most explicit in the narratives of M who is a less balanced trilingual and who has not been schooled in any of the less dominant languages (English and Spanish). Though M’s formal use of the aspectual markings in English and Spanish suggest an ‘immature’ narrative (relative to her age), the distribution of these forms in terms of narrative and thematic organization indicate that M is often using these forms to express both primary and secondary type of information. It is not clear (due to the scarce data set) whether M is attributing particular functions to these forms. However, the more ‘primitive’ use of these forms may be attributed to the fact that M is a less balanced trilingual and the transfer of some formal features of the dominant language together with other formal features of the ‘host’ language result in the innovative combination M produces. In all, temporality is clearly established in each narrative by means of tense (i.e. the relations each event bares to another event in the timeline) and aspect (i.e. the perspective the narrator chooses to give to the timeline within each event), each narrator has a preferred temporal frame into which s/he casts the story. At the same time each language avails to its user a set of possible

Trilingual children narrating in Hebrew, English and Spanish 

aspectual forms which acquire diﬀerent functions and purposes in a narrative (e.g. background and foreground).

Connectivity In discussing temporal cohesion so far we have dealt with the morphology and grammar of each clause and the clauses in its immediate discourse environment. However, the ‘text/ure’ of the narrative is determined by the ‘weaving’ of larger chunks of clauses and events in a coherent way. One way of ‘weaving’ events and creating a narrative ‘text/ure’ is through the connectivity devices one uses. Events, thus, can be connected by means of: syntactic coordination and subordination, nominalization and reference, and subject ellipsis. Hebrew, Spanish and English have similar, but not identical means of connectivity. The similarities may facilitate, for the trilingual, the construal of a cohesive text with ‘text/ure’, yet it is in the diﬀerences between these connectivity means that a trilingual is faced with a more diﬃcult task. All three languages create clausal combinations by means of markers, which express: deixis, sequence, subordination and coordination. Moreover, all three languages allow for connectivity between two clauses by subject ellipsis, though there are language speciﬁc forms, which translate into language speciﬁc discourse functions.

Connective markers Table 6 displays the percentage of clauses which begin with each connectivity marker. Both children use the coordinating marker by each subject and language, ‘and’ in all three languages extensively. This is not unusual to children of these ages as has been reported for monolinguals in of all three languages. For example, Berman and Slobin (1994) reported that English monolingual 9-year-olds use ‘and’ conjunction at clause initial position in 68 per cent of the Table 6. Percentage of clauses with connecting markers at clause-initial position Trilingual E (9;8)

Trilingual M (7;10)

Hebrew English Spanish

Hebrew English Spanish

Deictic Sequential Coordination Subordination

4 6 37 8

2.5 25 37.5 10

− 3 11 17

− 12.5 46 12.5

− 2 40 −

− − 31 −

Total

55

75

31

71

42

31

 Anat Stavans

clauses with the function of either connecting two clauses with the same subject, or introducing the next event. Hebrew and Spanish monolingual use the coordinating marker in similar ways for the same purpose. The use of deictic markers, a more juvenile form of cohesion, is particular to E. Both in Hebrew and in English, E chose to begin the story in a descriptive way as for example: ‘batmuna harishona anaxnu roim yeled vekelev vepo anaxnu roim . . .’ in the ﬁrst picture we see a boy and a dog and here we see . . .

E’s use of deictic markers is an inadequate ‘opening’ for the narrative and perhaps a ‘businesslike approach to the task’. Still these low percentages (which translate into a total of 2 clauses in Hebrew and 1 in English) are soon dropped and give way to the use of more sophisticated types of connectivity markers. The sequential markers as deﬁned and described by Berman (1988) are the next most commonly used and emerge later in the development of the narrative ability. English and Hebrew monolingual 9 years old seem to use and then, afterwards or after that to indicate that the coming event shares features with the previous one in terms of its occurrence on the timeline. E, the more balanced bilingual exploits these sequentiality markers in all three languages, as in: Hebrew: axarei ze hem yoc’im mehabait after that they exit (from) the house English: and then they saw his frog and his wife Spanish: entonces no lo encontraron. then (they) did not ﬁnd him (reﬂx)

However, the use of these markers in English (25 per cent) exceeds that of Hebrew (6 per cent) and Spanish (3 per cent). M, the less balanced bilingual, exploits these possibilities again in both Hebrew (12.5 per cent) and English (2 per cent) making more use of these forms in her dominant language, as in: Hebrew: az hem matxilim lisxot then they begin to swim English: and then the doggie and the boy look in the glass.

M does not use any marker other than the ‘additive’ and in Spanish, which suggests that these are not yet formally mature. The subordinating markers are the most frequently used in monolingual narratives by adults and school age children (Berman and Slobin 1994). The

Trilingual children narrating in Hebrew, English and Spanish 

subordinate clauses are formed much the same way in all three languages. In Hebrew there is a single subordinating marker she which functions the same way for all types of subordination (i.e., complementizers and relative clause). E uses more subordination markers in Hebrew and Spanish, while in English he prefers the sequential markers. Hebrew: ki hayanshuf shebaec doxef oto haxuca because the owl that is on the tree pushes him out-wards English : when he was in the deer’s antlers Spanish: cuando salieron afuera when (they) exited outside

M, on the other hand, uses equally the subordination and the sequential markers in Hebrew and avoids the sequential markers altogether in both English and Spanish. Hebrew: biglal she-ha’axbar axal lo et ha’af . because that the mouse ate to-him the nose

To sum up, E, the more mature and balanced trilingual, deploys connective markers much like his monolingual cohorts in all three languages. In doing so, E increases the use of the more mature forms of connectivity as is expected of a 9-year old, though in English there is a more extensive use of the sequential marker E, begins the narrative with deictic markings in the initial scene setting clauses and very soon drops those markers in favor of the sequential and subordinate markers. E uses the sequential markers as indicators of linearity rather than hierarchical or logical relations between the events. E uses the subordination markers to establish temporal, locative and causal relations between events. M, the less balanced trilingual, deploys connective markers much like a monolingual cohort in Hebrew — the dominant language. However, M avoids any connectivity markers other than the mechanically used and form with a purely additive function to ‘glue’ as it were the clauses into events. This use of the coordinating marker creates a narrative which is rather monotonous and long for it is made up of patched independent clauses which are only connected thematically through the protagonists and actions.

Subject Elision Another way of establishing connectivity is through tracking and maintenance of reference across the clauses and chunks of clauses. All three lan-

 Anat Stavans

guages provide for subject elision, though they diﬀer in the possibilities each language avails. In English as well as in Hebrew and Spanish, the Ø-Subject is allowed in conjoined clauses, that is the second clause has no overt subject. For example: She got in the car and drove away.

Still the languages diﬀer in the syntax used to express the subject. In Hebrew and in English a third-person subject pronoun can be added without changing the meaning of the second clause, as in: She got in the car and she drove away

while such addition will change meaning in Spanish . As Spanish and Hebrew mark number, person and gender on the verb, subject elision is common because referents are overtly maintained in the verb form. English, by contrast, has few number and person markings on the verb and require an overt subject in each independent clause. In the trilingual narratives three diﬀerent formations of Ø-Subject were found: Ø-Subject:

Clause 1, Clause 2 ( with Ø-Subject) Then he woke up, decided to go out.

Coord. + Ø-Subject: Clause 1 AND Clause 2 (with Ø-Subject) He ran outside, and climbed on the tree. Subord. +Ø-Subject: Clause 1 THAT Clause 2 (with Ø-Subject) There was a boy who run outside

Berman and Slobin (1994) report that for English monolingual 9-year-olds the use subject ellipsis in ﬁnite clauses with tensed verbs is an ungrammatical but conversationally accepted way (as in the Ø-Subject formation). The function of subject elision in English narratives of this age is a local grammatical devise and not a method of organizing information with a shared topic, much like children at this age inserting the ‘you know’ phrase without any intention to corroborate the interlocutor’s knowledge of the matter. Spanish diﬀers from English in that clauses with Ø-Subject are acceptable in both the spoken and the written modes. Tracking subjects between clauses is facilitated by the person/number and gender marking on the verb or by readily available contextual information. In Hebrew subject elision is not as uniform as in English. Subject elision in the Hebrew monolingual narratives are likely to

Trilingual children narrating in Hebrew, English and Spanish 

occur in the second formation as described above (Coord. + Ø-Subject) though subjects may be elided for diﬀerent discursive purposes taking other forms. The 9-year-old Hebrew monolingual uses syntactically well formed subject ellipsis formation of the Coord. + Ø-Subject and the Subord. + ØSubject type. Occasionally, children will use the Ø-Subject type to indicate connectivity between two grammatically separate but congruent clauses with a common topic. Table 7. Clauses with connectivity by null subject Trilingual E (9;8) Total clauses with Ø-S Percentage Ø-S clauses Percentage Coordinat.+ Ø-S Percentage Subordinat. + Ø-S

Trilingual M (7;10)

Hebrew English Spanish

Hebrew English Spanish

7 4 10 −

16 7 8 7

10 2.5 17.5 5

9 11 3 11

− − − −

1 3 − −

Table 7 shows the diﬀerent subject elision formations and their percentage of occurrence in each trilingual narrative. Both children exploit the possibility of connectivity through subject ellipsis in Hebrew and Spanish; E uses subject elision in English as well. The exploitation of subject elision in Spanish and Hebrew, regardless of whether the trilingual is balanced or not, seems to be a result of the alternative to overt subjects that both language enable by markings on the verb. As the subject elision is less licensed in English, M, who is less dominant in English, avoids such means of connectivity altogether. The elided subject in the second clause of a conjoined sequence is by and large the most popular means of connectivity if Ø-Subject is deployed. For example: Child E Hebrew: hu ole al sela veØ menase lexapes sham et hacfardea He climbs on a rock and Ø tries to search there for the frog. English: they tried andØ tried andØ didn’t ﬁnd it Spanish: el niño oyo unos sapos y Ø le dijo al perro shhh the boy heard some frogs and Ø told to the dog shhh Child M Hebrew: hacfardea yoce haxuca veØ boreax mehaxalo The frog exits out-wards and Ø escapes from the window English: — Hebrew: —

 Anat Stavans

In all, both children conform with the monolingual patterns of using the elided subject in conjoined clauses as means of tracking subject and topic through the narration. It is not at all clear for both monolinguals as well as these trilinguals whether a particular narrative organization function is attributable to these forms.

Multilingual trends When a multilingual person is asked to render two or more accounts of the same story, that person has at hand not only several tasks to attend to but also several options to chose from in order to formulate these renditions. We have thus far discussed in what ways the trilingual children in this study render three accounts of the Frog story, both formally and functionally. In terms of the linguistic form available to these trilinguals, we have seen how each language avails diﬀerent and similar ‘tools’ to carry out this narrative task. This array of ‘tools’ may or may not exist in one or more of the languages, yet when there is an overlap of forms there may not be an overlap of functions and when there is a lack of forms there may be compensatory devices for the diﬀerent languages. The trilingual must attend to all these possibilities, a task far more complex for trilinguals who are still in advanced stages of acquisition. Berman (1999) states that in adult bilingual narratives, the mature command of the narrative structure is unaﬀected by the proﬁciency of the speaker in a particular language and proposes studying the bilingual narrative productions as a continuum of non-native to native-like proﬁciency ranging from rules of linguistic structure (forms) to more discourse-motivated and culturally driven features of language use (functions). So far, this chapter outlines the form and functions that the trilinguals deployed in the narrative keeping in mind the narrative structure. Sketching of the trilingual narratives in terms of length, components, temporality and connectivity provides parallels between trilingual and monolingual productions. It has been a common practice in the last decades to compare the multilingual production with that of the monolingual in order to show that multilingual productions fall within the ‘‘norm’’. In this way research methods and conclusions were drawn in an egalitarian fashion. Still, there are very serious contentions to this way of perceiving and proving the ‘normality’ or rather commonplace quality of multilingualism. Many have strongly

Trilingual children narrating in Hebrew, English and Spanish 

argued that the multilingual is NOT the sum total of his monolingual cohorts (Grosjean 1985, 1992). In a recent article, Hernandez, Bates and Avila (1994) support this view and propose a diﬀerent way of looking at the multilingual production. In their study, Hernandez et al. show that adult Spanish–English bilinguals fall ‘‘in between’’ the two monolingual productions in an on-line sentence interpretation task. In their study, they argue that it is not enough to ﬁnd similarity and diﬀerences between the multilingual and the monolingual productions but rather to look at what the multilingual (language) processing is like. Hernandez et al. just as Berman and others, suggest an interpretation of multilingual productions as a continuum of ‘more or less’. Hernandez et al. propose four typological classiﬁcations to describe transfer and dominance of one language over the others which is common to multilinguals: diﬀerentiation (refers to the use of separate processing strategies for each language), forward transfer (refers to the use of strategies in the L1 or dominant language when processing the L2 or other languages), backward transfer (refers to the use of L2 or other languages’ strategies in processing the L1 or dominant language), amalgamation (refers to the combination of backward and forward transfer; or the combination of strategies used by the monolingual cohorts). This classiﬁcation suggests that a bilingual is not an ‘‘all or none’’ but rather a matter of degree of ‘‘more or less’’. With this view in mind, the trilingual children’s narratives presented thus far may be a reﬂection, more than anything else, of what type of trilinguals they are. In the case of these children we cannot talk about a ﬁrst or other language (as would be the case in the model proposed by Berman 1997), as they both were raised trilingual from birth. However, the ﬁndings thus far indicate that the command of these children in each of the three languages is not balanced. These trilinguals, in processing the narrative productions, showed some signs of diﬀerentiation, amalgamation and forward transfer which partially agrees with Hernandez et al ’s claim that: ‘‘in early bilinguals, the two most common patterns observed so far are forward transfer (L1 dominance) and amalgamation . . . full diﬀerentiation is the least common — although many individuals do display partially diﬀerentiated patterns of amalgamation.’’ (1994: 421). Lanza (1994, this volume) reported similar ﬁndings where three bilingual Norwegian–English girls’ narratives were analyzed for temporal and spatial expressions. What is at the center of this chapter is to see to what extent the forms and functions in each language interact or remain separated in the trilinguals’ use;

 Anat Stavans

and to what extent these interactions are a result of the children’s trilingualism or degrees of language dominance and transfer as suggested by the Hernandez et al. classiﬁcation.

Language dominance and transfer The analysis of the trilinguals’ narrative has shown that dominance of one or two of the languages over the third is relative and variable from language to language as well as from child to child. Hence we are dealing with degrees of dominance of each child in one of the three languages in relation to the child’s other languages and their monolingual cohorts. Moreover, we are dealing with typologically diﬀerent languages, which encode ideas by diﬀerent means, some more or less salient in one language as compared to another. One instance where language dominance — independent of language typology — is evident is in the length of the narratives both trilinguals produce. In all, the Hebrew and English are longer than the Spanish narratives produced by both children. Both children produce narratives in Hebrew and English, within the range of the narratives produced by the monolingual cohorts in those languages. In terms of language dominance it is clear that for both children Hebrew and English are more dominant than Spanish. This is also suggestive of the language choice in the order of renditions by the subjects whereby the narrative in the most dominant language is rendered ﬁrst (Hebrew), followed by the less dominant language (English) and lastly the least dominant one (Spanish). Another example where dominance and transfer seem of relevance is in the components which makes up the skeleton of the Frog Story. Both children mention in one way or another all six components in all three narratives; both children give greater emphasis to the ﬁrst three components: setting, initiation, and beginning of search (indoors). Temporality expressions are another example where the trilinguals show degree of their trilingualism. Both children’s preference for tense in the Hebrew production indicates diﬀerentiation among the languages singling out Hebrew (the dominant language). This diﬀerentiation is more overt in E who chose to use mostly the present tense for the Hebrew narrative and the past tense for the other languages. However, is not clear with M, as stated above in the quote from Hernandez et al., that the use of present tense as preferred tense in all the three language is not an indication of forward transfer. Another piece of evidence for dominance and transfer is in the trilinguals’ deployment

Trilingual children narrating in Hebrew, English and Spanish 

of tense/aspect forms lies in the children’s diﬀerentiation of the progressive in their other languages (Spanish and English) as compared to Hebrew. It is clear that the lack of durative aspect marking on the Hebrew verbs is more salient and thus easier to separate the strategies to deal with duration in Hebrew on one hand and in Spanish and English on the other. At the same time, M’s misuse of the progressive form often by using the ‘bare’ -ing in English or the ‘bare’ -ando in Spanish indicates both forward transfer and amalgamation. M’s progressive markings can be interpreted as forward transfer in that the copula BE is dropped and the V-ing pattern is kept as the inﬂected form of the verb much like in Hebrew (the dominant language). Then, there is forward transfer in the form of the verb inﬂection from the dominant language into the other languages. At the same time we may interpret M’s ‘bare’ progressive marking on the English and Spanish verbs and the omission of the compulsory copula as an amalgamation of salient forms and functions in the dominant language on one hand and the other languages on the other. Hernandez et al. model does not account for cases where there is no overt dominance or transfer from one language to the other but rather a lack of a strategy in one language which may exist in the other. In the case of these two trilinguals, an extrapolation of Hernandez et al. model may serve as possible frames for explicating some of the present ﬁndings. After all, language dominance and/or transfer may evoke strategic variations in multilingual processing. The way by which the trilinguals connect events in the text suggests the hierarchical nature of language dominance and transfer patterns. Both children’s Hebrew narratives deploy the same types and frequency of usage of connectivity markers as those produced by Hebrew speaking monolinguals. However, E shows a more immature (relative to his age) connectivity through more use of sequential markers compared to subordination markers. This indicates the frailty of E’s English. In the case of E’s connective markers it is clear that E avoids the subordinating markers in favor of the less mature connectivity forms because he is less comfortable with those. M on the other hand avoids many mature connectivity forms in her other languages though she shows clear signs and capabilities to use them in her dominant language. The Hernandez et al. model accounts very appropriately for the way by which subject elision is used as connectivity in the narratives. Typologically, both Hebrew and Spanish allow much more for subject elision than English does. This is clear in the diﬀerentiation that both children make by using subject elision to connect events in the Hebrew and Spanish narratives. However,

 Anat Stavans

E also uses subject elision (in the form licensed in colloquial English) which indicates amalgamation.

Conclusions and discussion Over the last few decades studies in bilingualism have aimed at singling out each of the bilingual’s languages and compare these with monolingual cohorts. Lately, scholars have proposed that such a view of bilingualism suggested indirectly that the monolingual is the ‘template’ for normal language processing and the aim of research was to establish the extent to which bilinguals were similarly ‘normal’ to their monolingual cohorts (Berman and Slobin 1994). These more recent studies focused on looking at the bilingual as a normal multi-language user. The aim of this line of research was to study the development, processing and production of bilinguals independent of monolingual productions. However, often these studies falls in the pattern of comparing monolingual productions with bilingual productions as means of methodological ‘control’. This clearly shows that it is inevitable to look at multilinguals without looking at monolinguals, suggesting that a combination of both views is in order. On the one hand, we must look at the bilingual production as a diﬀerent, albeit normal language production, which proﬁts from the beneﬁts and shortcomings of each of the languages and of the possibilities of combining them. On the other hand, we do accept that bilingualism which is becoming more the norm than the exception, is a less common phenomenon especially in early stages of language development, and therefore it is important to look at the bilingual child’s language development vis a vis the monolingual one — at least as long as becoming bilingual from birth is not yet a widespread practice. In addition, it is often claimed that it is not clear whether trilingual processing is diﬀerent from bilingual processing. The layman would claim that trilingualism is more taxing and therefore the more languages you know the slower you process. This claim can be refuted with simple examples that demonstrate that polyglots use language ﬂuently and competently. With these caveats in mind and adopting a more eclectic perspective on child multilingualism, the study of the two trilingual children’s narrative productions in the three languages was carried out. The similarities and diﬀerences were compared in terms of their form and function relations

Trilingual children narrating in Hebrew, English and Spanish 

within each language (i.e. compared to monolingual productions) and across the languages (i.e., multilingualism issues). Special emphasis was placed on means of expressing and maintaining temporality and connectivity throughout each narrative. The results show that in the case of these two trilingual children the manipulation of form-function relationships available in each language is used in similar ways to that of a monolingual. Diﬀerence or divergence in usage of these form-functions relationships suggests an exploitation of the various linguistic means available to the child in each language. Moreover, these diﬀerences may be attributed not only to the interaction among the various languages but also to psycholinguistic factors such as: rendering the same story more than once, or the extent to which the child is equally able in all three languages. The trilingual renditions suggest that linguistic, narrative and expressive ability are intertwined in a way which creates a sliding-hierarchy of language dominance. Such trilingual-based sliding hierarchy is diﬀerent from a bilingual ‘binary’-choice by virtue of the number of competing languages and by time. That is to say, dominance in trilinguals is based on a network of dual-language relations (i.e., English vs. Spanish) which may be true for some elements of language (i.e., temporal frame which is either more salient or simple to process) and depends on time (i.e., one may have dominant language at age 7 which may become less dominant at age 9 given social and geographical changes). The present study is exploratory in nature and as such it is not exhaustive in its ﬁndings. The number of subjects is very small and the amount of data cannot be more than suggestive. A large-scale study should be conducted addressing not only developmental issues but also crosslinguistic and multilingualism issues. From a psycholinguistic perspective we should be able to come up with more clear description as to bilinguals’ processing. Perhaps by looking not only at bilinguals but also at multilinguals with diﬀerent command in more than two languages which are either typologically similar or diﬀerent. From a linguistic point of view, a more in depth analysis of which forms are more ‘permeable’ to be exploited in one speciﬁc language may suggest that it is not only the processing that enables the multilingual to manipulate the systems but rather that the system is more or less ‘fragile’ in certain points of its structure. The narrative provides an excellent means to study the various language speciﬁc and language universal domains of formfunction relations as well as discourse variation and style.

 Anat Stavans

References Bates, E. and B. MacWhinney (1982). Functionalist approach to grammar. In: E. Wanner and R. Gleitman (eds.), Language Acquisition: The State of the Art. New York: Cambridge University Press, 173–219. Bates, E. and B. MacWhinney (1987). Competition, variation and language learning. In: B. MacWhinney (ed.), Mechanisms of language acquisition. Hillsdale, NJ: Lawrence Erlbaum. Bates, E. and B. MacWhinney (1989). Cross-linguistic studies of sentence processing. New York: Cambridge University Press. Berman, R. (1988). On the ability to relate events in narrative. Discourse Processes, 11, 469–97. Berman, R. (1999). Bilingual proﬁciency/proﬁcient bilingualism: Insights from Hebrew– English narrative texts. In G. Extra and L. Verhoeven (eds.), Bilingualism and migration. Berlin: Mouton de Gruyter, 187–209. Berman, R. and Y. Neeman (1994). Development of linguistic forms: Hebrew. In: Berman and Slobin (eds.). Berman, R. and D. I. Slobin (eds.) (1994). Relating events in narrative: A cross-linguistic developmental study. Hillsdale NJ: Lawrence Erlbaum. de Houwer, A. (1990). The acquisition of two languages from birth: A case study. Cambridge: Cambridge University Press. de Houwer, A. (1994). Bilingual language acquisition. In: P. Fletcher and B. MacWhinney (eds.), The Handbook of Child Language. Cambridge: Blackwell, 219–51. Genessee, 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. Grosjean, F. (1992). Another view of bilingualism. In: R. Harris (ed.), Cognitive processing in bilinguals. New York: Elsevier, 51–87. 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. Lanza, E. (1992). Can bilingual two-year-olds codeswitch? Journal of Child Language, 19, 633–58. Lanza, E. (1994). Telling the same story twice: Expressing time and space in two languages. Paper presented at NELAS 3 (Northern European Language Acquisition Symposium), July. Reykjavik, Iceland. Meisel, J. (1989). Early diﬀerentiation of languages in bilingual children. In: K. Hyltenstam and L. Obler (eds.), Bilingualism across the lifespan. Cambridge: Cambridge University Press, 13–40. Mayer, M. (1969). Frog, Where are you? New York: Dial Press. Minami, M. and A. McCabe (1995). Rice balls and bear hunts: Japanese and North American family narrative patterns. Journal of Child Language, 22, 423–45. Paradis, M. (1987). The assessment of bilingual aphasia. Hillsdale, NJ: Lawrence Erlbaum.

Trilingual children narrating in Hebrew, English and Spanish 

Sebastián, E. and D. I. Slobin (1994). Development of linguistic forms: Spanish. In: Berman and Slobin (eds.). Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10, 209–31. Stavans, A. (1990). Codeswitching in two trilingual children acquiring Hebrew, English and Spanish simultaneously. Doctoral Dissertation, University of Pittsburgh. Stavans, A. (1997). Input eﬀects on trilingual narratives in Hebrew, English and Spanish. Unpublished manuscript. Stavans, A. (Forthcoming). A comparison of bilingual and monolingual narratives. Verhoeven, L. (1993). Acquisition of narrative skills in a bilingual context. In: B. Kettemann, and W. Wieden (eds.), Current issues in European second language acquisition research. Tübingen: Gunter Narr. Volterra, V. and T. Taeschner (1978). The acquisition and development of language by bilingual children. Journal of Child Language, 5, 243–64.

Chapter 14

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

In the course of a story told in response to Mayer’s wordless picturebook, Frog, Where are you? (1969), children have the opportunity to display their awareness of the story characters’ thinking and to demonstrate the linguistic means they have available for expressing that awareness. Such stories permit us to explore relationships between children’s theories of other minds, the language that encodes ideas about ideas, and the conditions which prompt its use. By looking at stories from bilingual children with diﬀerent levels of language proﬁciency in two languages but presumably a single cognitive level per individual, we can explore the nature of the relationship between particular cognitive achievements and their linguistic expression. Such an investigation sheds light on developmental changes in the use of and reference to logic and other forms of thought in children’s naturalistic speech and suggests areas for follow-up through experimentation. The direction of the relation between cognitive and linguistic development is an ongoing debate. Does cognitive growth drive linguistic growth, as a Piagetian model would suggest? Or does language play a more enabling role in the development of children’s reasoning? A speciﬁc mechanism by which language could lead cognition has been suggested in recent research on what is called the ‘‘Theory of Mind’’ by J. de Villiers and her colleagues (de Villiers and Pyers 1997a; Gale, de Villiers, de Villiers, and Pyers 1996). These authors propose that sentences with full noun-clause complements for mental predicates (to know, think, pretend, etc. 〈that something is so〉) not only communicate about the contents of other minds. They claim further that the syntax of such sentences provides the mental machinery necessary to hold the two ideas in mind simultaneously, having separate referents and a separate truth value for each. The ability to have two separate linguistic representations for the same

 Barbara Zurer Pearson

event (described by Nelson 1996) appears crucial for success in formal propositional reasoning. The ﬁrst manifestation of this skill may be the ability to reason about the false beliefs of others, generally observed to develop between ages 3 and 5 (Wimmer and Perner 1983; de Villiers and Pyers 1997a). It is also important in coordinating information from diﬀerent sources about a single situation. Such reasoning requires the individual to diﬀerentiate and coordinate what Sokol, Muller, and Overton (1997) call ‘‘propositional’’ and ‘‘meta-propositional’’ levels. That is, children must comprehend and encode in their sentences both the facts of a scene and relevant modal information about it; for example, how credible the events represented are. These notions can be expressed lexically, in separate words, but they are most generally ‘‘folded into’’ the choice of grammatical form and expressed through the linking of multiple clauses. It is well known that children begin using sentences composed of more than a single clause well before age 5 (Bowerman 1979), but the complexity of the multi-clause units expands with exposure to literate academic texts once children are being schooled (Scott 1988). In particular, the meanings of sentence connectives continue to develop through the teen years (Nippold 1988); relationships between clauses become more elaborate and dependencies across clause boundaries are more frequent (Hunt 1977). The choice of subjunctive mood in Spanish, for example, is most often a function of the verb in a previous clause, or as described by Perez-Leroux (1996) for relative clauses, of diﬀerent presuppositions about the existence of the entities referred to. Also, as children become more conversant with factivity, they become more aware of how the truth value of embedded clauses can change depending on the verb in the adjoining clause. So, an assertion like ‘‘John failed the test’’ will be necessarily true if it comes after ‘‘we know,’’ necessarily false if embedded with ‘‘we pretended,’’ and indeterminate (but not likely) with ‘‘we thought.’’ More than in dyadic conversation, stories children hear and produce provide a context for linking propositions with logical ties and cohesive devices. As children get older they are more likely to report on others’ evaluations of events and to explain actions with reference to the intentions and acts of ‘‘other minds’’ using metacognitive statements, talking about what someone else said or thought, (Kemper 1984). In the case of the Mayer frog story, it is the lack of logical thinking on the characters’ part that may prompt children to comment on such thinking.

Logic and mind in Spanish–English children’s narratives 

The role of logic and theory of mind reasoning in the Frog story When the boy and his dog lose their pet frog in Frog, Where are you?, they do the logical thing: they look ﬁrst where the pet had been kept, in its bowl and all around their room for it. When they do not ﬁnd it there, though, they do not stop and think logically. They could say to themselves: ‘‘Frogs like to be near water. Our frog is a typical frog. Therefore, we should look for our frog near some water.’’

Translating this little syllogism into a plan of action, they might have asked themselves: (1) Where is a froggy body of water? and (2) How do we get there? Instead, they let a combination of chance and childish curiosity draw them in small steps away from their house and closer to water. They don’t actually decide to go outside: the dog falls out and the boy goes out after it. Then they go into the woods, conveniently located right outside their window. But once there, they look ﬁrst in places a frog is unlikely to go, up in a beehive or an owl’s nest, for example. Thankfully, though, the bees and the owl chase the pair further into the woods where the boy manages to fall onto a deer who throws him unceremoniously into a small body of water. He is soon joined there by the dog, who once again falls where he would have gone had he thought about it. Here they ﬁnd a frog, possibly their original pet, and the story resolves with a happy ending (for the boy, not necessarily for the frog). So the story ends up following a reason-able course, but without any apparent planning from the protagonists. It is not a cognitively taxing story: the action depends on motivation and reaction, with little explicit reasoning. The boy’s lack of reﬂection — and of course the dog’s — provide for much of the plot complication. Indeed, if there were more logical thinking involved, there would be less story to tell. As children get older, we may expect them to notice the boy’s lack of logic and make increasing reference to thinking and inference even though such references are purely optional in the recounting of the story. That is, as the distance grows between the characters’ mental acts and the (child) narrator’s own awareness, it becomes more likely that the characters’ state of knowledge will appear newsworthy to the narrator, provided she or he has the linguistic means to express it. So, Frog, Where are you? is almost entirely an action story, but there is room and rationale for reference to cognition in it.

 Barbara Zurer Pearson

At least three points in the story speciﬁcally invite reference to the characters’ state of awareness. First, in panel 2, the boy and dog are conspicuously asleep when the frog climbs out, so when they awaken, they must have a speciﬁc realization of the frog’s disappearance before a search for it can begin. Later, in panels 13 to 15, there is a depiction of false belief on the boy’s part. Before the boy ﬁnds the deer, he has climbed up on a big rock and grabbed onto what appear to be branches. By the time he realizes they are not branches, but antlers, he is already high atop the deer’s head and about to be carried oﬀ by the animal. Finally, when the boy and dog have landed in the water at the end, the boy is shown with his hand cupped around his ear, a conventionalized representation of ‘‘listening.’’ He hears something probably froglike before he can see the frogs, and he must interpret the sound as a cue for the search. In their corpus of frog stories in ﬁve languages, Berman and Slobin (1994) found that about three quarters of the adult stories made speciﬁc mention of the ﬁrst of the realizations above, using a mental verb like ‘‘realize,’’ ‘‘discover,’’ etc. (pp. 53–4). By contrast, only about half of the adults made explicit reference to the boy’s false belief (p. 55). The failure to do so may be a result of the way the story is drawn. On close examination, the antlers are very deliberately drawn to suggest branches, and they are also repeated in exact detail in the next panel where it is clear that they are antlers. The drawings in general, though, are small and rather sketchy, so they discourage strict attention to the small details. Furthermore, explanation of the boy’s misperception is not absolutely required for the story to make sense. Some perfectly mature stories use the element of surprise instead: ‘‘then up popped a deer’’ or ‘‘along came a deer.’’ On the other hand, the story provides a clear opportunity for the narrator to remark on the boy’s false belief. Ending up on a large animal which could bring the boy to the pond where he needs to go is a central point in the story, and so it is a strong candidate for inclusion. Berman and Slobin (1994, chapter IIA) present evidence that older narrators are more likely to use this scene in their stories than are younger narrators. In their analysis of the deer scene among the 12 English and 12 Hebrew subjects at each age, there is a tendency for more mentions of the false belief among older subjects. For example, only four ﬁve-year-olds (of 24) even imply the boy’s mistake about the antlers, compared to 13 nine’s and 20 adults. For those who explain it explicitly, Berman and Slobin report 2, 3, and 12 respectively at each age. There is a

Logic and mind in Spanish–English children’s narratives 

similar developmental trend for the use of an explicit mental verb in recounting the boy’s discovery of the missing frog: from 2/24 at 5 years, to 16/24 at 9 years, and 17/24 for adults (pp. 53–4). Their study, though, is largely qualitative and there is no statistical testing of the suggested trends. Along with the more advanced cognition required to report the boy’s false belief in the story, there is also a call for more complex linguistic structures capable of conveying the more complicated message. In particular, the usual syntax to express this notion would be a full noun-clause complement: ‘‘He thought (that) it was a branch, (but it turned out to be the antlers of a deer).’’ As explained in de Villiers and Pyers (1997a), full noun-clause complements provide the means by which a false proposition can be embedded under a verb and yet the whole sentence will remain true. Thus, ‘‘it was a branch’’ can be false, but the sentence ‘‘He thought it was a branch’’ is still true. (This is the same for other intensional verbs, like ‘‘believing,’’ ‘‘supposing,’’ etc. and also verbs of communication like ‘‘he said.’’) De Villiers and her colleagues note the extremely close coincidence in the time between acquisition of full-clause complements and the ability to pass false belief tasks. What de Villiers calls the ‘‘usual orthodoxy of cognitive determinism’’ would oﬀer the explanation that understanding beliefs and states of mind is a prerequisite for using the linguistic forms correctly (de Villiers and Pyers 1997a, p. 1 in ms). She and her colleagues propose a more radical interpretation of the close correlations, a ‘‘strong form of linguistic determinism’’ whereby the mastery of the ‘‘linguistic forms of complementation provides the representational structure for handling false belief reasoning’’ (p. 1). Through their experiments, they provide plausible support for this stronger position. In particular, in a test of 3 to 5-year-olds, the occurrence of noun-clause complements — and not other complex syntax — in the child’s spontaneous speech was the best single predictor of passing false belief tasks (r 2=.47, p <.01), while the reverse prediction, i.e. using passing false belief to predict complement syntax yielded an r 2 of only .095, which was nonsigniﬁcant. De Villiers and Pyers claim that this asymmetry supports the conclusion that ‘‘a certain level of mastery of complements is a prerequisite for false belief and not vice versa’’ (p. 8). In making this claim, though, they make what might be an overstatement: that ‘‘[without] the grammar and semantics of an embedded complement, [children] have no system to represent and reason about others’ false beliefs’’ (p. 1). According to such a statement, noun-clause complements are a neces-

 Barbara Zurer Pearson

sary and possibly even a suﬃcient condition for reasoning about false beliefs. Such a claim would be weakened if one routinely found false belief reasoning in stories that do not include noun-clause complements.

The present study The current study explores the strength of the relationship between false belief reasoning and noun-clause complementation in a story generation context. We focus on two scenes of the frog story where there is speciﬁc attention to other minds, the discovery of the frog’s escape and the deer episode, as well as other language about thinking throughout the stories. The children are English- or Spanish-learning monolinguals and bilinguals at the elementary school level. They are older than those typically studied in connection with the issue of false belief and noun-clause complementation (de Villiers and Pyers 1997a), but following Berman and Slobin (1994), we are not so much interested in documenting the ﬁrst emergence of linguistic forms. Rather we seek to observe how speciﬁc forms gradually become recruited for developing functions. Children must not only be able to produce the forms at a local sentence level, but also in the more demanding situation of organizing a narrative at the global level (Berman and Slobin, Part IV). As mentioned above, children can tell the plot of the frog story quite adequately with little or no reference to cognitive states. In previous studies (Pearson 1996), we found that the average second-grade monolingual subjects (ages 7 and 8) did just that: ‘‘this happened and then this happened and then this happened.’’ But as their own thinking becomes more mature, and as they become more aware of their own reasoning, children appear more inclined to remark on the thinking they infer on the part of the characters whose exploits they are describing. Following de Villiers et al., we hypothesize that children with greater ability in the speciﬁc syntactic domain of the noun-clause complement will be more likely to include references to beliefs and logic in their frog stories. Therefore, we examine the children’s use of logic and of noun clauses, and the relation between the two. Clearly, we will not be claiming that false-belief ‘‘mentioning’’ in the stories is equivalent to false belief ‘‘passing’’ in the sense used by de Villiers (Wimmer and Perner 1983; Perner, Leekam, and Wimmer 1987). Nor will the actual utterances of the children in the stories be considered an exhaustive reﬂection of what the children could produce in the noun-clause domain, say

Logic and mind in Spanish–English children’s narratives 

under experimental conditions. Nonetheless, it is interesting to explore the children’s performance in both areas while under the pressure to produce a narrative, where the story-telling context can provide a unique perspective on emerging skills and their relationships. Our ﬁrst goal in this study is to evaluate all expressions of logic and metacognitive knowledge in stories told by children ages 7 and 11 to discover whether there is observable growth in this domain within this age range. We replicate Berman and Slobin’s false-belief and mental-verb-use ﬁndings with a larger group of children. We relate mentions of logic and false belief to the demographic and linguistic factors in the children’s background, especially age, socioeconomic status (SES), and language background. A second objective is to examine the syntactic devices used by the children to express the contents of the characters’ thoughts. We look at how often the full-clause noun phrase is used to embed the false belief within a larger proposition, as suggested by de Villiers and Pyers (1997a 1997b). To test de Villiers’ claim that linguistic growth precedes cognitive growth, we make use of the fact that some bilingual children can have diﬀerent levels of skill in their two languages. That is, we presume that each child brings to the task a single set of general intellectual and conceptual abilities, but because of diﬀering amounts of exposure to the two languages children may exhibit a disparity in the level of expression of noun clauses in the two languages, some stronger in Spanish and others in English. Furthermore, we examine the possibility that the developmental trajectory of the noun-clause is diﬀerent in English and Spanish, even for monolinguals, and thus may play a diﬀerent role in supporting expressions of logic. In a study of ‘‘that-trace,’’ (a component of nounclause syntax that governs whether the complementizer will be expressed after a movement operation), a group of Spanish monolingual children were shown in a grammaticality judgment task to have more knowledge of that-trace expression than a matched group of English monolingual children (Gathercole 1997; Gathercole and Montes 1997), presumably because the presence of that (or que in Spanish) was more consistent in all its uses than that in English. We conﬁrm whether this pattern of greater noun-clause use in Spanish than English is present in the stories of monolingual and bilingual children. Our third goal is to examine the relation between the two domains. We explore the eﬀect of false belief expression on other qualitative aspects of the stories. With separate stories in English and Spanish from the children, we evaluate the contribution of diﬀerent levels of language ability to how the children chose to recount the meta-cognitive aspects of the story. In particular,

 Barbara Zurer Pearson

we look for evidence that children are more likely to report the false belief in a language where they demonstrate full noun-clause complements than in one where they do not use full clauses, or in which their clauses are in some way defective. We also evaluate whether the use of the full noun clauses has a diﬀerent predictive relationship in the two languages, due in part to the diﬀerent status of the embedded clauses in the syntax of the two languages (Gathercole and Montes 1997) or for the two language groups, monolinguals and bilinguals.

Method Participants The primary data are a corpus of 257 frog stories in English and Spanish collected from 79 monolingual and 89 bilingual elementary school children in Miami, Florida. In addition, 43 stories in the CHILDES archive (MacWhinney 1995) from Spanish monolingual children were consulted. The Miami children took part in a larger study of language and literacy development which used standardized and non-standardized tests to assess academic development in both English and Spanish. A quasi-experimental factorial study with two nested factors pro-actively controlled for language of the home, language of instruction at school, and socio-economic status. In the full study, there were 20 to 40 subjects in 8 bilingual cells (2×2×2, language of the school, language of the home, and SES) and 2 monolingual cells (SES, high versus low) at each of three grades, kindergarten, second, and ﬁfth (average age 5, 7, and 11 years). All of the children in the Miami study were born in the United States. Testing took place in ten schools in Miami, equated for teacher preparation, school size, expenditure per pupil, and other demographic variables published annually by the Dade County School Board. Two schools where instruction is given in both English and Spanish approximately equally in what is called ‘‘two-way’’ bilingual education (Christian and Maher 1992) were the basis of the matching. They were matched demographically by four schools serving the same populations, but where instruction follows a more standard pattern with all instruction in English, (except an optional half-hour a day of Spanish for Spanish-speakers). The monolingual English children were in schools matched to have a similar curriculum to the schools with Hispanic children as well as similar teacher characteristics and pupil expenditures. The determination of

Logic and mind in Spanish–English children’s narratives 

socio-economic status was made from parent report of years of education and current occupation following a procedure adapted from the Hollingshead (Eilers et al., 1993). Frog stories were collected from groups of 9 to 12 each in the 8 bilingual cells and 18 to 22 each in the 2 monolingual cells at second and ﬁfth grade (ages 7 and 11). (All of the stories are archived on the 1997 CHILDES CD and Website (MacWhinney 1997) and are available in print versions from the author.) In all, there are 159 English stories from bilingual children, 159 Spanish stories from the same bilingual children, and 79 stories from monolingual English children. For the current project, 79 stories from the monolingual English children and 178 stories (89 in English and 89 in Spanish) from the bilinguals were analyzed (plus the 43 monolingual Spanish stories from CHILDES). Language of the school and language of the home were not used as separate variables, but were balanced across high and low SES at the two ages, 7 and 11. The 43 Spanish monolingual children, 22 nine-year-olds and 21 ﬁve-year-olds, come from Argentina and Spain (MacWhinney 1995, 1997; Bocaz 1987; Sebastian 1989). The stories from Spain are from children ‘‘from middle-class literate backgrounds growing up in urban, industrialized settings,’’ (Berman and Slobin 1994: 28). The Argentines are from a university community. They were all classiﬁed for our purposes as high-SES.

Procedures Data collection Children told the story according to the standard protocol proposed by Berman and Slobin (1994: 22–3). They looked through the book once in advance and then told the story as they turned the pages of the book a second time. The bilingual children told the story twice — in one language on one day and in the other about one to two weeks later, with language order counterbalanced across subjects and groups. To evaluate the eﬀect of telling the story twice on the measures we were collecting, about one-third of the monolinguals also told the story twice after the same interval of time (Pearson 1996). The Miami stories were transcribed and checked by two bilingual research assistants. Fifteen percent received two independent listenings for reliability checking. Discrepancies were resolved by consensus. The stories averaged 275 words in length, with a mean number of 45 verbed clauses. The monolingual stories from ﬁfth graders were shorter on average than those of monolingual

 Barbara Zurer Pearson

second graders (259 versus 324 words) while the bilinguals’ stories got slightly longer from second to ﬁfth grade, from an average of 246 to 280 words in English; 230 to 250 in Spanish (Pearson 1997). In addition, standardized tests in English and Spanish from the Woodcock Language Proﬁciency Battery (Woodcock 1991; Woodcock and Muñoz 1995) and the Peabody Vocabulary Tests (Dunn and Dunn 1981; Dunn, Padilla, Lugo, and Dunn 1986) were administered to all the Miami subjects.

Data coding Stories were coded for false-belief mentioning, for use of a mental verb for the discovery of the frog’s escape, and for the presence of full-clause noun complements as follows. For false belief mention, we found the following steps: Score Criterion 0 No mention of the boy’s mistake (sometimes no mention even of the deer). 1 Some children refer to the change, but not clearly. Ex. ‘‘He was holding a branch. It was a deer.’’ 2 Another set are clearer about the change, but do not mention whether the boy is aware of his mistake. Ex. ‘‘He holded the branches that weren’t really branches.’’ 3 Finally, the most mature level would be to mention both the boy’s misapprehension and his realization of it. ‘‘Y se aguantó y no eran palos, era(n) un venado, pero él no lo sabía’’ [He grabbed on and they weren’t branches, they were a deer, but he didn’t know it] or ‘‘He didn’t realize he was holding a deer’s antler.’’

Thus, in evaluating the false-belief mention for our database, we made four levels: 0–3, representing the levels above. For the mental verb, the coding was similar: Score 0 No speciﬁc mention of the boy’s awareness. ‘‘The boy gets up and starts looking everywhere for the frog.’’ 1 The boy’s realization can be inferred from the boy’s looking at the bowl, but it is not explicitly stated, as in ‘‘The boy looks [at the bowl], and the frog is not there.’’ 2 An explicit mention of the boy’s discovery, as in ‘‘He realized that the frog was gone.’’ (Statements that he was upset, worried, or surprised, that the frog was gone were considered to include the realization of the disappearance even without a speciﬁc ‘‘thinking’’ verb.)

Logic and mind in Spanish–English children’s narratives 

A similar hierarchy was established for noun clauses found in the stories. A simple non-clausal noun-phrase object was given zero noun-clause credit. The highest score was given to the prototypical noun clause for mental verbs or verbs of communication, with an intermediate value for formulations that were either ambiguous or in someway ill-formed. Score 0 No clausal complement, as in ‘‘He told the story,’’ even with extended noun phrases or small clauses, as in ‘‘He saw the bowl without the frog’’ or ‘‘He saw his frog gone.’’ 1 Complements with a non-ﬁnite verb (‘‘he told the dog to stop talking’’) or with an embedded clausal element that was not a proposition (capable of bearing a truth value), as in ‘‘. . . to see if the frog was there’’ or ‘‘. . . to see what it was.’’ 2 A ﬁnite verb clause embedded under the main verb: ‘‘he thought that the deer’s horns were the branches of a tree’’; ‘‘he realized that the frog was gone.’’

An intermediate score was also given for direct discourse when it included a ﬁnite verb phrase: ‘‘He said, ‘Where is the frog?’’’ or in ambiguous cases without a complementizer ‘‘He thought maybe the frog is here,’’ or when the noun clause used was ill-formed in some way. Thus ‘‘he told him *to don’t talk’’ was coded with a ‘‘1.’’ The same coding principles were used for Spanish. Score 2 The full complement structure, ‘‘ven que no está la rana’’ [they see that the frog is not there], or ‘‘creyendo que eran unas ramas’’ [thinking that there were some branches] 1 Intermediate structures, ‘‘ver si está la rana’’ [to see if the frog is there]. Also an intermediate noun-clause form which was not the intensional context: ‘‘viendo qué había abajo’’ [seeing WHAT was below], or ‘‘diciendo que porqué habían entrado’’ [saying *that why had they entered], or the ambiguous ‘‘no sabe el niño que está aquí’’ which could be ‘‘. . . qué está aquí’’ [the boy doesn’t know that it is here, or perhaps . . . what is here].

The Spanish children also gave evidence of some problems with the forms, or perhaps beginning to use a noun phrase and then changing to a full clause. Thus we saw: ‘‘ven al sapo que él había juntado con una novia’’ where one might have expected ‘‘ven que el sapo había juntado . . .’’ [they see the frog that he had gotten a girlfriend, instead of they see that the frog had gotten a girlfriend];

 Barbara Zurer Pearson

or ‘‘el niño decía que se callase el perro y que silencio’’ [the boy told the dog to be quiet and *that ‘‘(be) quiet’’]

and ‘‘vieron la botella que él se fue’’ [they saw the bottle *that he was gone].

Stories were also coded through the CLAN programs (MacWhinney 1995) for mean length of utterance (mlu), number of clauses, and a subordination index (number of clauses per sentence). Another set of codings compared stories on ten measures of cohesion, lexicon, and syntax developed for the project (Pearson 1996). These encompassed elements like how well the story events were sequenced, how clearly reference to the characters was maintained, and how elaborate the children’s sentence structures were.

Data analysis 1. To evaluate the logical and metacognitive statements in the stories, all instances of explicit logical language were identiﬁed. Tokens of modals (‘‘would,’’ ‘‘must,’’ etc.) and intensional verbs (‘‘think,’’ ‘‘wonder,’’ ‘‘realize,’’ etc.) were tallied for each age, socio-economic group, and linguality (monolingual versus bilingual). False-belief mentions in the deer scene and use of a mental verb in the discovery scene were also tallied and compared to the Berman and Slobin (1994) ﬁndings. Incidence of these three types of cognitive expressions for each subgroup by age, language, socioeconomic status and linguality was tested with chi-square statistics, using the values in Berman and Slobin’s Chapter Two for the expected values. 2. The use of full noun-clause complements and intermediate forms was tallied according to the same design, and the groups were compared to each other by chi-square. 3. To evaluate the necessity and suﬃciency of noun-clause usage for falsebelief mentioning, as suggested by de Villiers, the co-occurrences of noun clauses and false-belief mentions were tracked by language and language group and the correlation between them was computed. Correlations were also computed between false-belief mentions and other syntactic and narrative measures derived from the Miami corpus through CLAN analyses. These correlations were done for the whole Miami group and then for the monolinguals and bilinguals separately. 4. A ﬁnal analysis evaluated our hypothesis of how strongly the ability to encode cognitive elements in the stories was tied to the child’s level of expression generally. Stories with false-belief mentions were compared to those

Logic and mind in Spanish–English children’s narratives 

without false-belief mentions on a host of other qualitative measures. Some of those other measures were taken from the stories and others were the standardized measures of language (expressive and receptive vocabulary, and reading) tested in the parent grant (Eilers 1997; Dunn and Dunn 1981; Dunn et al., 1986; Woodcock 1991; Woodcock-Muñoz 1995). Diﬀerences on these measures between the two sets of stories were evaluated for signiﬁcance by t-test. A similar analysis compared stories that used noun clauses to those without noun clauses and those with intermediate noun-clause forms.

Results Logic and metacognitive language The language of logic As a whole, there was a fair amount of informal reasoning in the stories about causes and motivations based on expectations created by the children, but these were not couched in the typical language of logic. In fact, there was only one prototypical ‘‘if-then’’ statement in the whole corpus, from a monolingual English (MLE) high-SES ﬁfth grader: ‘‘Thank goodness the pond was there. If it wasn’t, we would be hurting.’’

There was also this statement, which could be translated into a syllogism, from a MLE high-SES second grader: ‘‘They thought maybe the frog’s here because there’s a stream and frogs live near streams.’’

Another child in the same demographic group exhibited a similar sense of logic when he exclaimed: ‘‘[the boy] decides [the swamp] might be the perfect place to look around.’’

One low-SES second-grade girl said: ‘‘This is the right place for Berty [the frog] to be. Let’s check around.’’

And a high-SES bilingual ﬁfth-grader oﬀered this inference: ‘‘thinking that their frog must have had babies.’’

In the whole English corpus, there were only 148 tokens of modals, the words most closely associated with inference. The modals used most were ‘‘could,’’

 Barbara Zurer Pearson

‘‘can,’’ and ‘‘would,’’ as opposed to ‘‘may,’’ ‘‘must,’’ and ‘‘might.’’ There was little diﬀerence in their frequency between second and ﬁfth grade or between monolinguals and bilinguals. With respect to intensional verbs (‘‘think,’’ ‘‘realize,’’ ‘‘wonder,’’ ‘‘know,’’ etc.), there was a diﬀerent pattern among bilinguals and monolinguals. For the monolinguals, there was no age eﬀect (p>.2 by P2, but there was a decided diﬀerence by socio-economic status, with high-SES children using almost three times as many of the 32 instances of these verbs (p<.01 by P2). Among the bilinguals, there was a signiﬁcant increase in talk about thinking from second to ﬁfth grade as well as the high-SES eﬀect noted with monolinguals (21 versus 42, 38 versus 25, respectively, both p<.001 by P2).

False-belief mentioning False-belief mentioning among the groups is shown in Table 1. The false-belief mentioning among the monolingual high-SES children was consistent with the pattern found in Berman and Slobin, (1994: 55, Table 4). That is, about one half of the 11-year-olds in the high-SES monolingual English groups and the 9-year-olds in the monolingual Spanish group explicitly mentioned the boy’s

Table 1. False-belief mentioning by language, language background, age and socioeconomic status Language status

Age

SES

MLE

11

High Low High Low

7 MLS BLE

9* 5* 11 7

BLS

11 7

All

Mentioners

Per cent

9/20 2/22 6/19 4/18

45% 9a 32 22a

12/22 4/21

54 19 a

High Low High Low

2/20 3/20 5/20 2/29

10a 15a 25a 7a

High Low High Low

1/20 3/20 1/20 0/29

5a 15a 5a 0a

55/300

18%

(High) (High)

*From A. Bocaz (1987) and Sebastian (1989), both from the CHILDES database (MacWhinney 1997). Note: superscripts denote percentages signiﬁcantly diﬀerent from expected values by P2, p < .05, and in most cases < .01.

Logic and mind in Spanish–English children’s narratives 

false belief, whereas children in the other demographic groups mentioned the misapprehension signiﬁcantly less often. Comparing to the 50 per cent expected value, derived from the Berman and Slobin data, the high-SES monolingual children 7 or older showed an equivalent percentage of ‘‘mentioning,’’ but all other groups, the 5-year-olds, the low-SES monolinguals and all the bilinguals, were signiﬁcantly lower. Bilinguals mentioned the false belief more often in English than Spanish (12 versus 5 times); only half of the time did the same child report it in both languages. In general, stories, whether told ﬁrst in English or Spanish, were more elaborate in English.

Mental verb results The use of a mental verb was more common than false-belief mentioning, 45 per cent versus 18 per cent of all stories. There was some growth from second to ﬁfth grade, from 41 per cent to 60 per cent (p<.05, by P2), but the SES eﬀect, 56 per cent for high-SES versus 46 per cent for low-SES, did not reach signiﬁcance. The Berman and Slobin ﬁndings (1994: 53–4) indicate 8 per cent for 5-year-olds, but 66 per cent for 9-year-olds and adults in a mid-SES sample; thus, these children used a mental verb slightly less often than might be expected.

Noun-clause results As with false-belief mentioning, not all stories included full-noun-clause complements, reﬂecting their optionality. The use of full-clause noun complements by the children in the diﬀerent demographic categories is shown in Table 2. In English, the percentage of noun-clause use in the monolingual groups was somewhat higher than for false-belief mentioning: 41 per cent nounclause use versus only 26 per cent mentioning, and the SES-eﬀect was less pronounced. In Spanish, there was more tendency for full ﬁnite noun clauses than in English. Seventy percent of the monolingual Spanish children used full noun clauses compared to 41 per cent of the monolingual English-speaking children, (P2=26.2, p <.001). This may be a result of diﬀerences in the syntax of the two languages, as suggested by Gathercole and Montes (1995/1997). For example, in the Spanish ‘‘say’’ clause, the only option for reported speech is a full clause with ‘‘que’’ (‘‘le dijo al perro que se callara’’) [he told the dog that

 Barbara Zurer Pearson

he be quiet], unlike English where the speaker can use an inﬁnitive complement: ‘‘he told the dog to be quiet.’’ The Spanish children could, of course, still use direct speech, as many did, for example, ‘‘Dijo, ‘Silencio.’’’ [He said, ‘‘(Be) quiet.’’’] It is also possible to use inﬁnitive forms in the intensional contexts in Spanish, as in ‘‘sin ellos saber’’ [without their knowing, or literally, without they to know] and ‘‘sin darse cuenta’’ [without taking account]. Unlike in English, these were relatively sophisticated uses and were seen in only two or three stories, all of which also contained other more standard noun-clause locutions. Monolingual English-speaking children showed more use of intermediate forms and more errors in their use than similar aged (or slightly younger) Spanish-learning monolinguals: 29 per cent versus 12 per cent, (P2=4.5, p<.05). Even the bilingual children did signiﬁcantly better with Spanish noun clauses than English ones. Their percentage of correct use (P2=7.11, p<.01) and intermediate or erroneous forms (P2=9.8, p <.01) in each language mirrored the imbalance in favor of Spanish seen in the two monolingual groups. This is all the more surprising because in most other measures the Table 2. Noun-clause usage by language, language background, age and socioeconomic status Language status

Age

MLE

All 11 7

MLS

BLE

All 9* 5* All 11 7

BLS

All 11 7

SES

Noun-clause use

Intermediate forms

High Low High Low

41% 10/20 9/22 9/19 4/18

29% 8/20 5/22 5/19 5/18

70% 21/22 9/21

12% 0/22 5/21

High Low High Low

40% 12/20 7/20 10/20 7/29

37% 6/20 9/20 7/20 11/29

High Low High Low

58% 14/20 11/20 12/20 15/29

17% 4/20 6/20 3/20 2/29

(High)

*From A. Bocaz, (1987) and Sebastian (1989), both from the CHILDES database (MacWhinney 1997).

Logic and mind in Spanish–English children’s narratives 

bilingual children’s English was generally stronger than their Spanish. The disparity may reﬂect a diﬀerence in the distributions of the forms in the target languages and may also be a clue to their relative diﬃculty for learners.

The relation between noun-clause complements and false-belief mentioning Finally, what does our corpus reveal about the relationship between nounclause complements and false-belief mentioning? Since adults, who may be presumed to have the ability to pass false belief tests, mention the false belief in the stories only half of the time, failure to mention false belief in the story cannot be taken as an indication that the narrator cannot do so. In the age group producing the stories, there is little question that almost all, if not all of the children, in a proper experimental circumstance, would be false belief passers. We make no claim that false-belief ‘‘mentioning’’ in the frog story is equivalent to false belief ‘‘passing’’ in the work of de Villiers and her colleagues (or others). Mentioning may, however, index how salient the concepts are and how relatively accessible they are to the narrator. Similarly, noun-clause use within the story is not claimed as a criterion for knowledge of the noun-clause forms. The use of the forms implies noun-clause knowledge, but failure to use noun clauses may imply a lack of motivation rather than a lack of mastery. So, demonstrations from these two aspects of the corpus will not be interpreted as indicators of what the children CAN do, but just what they DID do in this naturalistic, but constrained task of production. Our question is how the child’s knowledge of these two competencies — one cognitive and one linguistic — is reﬂected in their actual performance in the semi-spontaneous situation of telling the frog story. Was a speciﬁc relationship observed between false-belief mentioning and noun-clause use, as de Villiers claims for the early acquisition? The most obvious relationship between them comes from the fact that the most typical way to report a false belief is with a sentence that includes a noun-clause complement: ‘‘He thought that they were branches, but then he learned/found out/ discovered that they were the horns of a deer.’’ However, the noun-clause syntax is not obligatory and we observed several instances where the child reports the false belief using less complex syntax. Especially in a story context, the child can use direct speech instead of an embedded clause to convey the change of the boy’s awareness:

 Barbara Zurer Pearson

‘‘He said, ‘Oh, it’s a deer and not a tree.’ ’’

Other examples use coordination rather than embedding: ‘‘Se aguantó y no eran palos, era(n) un venado, pero él no lo sabía’’ [He grabbed on and they weren’t branches, they were a deer, but he didn’t know it].

Since there are alternative locutions, it is not a necessary relationship between false-belief mentioning and noun-clause complements. Still, it could be the case, as de Villiers suggests, that children would not have ‘‘the mental machinery’’ to handle false belief perceptions and reports until they had mastered noun-clause complements. Our data cannot test that claim directly, but may provide indirect evidence for or against it. In particular, de Villiers’ position would be weakened if one were as likely to ﬁnd false belief reports in the absence of noun-clause complements as noun-clause complements without false belief reports; it would be strengthened if noun-clause use was more common earlier than false-belief mentioning. In fact, Table 3 shows that the latter situation prevailed: noun-clause use without false-belief mentioning was more frequent than the reverse in both languages. There was more noun-clause use than mentioning, so we may hypothesize that noun-clause use is prior to — or at least easier to access than — mentioning. Indeed, noun clauses facilitate the expression of the false belief so it may be that the subjects who can express the false belief without the help of noun clauses are in some other way precocious. By the same token, use of full noun clauses does not guarantee good perception of the events. One child, a lowSES monolingual English second-grader said: ‘‘[The boy] saw that they were branches and then a deer came by.’’ He or she has a prototypical noun clause, but completely misses the point of the boy’s perception. So, noun-clause use is not a suﬃcient condition for making the inference, but noun clauses may be a step toward the inference for most children. The case for a speciﬁc link between noun clauses and theory of mind would be strengthened further if noun-clause use, and NOT other syntactic Table 3. False-belief mentioning and noun clause usage

False-belief/no noun clause Noun clause/no false belief

English

Spanish

Monolingual Bilingual N=79 N=89

Monolingual Bilingual N=43 N=89

2 19

1 16

1 24

0 48

Logic and mind in Spanish–English children’s narratives 

learning, were shown to be associated with false-belief mentioning. We compared the correlation of noun-clause use with the speciﬁc cognitions in the story and also with other scores of story quality to similar correlations done with those same scores and two other more general syntactic measures, a subordination index and a Complex Syntax Score. These other measures included adverbial and adjectival clause tallies as well as noun-clause use. (See Pearson 1997, for a fuller description of the other frog story measures.) All three syntactic measures (noun clauses, subordination index, and Complex Syntax) showed about the same association with the occurrence of the cognitive descriptions in the story: r=.3 to .35 for either the use of a mental verb in the escape scene or the false belief in the deer scene or about .45 with a composite score for either cognition. (These are signiﬁcant, p<.05, but fairly modest.) In the correlations to other measures, noun-clause use correlations were lower and thus appeared more speciﬁc to these cognitive elements. By contrast, the other two syntax scores also correlated highly to other measures from the story unrelated to the cognitive elements: Story Sequencing and Lexicon. Thus, this may be a weak indicator in support of greater speciﬁcity for the noun-clause measure than for the more general syntactic measures. The correlations between references to cognitive states and syntax measures were also done separately for the Miami monolingual and bilingual groups. Values for monolinguals alone and for bilinguals in English were similar to those reported above for the whole group. However, for bilinguals in Spanish, the correlation was very low (r=.08, n.s. between false-belief mentioning and noun-clause use, r=.11, n.s. between false-belief mentioning and the more general subordination index), most likely because there were only ﬁve false-belief mentions in the bilinguals’ Spanish corpus, too few to make an adequate test. For the most part, both false-belief mentioning and noun-clause use were associated with higher levels of language and literacy generally (counting of course measures that do not include either factor as a component). That is, although correlations of false-belief mention or noun-clause use with other language and literacy measures on the children were modest (in the .2 to .4 range), the children who used noun clauses or mentioned the false belief had higher scores on most of the other measures evaluated than children who did not, indicating that their level of language was higher in general. When we divided the corpus according to whether the false belief was mentioned or not, we found the results described in Table 4. The ﬁrst set of measures in the table are derived from diﬀerent aspects of the frog stories

 Barbara Zurer Pearson

Table 4. Average scores on other language and literacy measures according to false-belief mention in the Frog Story (monolingual English and bilinguals, English and Spanish) Measure

False-belief mentioners (N=26)

Non-mentioners (N=135)

Story Measures Number of Clauses Story Sequencing (of 12) Referential Adequacy (of 6) Verb Phrase Elaboration (of 12) Story Lexicon (of 15) Morphosyntactic Accuracy (of 12) Mean Length of Utterance (MLU)

54** 9.0** 5* 10** 9.7** 9.4** 8.2

44** 7.9** 4.5* 8.7** 7.4** 7.9** 7.8

Standardized Language Measures Peabody Picture Vocabulary (receptive) Woodcock Picture Vocabulary (expressive) Woodcock Verbal Analogies Woodcock Reading Comprehension

99** 96* 102* 102

89** 89* 95* 102

Note : p-value from t-test <.05 marked with *; p < .01 marked with **

(Pearson 1996); the others are standardized test scores collected for the parent grant. Mean scores on almost all measures were higher for children who mentioned the false belief, in most cases signiﬁcantly so. Likewise, scores on these other measures were higher for the children with full noun clauses than those with intermediate forms or no noun clauses, as shown in Table 5. Although many qualities beyond logical inferencing ability can contribute to a good frog story, in actual fact, the children who paid close attention to the details of this scene involving the false belief appeared to have consistently higher scores on other aspects of the story as well. It was not just a question of attending to details throughout that made a story good. Many immature stories were painfully detailed picture descriptions, but they failed to provide any interpretation of the story characters and their actions. By contrast, stories that picked out the logical connection between the key pictures involving the deer’s appearance and elaborated on it with embedded sentences were superior in most other aspects as well.

Conclusions and discussion Overall, the frog stories have provided an opportunity for children to demonstrate their skills in these two related cognitive and linguistic domains. They

Logic and mind in Spanish–English children’s narratives 

do not give the whole picture, but they shed an interesting light on it. We have seen that the frog story is not a particularly ‘‘cognitive’’ or ‘‘logical’’ story. Indeed, we have examined a large number of stories and found exceedingly few explicit logical references in them. Even for adults, the particular moments in the picture sequence that invite reference to the characters’ thinking do so only in about 75 per cent of the cases; and only about 50 per cent of the time do adults mention the false belief suggested in the story. For the 11-year-olds in our dataset, about 66 per cent of the children used a mental verb in their description of the events, and 50 per cent mentioned the false belief, compared to 41 per cent and less than 30 per cent, respectively, for the 7-year-olds. There is, thus, some growth in the use of mental verbs and false-belief mentioning in the stories during the age range under investigation. High socioeconomic status was seen to contribute to the child’s reference to thinking in the stories, although the SES eﬀect was not observed among bilinguals. We also observed that noun-clause use rose from 35 per cent at age 7 to 45 per cent at 11 (among the English monolinguals). We can infer that, even though noun clauses are technically ‘‘acquired’’ already, the handling of such Table 5. Average scores on other language and literacy measures according to level of use of noun clauses (NC) in the Frog Story (monolingual English and bilinguals, English and Spanish) Measure Story Measures Number of Clauses Story Sequencing (of 12) Referential Adequacy (of 6) Evaluative Language (of 6) Verb Phrase Elaboration (of 12) Story Lexicon (of 15) Morphosyntactic Accuracy (of 12) Mean Length of Utterance (MLU) Standardized Language Measures Peabody Picture Vocabulary (receptive) Woodcock Picture Vocabulary (expressive) Woodcock Verbal Analogies Woodcock Reading Comprehension

NC Zero (N=46)

NC=1 (N=56)

NC=2 (N=66)

42.5a 7.2a,c 4.2a 1.9a 7.8 a,c 6.7a,b 7.3 7.1a,c

44.2b 7.9a,b 4.6 2.1b 8.6a,b 8.0a 8.3 7.8a,b

51.3a,b 8.7b,c 4.8a 3.0a,b 9.8b,c 8.2b 8.6 8.4b,c

87.1a 85.5a 90.1a,b 101.0

89.0b 89.9 96.9a 103.1

96.6a,b 94.9a 100.9b 104.6

Note : paired superscripts within a row identify values that diﬀer signiﬁcantly from each other by t-test, p < .05.

 Barbara Zurer Pearson

clauses within the story context is undergoing development within this age span. In addition, the presence or absence of noun-clause complements in stories was a powerful indicator of the extent to which the story addressed the characters’ cognitions, although there was no speciﬁc support for it to be considered a precondition. We interpret the timing of these developments as weak support for de Villiers’ position that syntactic growth is leading cognitive growth within this domain. Our ﬁndings are, at least, not inconsistent with that claim. We observed a pronounced cross-linguistic diﬀerence in English and Spanish with respect to the use of full-clause noun complements. There is a suggestion in these data that the monolingual Spanish children demonstrated greater facility with noun-clause syntax than the monolingual English children did. This may be due, perhaps, to its greater regularity in Spanish than in English, and this may have encouraged earlier and more frequent references to the characters’ cognitions in the stories of the Spanish children. The number of stories available and the imperfect matching of the groups being raised in diﬀerent countries limit our ability to make ﬁrm conclusions in this regard, but this may be a proﬁtable avenue for future research. It will also be informative to probe more deeply the ‘‘intermediate’’ noun-clause forms observed in this study to see both the speciﬁcs of their developmental sequencing and their relation to speciﬁc cognitive developments such as those in the frog story. Results for the bilingual children are in the same direction as those for monolinguals. Bilinguals reported the false belief about half as often as the monolinguals did. Children who reported the false belief in one language sometimes reported it in the other as well, although 7 of them reported it in only one language or the other, not in a pattern predicted by whether it was a ﬁrst or second telling for the child. When we compared noun-clause use in each language, both the frequency and accuracy of the bilinguals’s use mirrored the direction of the discrepancy noted between English and Spanish monolinguals. The bilinguals made more errors in noun-clause use in both languages than did monolinguals, but like the monolinguals they used more full noun clauses in Spanish and made more errors in their use in English. There are several possible explanations for why the bilingual children were less likely to produce false belief reports (or mental verb mentions) in either language. It may be that their language facility was somewhat lower in both languages (as was found in standardized testing, Umbel and Cobo-Lewis 1997), and so they might have had fewer linguistic resources available to treat the cognitive aspects of the stories. Or, they may simply have interpreted the task more narrowly as a language task: to demonstrate whether they were able

Logic and mind in Spanish–English children’s narratives 

to talk about the story events in English one day and in Spanish on another day. Further research on this question with bilinguals should use a more speciﬁc test of false belief in addition to the frog story. De Villiers singled out the syntax of noun clauses as particularly predictive of the speciﬁc cognitive developments of Theory of Mind reasoning. Indeed, the syntax of noun clauses is ‘‘specialized’’ for articulating the contents of other minds and maintaining the distinction that those facts and opinions can be diﬀerent from what the speaker knows and thinks. So one can say, ‘‘The boy thinks the antlers are branches, but I can see that they are not’’ and there is no confusion about who thinks what. Beyond the syntactic structures themselves, there are pragmatic and semantic conventions that help speakers communicate factually and evaluatively about what others think. Children who have learned noun-clause syntax still need to learn the conventions that interact with the syntax in governing, for example, which referents are substitutable (de Villiers, Pyers, and Broderick 1997), or when adverbial modiﬁcation changes the scope of the predicate (de Villiers, Curran, DeMunn and Philip 1997). Once they can use noun clauses and pass theory of mind tasks, children are in a position to begin learning more subtle aspects of their grammar, but knowledge of other minds in itself is not a suﬃcient condition for the further developments. Noun-clause syntax may play a similar pivotal, but insuﬃcient, role in children’s narrative development. That is, the mastery of noun-clause syntax gives children a more eﬃcient means to talk about the contents of other minds and to perform logical operations on propositions held by others (de Villiers and Pyers 1997b). Such syntax can alert them to notice a distinction between what ‘‘he said’’ and ‘‘she knows,’’ and it may even hasten their doing so. But in a story context, there are many other demands. Whether the child notices and describes the characters’ cognitive states depends at least in part on how well she can accomplish less speciﬁcally cognitive tasks and how much processing capacity she has left over for the cognitive side of the story. Before the child can think and talk clearly about the story characters’ thoughts, she must ﬁrst be able to create the proper context within which those thoughts will make sense to the listener. She must ﬁrst organize and recount a sequence of events, describe external states, and so forth to accomplish a number of tasks not governed by this particular cognitive advance on her part. Overall, elaborate stories were not necessarily metacognitive, but all stories with more metacognitive references were more complex than those without them. Of the 211 children, only four mentioned the boy’s false belief without

 Barbara Zurer Pearson

using an embedded noun clause to do so, while over half used noun clauses but did not mention the false belief. Thus, for most children, bilingual and monolingual, it appears that the more complex syntax is prior to — or at least easier to access — than the more complex cognitive messages it can convey.

Acknowledgement This research was supported in part by NIH grant RO1HD30762 to D. K. Oller. Parts of this chapter were presented at the Second-Language Research Forum, Michigan State University, 1997.

References Berman, R. A. and D. I. Slobin (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Bocaz, A. (1987). Argentinean Spanish story data: Temporality in discourse. Unpublished manuscript, Universidad de Chile, Santiago, Chile. Bowerman, M. (1979). The acquisition of complex sentences. In: P. Fletcher and M. Garman (eds.), Language acquisition: Studies in ﬁrst language development. Cambridge: Cambridge University Press, 285–306. Christian, D. and C. Maher (1992). Two-way bilingual programs in the United States, 1991–2. Center for Applied Linguistics, Washington, DC; National Center for Research on Cultural Diversity and Second Language Learning, Santa Cruz, CA. 1997 Update: Center for Applied Linguistics: http://www.cal.org/cal/db/2way/ Dunn, L. and L. Dunn (1981). Peabody Picture Vocabulary Test — Revised. Circle Pines, MN: American Guidance Service. Dunn, L., E. Padilla, D. Lugo and L. Dunn (1986). Test de Vocabulario en Imágenes Peabody — Adaptacíon Hispanoameriana [Peabody Picture Vocabulary Test — Latin American adaptation]. Circle Pines, MN: American Guidance Service. Eilers, R. E., D. K. Oller, S. Levine, D. Basinger, M. P. Lynch and R. Urbano (1993). The role of prematurity and socioeconomic status in the onset of canonical babbling in infants. Infant Behavior and Development, 16, 297–315. Eilers, R. E. (ed.). (1997). Language and literacy development in bilingual children. Poster symposium presented at the biennial meeting of the Society for Research in Child Development, Washington, DC. Gale, E., P. de Villiers, J. de Villiers and J. Pyers (1996). Language and theory of mind in oral deaf children. In: A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski (eds.), Proceedings of the 20th annual Boston University Conference on Language Development. Somerville, MA: Cascadilla Press. Gathercole, V. C. (1997). The acquisition of syntactic structures by bilinguals and monolinguals. Poster Symposium presented at the Society for Research in Child Development, Biennial Meeting, Washington, DC.

Logic and mind in Spanish–English children’s narratives 

Gathercole, V. C. and C. Montes (1995/1997). That-trace eﬀects in Spanish and English speaking monolinguals and bilinguals. Paper presented at the Acquisition of Spanish as a First or Second Language conference, Penn State University, State College, PA. In: A. T. Perez-Leroux and W. R. Glass (eds.), Contemporary perspectives on the acquisition of Spanish. Somerville, MA: Cascadilla Press. Hunt, K. W. (1977). Early blooming and late blooming syntactic structures. In: C. Cooper and L. Odell (eds.), Evaluating writing. National Council of Teachers of English, 91–106. Kemper, S. (1984). The development of narrative skills: Explanations and entertainments. In: S. Kuczaj (ed.), Discourse development. NY and Berlin: Springer Verlag, 99–124. MacWhinney, B. (1995). Tools for analyzing talk. Norwood, NJ: LEA. MacWhinney, B. (1997). CHILDES Data. (URL: http://psyscope.psy.cmu.edu). Mayer, M. (1969). Frog, Where are you ? New York: Dial Books. Nelson, K. (1996). Language in cognitive development: Emergence of the mediated mind. Cambridge and New York: CUP. Nippold, M. A. (1988). The literate lexicon. Later language development. Boston: College-Hill Press. Pearson, B. Z. (1996). Assessing narratives from bilingual children: Borrowing a system from gymnastics judging. Paper presented at the Modern Language Association annual meeting, Washington, DC. Pearson, B. Z. (1997). Narrative competence among monolingual and bilingual school children in Miami. Poster Symposium presented at the Society for Research in Child Development, Biennial Meeting, Washington, DC. Perez-Leroux, A. (1996). Mood selection and the acquisition of Spanish relative clauses. Paper presented at the VIIth International Congress for the Study of Child Language, Istanbul, Turkey. Perner, J., S. Leekam and H. Wimmer (1987). Three-year-olds’ diﬃculty with false belief: The case for a conceptual deﬁcit. British Journal of Developmental Psychology, 5, 125–37. Scott, C. (1988). Spoken and written syntax. In: M. Nippold (ed.), Later language development: Ages 9 through 19. San Diego: College-Hill, 49–95. Sebastian, E. (1989). Tiempo y aspecto verbal en el lenguage infantil [Verbal tense and aspect in child language]. Unbilished doctoral dissertation, Universidad Autónoma de Madrid. (Data available through CHILDES database, MacWhinney, 1997.) Sokol, B., U. Muller, et al. (1997). The development of propositional reasoning. Poster presented at the Society for Research in Child Development biennial meeting, Washington DC. Umbel, V. M. and A. B. Cobo-Lewis (1997). Performance by bilingual children in standardized tests of English- and Spanish-language proﬁciency. Poster symposium presented at the Society for Research in Child Development, Washington, DC. de Villiers, J. and J. Pyers (1997a). Complementing cognition: The relationship between language and theory of mind. In: Proceedings of the 21st annual Boston University Conference on Language Development. Somerville, MA: Cascadilla Press. de Villiers, J. and J. Pyers (1997b). On reading minds and predicting action. Poster presented at the Society for Research in Child Development biennial meeting, Washington, DC.

 Barbara Zurer Pearson

de Villiers, J., J. Pyers and K. Broderick (1997). A longitudinal study of the emergence of referential opacity. Paper presented at the 22nd annual Boston University Conference on Language Development, Boston, MA. de Villiers, L. Curran, H. DeMunn and W. Philip (1997). Acquisition of the quantiﬁcational properties of mental predicates. Paper presented at the 22nd annual Boston University Conference on Language Development, Boston, MA. Wimmer, H. and J. Perner (1983). Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children’s understanding of deception. Cognition, 13, 103–28. Woodcock, R. W. (1991). Woodcock Languge Proﬁciency Battery: English Form — Revised. Chicago: Riverside. Woodcock, R. W. and A. F. Muñoz-Sandoval (1995). Woodcock Language Proﬁciency Battery: Spanish Form — Revised. Chicago: Riverside.

Chapter 15

From aﬀect to language Development of evaluation in narratives in spoken English and American Sign Language1 Judy Reilly

Whereas the focus of this volume is primarily concerned with bilingual acquisition and narratives, this chapter departs from the canonical form to explore a speciﬁc aspect of narratives: the development of evaluation across two diﬀerent languages conveyed in two diﬀerent modalities, spoken English and American Sign Language (ASL). In their seminal article in 1967, Labov and Waletsky (1967) ﬁrst introduced the concept of evaluation in narratives. They suggested that narratives included both referential and evaluative functions. From their perspective, referential aspects of narratives include information about the characters and events of the story; it is what moves the story forward, and constitutes the plot. In contrast, the evaluative aspect of narratives gives sense or meaning to the story. According to Labov and Waletzsky, ‘the evaluation of a narrative is deﬁned by us as that part of the narrative which reveals the attitude of the narrator towards the narrative by emphasizing the relative importance of some narrative units as compared to others’ (1967: 37). In this chapter, we use stories from preschoolers and school-aged children to explore how children express evaluative information and how the nature of evaluation and its expression change over time. Our focus is on the transition from the early utilization of emotional expression to convey evaluation by preschoolers to its lexicalization in school age children who primarily convey evaluation linguistically. We begin with a brief discussion of our own work, as well as that of colleagues, on the development of evaluation in spoken English narratives. We then consider aspects of the nature and development of the evaluative function in signed narratives. As noted above, the evaluative elements are those aspects of story telling which convey the signiﬁcance of certain referential events in the narrative; they give the story meaning by transmitting the personal signiﬁcance of

 Judy Reilly

particular aspects of the story. Whereas Labov and Waletzky initially focused on evaluative clauses and noted semantically deﬁned evaluation, Peterson and McCabe (1983) noted that children sprinkled evaluative devices throughout clauses and employ both lexical and phonological means to convey evaluation. Thus, evaluative information can be conveyed or packaged in several ways: •

•

•

Lexically, for example, by using intensiﬁers, modals or hedges to reﬂect speaker attitude, as in the following example, ‘‘He might like chocolate’’, (uncertainty) or ‘‘He really adores chocolate’’, (intensely positive). Syntactically, as in relative clauses, which commonly function as asides to comment on a person’s behavior/character, ‘‘You know, that one who will do anything to win’’; Paralinguistically, by emotional facial expression, gesture and aﬀective prosody which can eﬀectively convey narrator attitude, or reﬂect the inferred emotions of a character, as in, ‘‘John’s gone’’ (distress face and sobbing) or ‘‘John’s gone!’’ (smiling with triumphant voice).

Evaluation is atemporal. Rather than carrying the action forward, as does referential information, evaluation can actually interrupt or suspend the ﬂow of the plotline to convey the narrator’s stance or perspective vis-a-vis the characters or events in the story. When evaluation is linguistically conveyed, it constitutes part of the body of the story proper. However, when evaluation is paralinguistically conveyed, e.g., via facial expression or prosody, it can cooccur with the referential aspects of the story, and the paralinguistic content can be consistent with or conﬂict with the linguistically encoded message, as in sarcasm. Since Labov and Waletzsky’s article in 1967, researchers have considered aspects of evaluation in adult texts (e.g., Labov 1984; Biber and Finnigan 1989) and from a developmental perspective (e.g., Peterson and McCabe 1983; Reilly, Klima and Bellugi 1990; Bamberg and Damrad-Frye 1991; Reilly 1992; Berman and Reilly 1995; Berman 1993, 1997); and the topic has been extensively revisited in the recent tribute to Labov that appeared in 1997 (Special issue edited by Bamberg 1997). Before we consider the nature of evaluation in a signed language, it would be prudent to present a brief introduction to ASL.

American Sign Language Common to the natural sign languages of the world, American Sign Language (ASL) is the visual-gestural language of the Deaf community of the United

Evaluation in spoken English and ASL narratives 

States; it is an independent linguistic system, not derived from any spoken language. Moreover, it exhibits both the grammatical complexity and organizational principles common to spoken languages of the world (Poizner, Klima and Bellugi 1987). For example, unlike English, ASL is morphologically complex and has been compared in typology to polysynthetic spoken languages (Bellugi and Klima 1982). Some aspects of the language are, however, unique to its modality. Not so constrained by the general linearity of the aural/oral modality, signed languages are organized with co-occurring features or ‘‘layers.’’ These are evident at all levels of the grammar: phonology, morphology and syntax. For example, ASL relies on a multi-layered organization in which the verb stem, e.g. look-at, and its aspectual and inﬂectional markers, (e.g., habitual and iterative) are signaled by characteristic movement patterns which all co-occur in space (see Klima and Bellugi 1979, for a discussion). Thus, ASL makes extensive use of co-occurring grammatical devices. Another way in which ASL can be viewed as multi-layered is its use of non-manual behaviors. While the most obvious articulators are the hands, much syntactic and lexical information is encoded on the face (Baker 1983; Baker and Cokely 1980; Baker and Padden 1978; Liddell 1978, 1980; Corina, Bellugi and Reilly 1999) as is evaluative information in discourse (Reilly, McIntire and Anderson 1994; Emmorey and Reilly 1997; Reilly, 2000). Shifts in eye gaze serve both pronominal and discourse functions (e.g., Bahan and Supalla 1995) and changes in body position can mark discourse and syntactic information (Loew, Kegl and Poizner 1997; also see Engberg-Pedersen 1995, 1993 for a discussion of similar behaviors in Danish Sign Language; Rossini, Reilly, Fabbretti and Volterra 1998 for an initial investigation of non-manual behaviors in narratives in Italian Sign Language). In striking contrast to the predominantly linear nature of spoken languages, in signed languages, several channels (eye gaze, hands, faces and torso) cooperate to convey the linguistic signal.

Candidates for conveying evaluation in ASL Given the multi-channeled nature of signed languages, there is a variety of potential means of conveying paralinguistic information: modiﬁcations of sign speed and movement (which might be most similar to vocal prosody); facial expression which serves both linguistic purposes, as noted above, as well the aﬀective/communicative role it plays in both signed and spoken discourse; and ﬁnally, gesture. As a discussion of all these possibilities is beyond the scope of

 Judy Reilly

this chapter, we will consider modiﬁcations of signing speed and movement in the service of evaluation.2

Movement in ASL There is no question that movement of the hands is a critical feature of signed languages. In addition to handshape, location, and orientation, movement is one of the basic parameters of any manual sign and has been given serious consideration in the context of the ‘visual’ phonology of ASL (e.g., Stokoe, Casterline and Croneberg 1965; Coulter 1993; Corina and Sandler 1993; Liddell 1995). Thought and attention has also been devoted to patterns of movement shape and speed as well as pauses that might constitute the elements of linguistic prosody and the suprasegmentals, for example, what might represent phonological stress and intonation in signed languages (for a discussion, see Wilbur 1997; Boyes-Braem 1997; Lillo-Martin 1999). At the morphological level, movement patterns play a critical role, especially with predicates (see Klima and Bellugi 1979, for discussion of verb morphology; Supalla and Newport 1978, for noun verb distinctions). However, little was known regarding the elements of visual aﬀective prosody, that is, how emotion might be conveyed through manual signing. So, we conducted two studies (Reilly, McIntire and Seago 1992) to address the following questions: 1. Do signers show consistency in manual signing style under speciﬁc emotional conditions? 2. To what extent are signers able to interpret these aﬀective diﬀerences in signing?

The production of aﬀective prosody in ASL For the ﬁrst question, we videotaped seven deaf signers signing 12 sentences of neutral content in ﬁve diﬀerent aﬀective states (Happy, Sad, Angry, Surprised, and Neutral). When we measured the length of the sentences for the diﬀerent emotions, the sentences in the negative emotions exhibited distinctive proﬁles in that sad sentences were longer than neutral and angry sentences were signiﬁcantly shorter. To clarify aﬀective diﬀerences, we conducted a frame-byframe analysis of the speed and the shape of the movement path of individual signs. Overall, we found consistency across signers for particular emotions. Again, the two negative emotions (angry and sad) exhibited the most consis-

Evaluation in spoken English and ASL narratives 

tent diﬀerences, and they diﬀer from each other on at least two parameters — duration and the shape of the movement path that the sign follows as can be seen in Figure 1. ANGRY

SAD

NEUTRAL

*

Subject 1 *

Subject 2

*

* * *

Subject 3 *

*

*

*

Subject 4

* *

Figure 1. Movement path for the sign look-for in diﬀerent emotions. *=chin of signer

The perception of aﬀective prosody in ASL A second experiment addressed the interpretation of diﬀerences in aﬀective manual prosody. Twenty-ﬁve deaf adults watched the videotape of a professional actress in the aﬀective prosody production experiment above, and indicated the emotion (forced choice) for each sentence. In one condition, she wore a mask so her face was not visible. All subjects were able to abstract

 Judy Reilly

emotional information from the manual signing alone. Interestingly, errors occurred in both emotional valence and intensity, although more errors cluster along the dimension of intensity. For example, Sad was often confused with Neutral, and Angry and Surprise were more often confused with Happy. Overall, we found subtle, yet consistent perceptible signing diﬀerences in the various aﬀective conditions, and these appear to be comparable with distinctions and patterns identiﬁed in vocal aﬀective prosody of spoken language (Scherer 1986). Thus, modiﬁcation of sign speed, duration and the shape of the movement path are good candidates for conveying evaluative information in sign. In the next section, we look brieﬂy at how evaluation is conveyed in English and ASL narratives by mothers telling stories to their young children.

Aﬀective expression in adult narratives: English and ASL Studies of narratives from English-speaking adults have found that linguistic evaluation (e.g., emotional words, causals, negatives, mental verbs, intensiﬁers) clusters at two points in the story — immediately before the story conﬂict and at its resolution (Bamberg and Damrad-Frye 1991; Bamberg and Reilly 1996). These linguistic evaluative devices thus mark the signiﬁcance of these episodes to the structure of the story. To understand how hearing parents use paralinguistic expression in their stories, we collected narratives in English from 12 hearing adults who are telling Mercer Mayer’s Frog, Where are you? (1969) to their young children (Provine and Reilly 1992). Analysis of their use of paralinguistic expression (e.g., aﬀective prosody and stress) revealed some striking diﬀerences across storytellers. In contrast to the clustered patterning of linguistically encoded evaluation, we found wide individual variation in both frequency and distribution of paralinguistic expression in these stories. Although this may in part be due to the additional attentional function that aﬀective prosody plays, overall, the use and distribution of aﬀective paralinguistic expression by hearing adults appears to be somewhat idiosyncratic. To complement the English data, we have also analyzed narratives from 12 deaf adults telling the same story (Frog, Where are you?) to their children in ASL. Surprisingly, we found that unlike Bamberg’s hearing subjects, the deaf parents did not use lexical evaluative devices to emphasize the conﬂict and resolution of the story. Rather, the deaf adults emphasized the same structural elements of the story as the hearing adults, but through paralinguistic devices, (i.e. aﬀective facial expression, modiﬁed sign location, modiﬁcation of the

Evaluation in spoken English and ASL narratives 

shape or size of the movement path, or of the sign speed). That is, all 12 deaf adult subjects used a signiﬁcantly higher number of paralinguistic evaluative devices during the conﬂict initiation and conﬂict resolution episodes of the story than for other episodes. (Provine and Reilly 1992). In a more detailed analysis of 6 mothers from each group, we found that the deaf mothers also use signiﬁcantly more linguistic devices of cohesion (e.g., anaphora, verb morphology) and coherence (e.g., re-iterating the theme) at these two structural points than their hearing counterparts (McIntire and Reilly 1996). In the ASL narratives, then, adults use paralinguistic evaluative devices and linguistic markers of coherence and cohesion to highlight important junctures in the story. Both deaf and hearing parents use evaluation to signal the same structurally relevant episodes; it is the channels and devices they employ which diﬀer. With this brief overview of evaluation in the adult stories, and the rather striking diﬀerences in the channels recruited to convey evaluation, we are now ready to address how children acquire these narrative skills.

The developmental story of evaluation The lexicalization of evaluation in spoken English Motivated by the broader question of how children integrate emotional and linguistic communication, about 12 years ago, we began to collect ‘frog stories’. We wanted to know not just how children develop narrative skills, but speciﬁcally how they learn ‘‘to tell a good story,’’ because part of telling a good story is that the narrator give it meaning through evaluative devices, and emotion and its expression are deeply intertwined with our personal evaluation. Our focus was on the performance aspects of the narrative (as opposed to the structure of the story) and how children conveyed evaluation. We were particularly interested in the way in which they incorporated aﬀective or emotional expression, both lexically and paralinguistically into their narratives. We used Mercer Mayer’s Frog, Where Are You? to elicit narratives from preschool and school aged English speaking children (Reilly 1992). We coded for both structural coherence as well as types and tokens of the following evaluative devices: a. Characterization or quoted speech: the child speaks for one of the characters, e.g., ‘‘He said, ‘Froggie, come back again’’’; b. Evaluative comments: when the child infers the emotions of the characters using labels for emotional states and behaviors, e.g. ‘‘an’ when he woke up

 Judy Reilly

he was very sad’’ or ‘‘he was crying’’. Other examples include the evaluation of an action or a character (from the point of view of the narrator), e.g., ‘‘He was a nasty owl’’. Mental verbs reﬂecting cognitive processes of characters are also included in this category, e.g. ‘‘He was wondering where that frog had gone’’; c. Facial expressions: e.g. smile, frown; d. Gestures: those that appear to be related to a particular utterance, e.g. covering the head to demonstrate hiding from the (apparently) attacking owl; e. Prosodic features: pitch, length, volume, and voice quality; ‘‘He said, ‘((Fro:ggie, come back again/))’ ((high))’’; f. Lexical/phonological stress: ‘‘He said, ‘Froggie, come back# again’’’. Similar to other studies of narratives from young children, (e.g., Appleby 1978; Bamberg 1987; Peterson and McCabe 1983; Berman and Slobin 1994) we found that even though the three- and four-year-old children had some notion of a story, their narratives were structurally quite thin, and tended to be more interactive than monologic. With respect to evaluation, however, we found that these young children made extensive use of aﬀective prosody, and that stress, vowel lengthening and intonation contours occurred frequently. The end result was a slightly breathless sing-song intonation pattern in which prosodic contours do not follow the adult syntactically motivated shapes. In comparing the stories of the preschoolers to those of seven and eight-yearolds, as expected, we found that the narratives from the older children were structurally much more coherent, and complete; they included all the required 60

3 and 4 years 7 and 8 years

Frequency means

50 40 30 20 10 0 Faces

Gesture Prosody

Stress Eval. com. Char

Figure 2. Paralinguistic evaluation in preschoolers and school aged children

Evaluation in spoken English and ASL narratives 

aspects of the canonical narrative, (initiation of the problem: the disappearance of the frog, the search, the individual search episodes, re-iteration of the search theme, and the resolution: the ﬁnding of the frog). However, in contrast to the narratives from the preschoolers, the older children produced stories that were aﬀectively quite bland and rather stereotypical. We should also note that Berman (1988) had found this same stereotypicality in stories from Hebrew speaking children at about this same age. Overall, as seen in Figure 2, the use of paralinguistic evaluation in these narratives decreased dramatically from preschool to early school age, with the seven- and eight-year-olds using almost none. As noted in the original study (Reilly 1992), the early reliance on aﬀective prosody may serve to help propel the child through the process of telling the story, or…, it may also serve as a type of global genre marking indicating that the child is participating in the very activity of storytelling. Given the structural immaturity of the preschoolers’ stories, we suggest that they are using their well developed aﬀective expressive abilities as a means to ‘‘glue’’ their stories together. (Reilly 1992: 372)

To give a ﬂavor for this striking diﬀerence in narrative quality, in the following examples a preschooler, and then a seven-year-old describe an episode from Frog, Where are you? in the picture book3 (Reilly 1992): E.B. (4;0) Interviewer

he caught# a frog# in there#/ uh huh/ what else?

an they’re look#in at# it# uh huh/

and then (.) he saw# ((something)) the frog# was gone/ (.3) ((low soft)) M. S.(7;0) Once# there was a boy who caught a fro:g/ We:ll#, one night (.) the frog jumped out of the jar (.) and ran away/ When the boy woke up he saw that the frog gotten away/

On listening to the preschool stories, the overall impression is of an excited and slightly breathless child narrating an exciting and somewhat scattered rendition of a story. In contrast, the seven and eight year olds sound more like

 Judy Reilly

they are providing a factual account of the series of events. In this ﬁrst study, we were concerned that the performance of the seven and eight-year-olds might reﬂect school conventions and that the task had been interpreted as informational rather than ‘telling a story’. Therefore, we conducted a second study in which seven and eight year-olds and then ten-and eleven-year-olds ﬁrst told the frog story to an experimenter and then to a three-year-old. In the second telling, they were asked to tell a good story and to make it ‘really interesting’. Whereas two children in the seven/eight year-old group raised the pitch of their voices, overall, their narratives were just as ﬂat as their ﬁrst telling to the experimenter and to the ﬁrst group of children in study 1. However, children in the older age group used multiple devices, both paralinguistic, e.g., aﬀective prosody and proxemics (for example, leaning toward the younger child), and linguistic, including the child’s name, and asking her questions, to actively solicit and maintain the attention of the preschool listener (Reilly 1992). Thus, our original hypothesis was supported: the decrement of aﬀective prosody in the school aged stories was not due just to ‘schooling’ but appeared to reﬂect a broader developmental trend in learning to tell a good story. Concurrently, Bamberg and Damrad-Frye (1991) had been examining the development of lexical evaluation, also using ‘frog stories’. Their framework included: 1. Frames of mind including two categories, emotional labels and mental verbs; 2. Causal signs that explain a character’s motivation; 3. Hedges which suggest that a narrator is making judgments about the likelihood of an event’s occurrence; 4. Negatives which can maintain a storyline by referring to a goal that is not yet attained or convey that an event is contrary to expectations; 5. Character speech, both direct and indirect discourse attributed to characters in the story. These authors found that the ﬁve-year-olds made few references to ‘frames of mind.’ However, there was a clear increase in the amount of lexical evaluation from age ﬁve to nine with a subsequent two-fold increase in the adult stories. In the children’s stories, they found that evaluation tends to reﬂect local events in the narrative, and it was not until the adult stories that evaluation was being used at the global level. Integrating their ﬁndings, Bamberg and Reilly (1996), proposed a transition from paralinguistic to linguistically

Evaluation in spoken English and ASL narratives 

conveyed evaluation that is, a lexicalization of aﬀective expression that appears to occur during the early school years. At this developmental point, lexical evaluation is used primarily at the local level of the narrative. Subsequently, the two communicative channels are re-united: paralinguistics are integrated with lexically conveyed evaluation so that by the end of elementary school, children can not only tell a coherent and well-structured story, they are beginning to tell a ‘good’ story.

The lexicalization of evaluation in ASL If our hypothesis is correct, that paralinguistic expression functions as a support system and stepping stone into the lexicalized expression of evaluation, we should see a similar transition in the developing narrative abilities of Deaf children signing stories in ASL. Our recent studies of aﬀective expression in narratives told by adult Deaf signers have shown that Deaf parents telling stories to their children in ASL typically rely on paralinguistic devices to convey the evaluative aspects of the narrative, speciﬁcally aﬀective prosody and emotional facial expression4 (Provine and Reilly 1992; McIntire and Reilly 1996). These ﬁndings stand in marked contrast to the storytelling styles of hearing parents who rely more heavily on linguistically encoded evaluation, e.g. emotional words, intensiﬁers, frames of mind. Given the diﬀering organization of signed and spoken languages which is exploited and reﬂected in parental narrative styles, we might predict that Deaf children would also rely heavily on paralinguistic devices in their storytelling and would continue to do so during the school age period. However, as we noted above, our studies of English speaking children’s narratives demonstrate that early on, they use a great deal of paralinguistic aﬀect, which decreases over time. Thus, a competing hypothesis stemming from the English children’s narratives predicts that the use of paralinguistic aﬀect in Deaf children’s ASL narratives will decline with age. To determine the developmental interaction of linguistic and emotional expression and the expressive means which children employ to convey evaluation, we analyzed videotaped narratives signed by 25 Deaf children (ages 3;4 to 7;10) whose ﬁrst language is American Sign Language. Each child ﬁrst looked at, and then signed, four short picture stories which were designed to elicit emotional narratives. In each picture story a child encounters a situation which elicits a negative emotion and then the child is helped by either another child or an adult to resolve that situation. For example, in the ﬁrst picture story, a boy falls from his bicycle and a girl helps him back onto it.

 Judy Reilly

Similar to our coding for spoken English, the types of linguistically encoded evaluation included manual signs in the following categories: 1. Frames of mind which included two categories: emotional labels, e.g., angry, excited, lonely, sorry, and mental verbs e.g., realize, worry, idea, dream, think; 2. Causal signs which suggested that the storyteller was making a judgment about the motivation for an event or a character’s behavior, e.g., because, why, how; 3. Hedges, which reﬂect the degree of certainty, e.g., seems, look-like, how; 4. Negative qualiﬁers, e.g., can-not, no, not-yet, none. To code paralinguistic expression in the children’s narratives, two deaf signers independently tallied all signs in their stories that deviated in location, or speed, or size or shape of movement path from ‘normal’ signing. Similar to their hearing peers, Deaf preschoolers made extensive use of paralinguistic expression and included very little lexically encoded evaluation in their narratives. Whereas the number of tokens of lexical evaluation increased dramatically at ﬁve and continued to increase through this age span, the number of types of evaluation leveled oﬀ at age six when children included all four categories of lexical evaluation in their stories. In contrast, the use of paralinguistic expression declined overall, decreasing signiﬁcantly by age seven. The signed stories of the seven-year-olds were as aﬀectively bland as those of their hearing counterparts. To provide an idea of the diﬀerences in prosody of signed stories from three-year-olds and seven-year-olds, below we ﬁrst show a picture of the sign fall and the adult shape of the movement path of fall. Note that only the dominant hand moves and it produces a rather smooth arc (see Figure 3).

Figure 3. Adult sign fall with shape of movement (from Sternberg 1981)

Evaluation in spoken English and ASL narratives

Figure 4 shows frame-by frame tracings of the movement path of the verb fall from two seven-year-olds. In addition to the shape of the movement path, signing location can be judged by the relationship of the movement path to the small circle in the ﬁgures which represents the child’s chin.

RH

RH

Figure 4. Tracing of movement paths of 7-year olds: FALL (‘o’=chin)

In these stories, the children generally used a base hand of an upright B classiﬁer (as the bicycle) and the dominant hand was V classiﬁer. In looking at the movement path of the signs from the seven-year-olds, we can see that the base hand moves very little (similar to the adult model) and that the movement path of the dominant hand is smooth and remains in the signing space. These closely reﬂect the citation form of the adult model, with slight modiﬁcations for discourse context. In Figure 5 we present the same sign from two three-year-olds.

RH

LH

RH

LH

Figure 5. Tracing of movement paths of 3-year olds: fall (‘o’=chin)

Comparing the signing of the three-year-olds, we ﬁrst notice that the movement paths are extremely large and irregular; they extend beyond the



 Judy Reilly

boundaries of normal signing space. Moreover, rather than the dominant hand moving away from the base hand, the preschoolers move both hands eﬀusively. In fact, these renditions of fall defy gravity by terminating in the air. Overall, in the signed stories, we found that similar to their hearing counterparts, Deaf preschoolers use evaluative paralinguistic devices extensively in their stories, and similar to the English speaking children at this age, they include little linguistically encoded evaluation. At age 5–6, both groups of children are including both more types and tokens of linguistic evaluation, and concurrently, the use of paralinguistic expression decreases: the prosody of individual signs and utterances more closely match the adult linguistic model. Thus, if we return to the hypotheses proposed above, Deaf children’s narratives more closely resemble those of age-matched English speaking children than those of Deaf adults with respect to the means by which evaluation is conveyed is their stories.

Conclusions and discussion Our original goal was to chronicle the development of evaluation in children’s narratives by comparing the performance of children acquiring spoken English with Deaf children acquiring a signed language, ASL, as their ﬁrst language. By looking both cross-linguistically and cross-modally, we can tease apart those aspects of performance that are truly developmental from those which are either language or modality speciﬁc. Since adults have recourse to all the rhetorical and narrative choices available in their language, we began by comparing stories told by Deaf adults in ASL and hearing adults in spoken English to their respective children. In both these language communities, adults provide the same structural information for their audiences; however, the channels they recruit directly reﬂect the organization of the two languages and the range of rhetorical choices that signed and spoken languages provide. Hearing mothers telling stories in English use a wealth of linguistically encoded aﬀect as well as other types of lexically encoded evaluation. Although they do employ prosody eﬀectively to draw in the audience, maintain attention and build tension, it is at a signiﬁcantly lower rate than Deaf mothers, and we found wide individual diﬀerences in its use and distribution. In contrast, all Deaf mothers signing to their children make extensive use of both aﬀective prosody and the multi-channeled nature of ASL. Whereas the hands

Evaluation in spoken English and ASL narratives 

convey lexical information, modiﬁcations in sign movement in the service of aﬀective prosody along with other channels (e.g., face, body and eye gaze shifts) assume the major responsibility for the evaluative aspect of narrative. In spite of these two strikingly diﬀerent models in adult story telling, in which English encodes evaluation lexically and ASL uses prosody and non-manual behaviors for evaluative information, Deaf and hearing children reﬂect virtually identical developmental proﬁles in their recruitment of paralinguistic versus lexically encoded evaluative information. We have seen that in both Deaf and hearing children, preschoolers rely heavily upon paralinguistically conveyed aﬀect and use very little lexically encoded evaluation. At about age ﬁve to six, there is a signiﬁcant drop in the use of vocal prosody (in English) and visual prosody (in ASL), as they make a transition to a lexical mode of conveying evaluation. Increasingly diﬀerent types as well as tokens of evaluation emerge over the school years. By age seven, both groups of children use little paralinguistic aﬀective expression, but are instead relying almost entirely on lexically encoded aﬀect and other types of lexical evaluation. Hence, in spite of the diﬀering adult models in the two groups, the developmental proﬁle is the same, regardless of language, and regardless of the modality in which the language is conveyed. In the preschool period, the children’s stories include both emotional and linguistic information, thus recruiting both aﬀective and linguistic systems. The school aged children have incorporated the emotional/aﬀective information into the linguistic system and although their stories are structurally and lexically richer, they are aﬀectively quite ﬂat. We might envisage this transition as a linearization with the linguistic system subsuming emotional expression, and lexicalizing that information. The result is richer content, but a more linear format. As we noted above, by the end of elementary school, the English speaking children are re-integrating paralinguistic aﬀective devices to engage the listener, however the bulk of evaluation continues to be conveyed lexically. The signing children make the same transition to linearized stories as they begin school. Unfortunately, we do not have comparable narratives of 10 and 11 year-old deaf children. However, given the adult model, we know that the deaf children eventually return to a multi-channel strategy to make use of aﬀective prosody and the non-lexical co-occurring linguistic components of the language. In sum, we have a type of U-shaped curve in which paralinguistic aﬀect is extensively used in the preschoolers, minimally recruited at the seven years old, and then again by the hearing children late in elementary school and extensively used by deaf adults.

 Judy Reilly

With our original question in mind, how aﬀective and linguistic expression are related at diﬀerent developmental points, and to understand this U-shaped pattern, we might now look at how aﬀective prosody functions at these diﬀerent points in time. For the preschooler, aﬀective prosody appears to be a global undiﬀerentiated signal, a kind of coarse genre marking indicating that a narrative is being told or some extended discourse is occurring; it is not an integral part of the story. Rather than being tied to any individual components of the narrative, either local or more global structural elements, aﬀective expression appears to supplement their emergent narrative skills. However, for the older children (10–11 years), emotion is integrated into their stories both lexically and paralinguistically. As such, they use both systems in an integrated manner. In contrast to the preschoolers, their use of aﬀective prosody bespeaks a ﬂuency in storytelling in that they can make use of several rhetorical and narrative options. However, from a structural perspective, their use of evaluation is still primarily limited to the local level of the narrative. Finally, considering the deaf adults, we see that paralinguistic expression carries much of the evaluative function of narratives and that it functions at a global level, serving to emphasize structurally signiﬁcant points in the story. In sum, at each developmental point, paralinguistic expression bears a diﬀerent relationship to the lexically conveyed message and plays a diﬀerent role in the narrative itself. What is perhaps most striking, however, is that in spite of the diﬀerences in language, language organization and modality, Deaf children acquiring ASL and hearing children acquiring English reﬂect common strategies and developmental milestones in the integration of linguistic and aﬀective expression.

Notes . This research was partially supported by NIH grants NIDCD-DCO1289–029 and NINDS NS22343. . However, for discussions of facial expression in ASL and the development of grammatical facial behaviors in ASL, see Reilly, McIntire and Bellugi (1991/1996); Reilly and Bellugi (1996); Anderson and Reilly (1997) , Corina, Bellugi and Reilly (1999) and Reilly (2000); and for a discussion of gesture and ASL, see Liddell (1978, 1980) and Emmorey (in press). . Transcription conventions for prosodic elements in spoken narratives include the following: ((double parentheses reﬂect scope, quality is speciﬁed at utterance end)) : = lengthened vowel

Evaluation in spoken English and ASL narratives 

italics = increased volume # = stress on the preceding syllable Intonation or pitch contour is reﬂected by a superscript line above the utterance. . One frequent context for aﬀective facial expression in narratives is in direct discourse or reported action which are sometimes referred to as ‘role shift’. For discussions of these phenomena, see Engberg-Pedersen (1995), Emmorey and Reilly (1997); Reilly (2000).

References Anderson, D. S. and J. S. Reilly (1997). The Puzzle of Negation: How children move from communicative to grammatical negation in ASL, Applied Psycholinguistics, 18, 411–429. Appleby, A. (1978). The child’s concept of story. Chicago: University of Chicago Press. Bahan, B. and S. Supalla (1995). In: K. Emmorey and J. Reilly (eds.), Language, gesture and space. Norwood, NJ: Lawrence Erlbaum, 171–191. Baker, C. (1983). A microanalysis of the non-manual components of questions in American Sign Language. Unpublished Ph.D. Dissertation, University of California, Berkeley. Baker, C. and D. Cokely (1980). American Sign Language: A teacher’s resource text on grammar and culture. Silver Spring, MD: TJ Publishers. Baker, C. and C. Padden (1978). Focusing on the nonmanual components of American Sign Language. In: P. Siple (ed.), Understanding language through Sign Language research. New York: Academic Press, 27–57. Bamberg, M. (1987). The acquisition of narratives. Berlin: Mouton de Gruyter. Bamberg, M. (1997) (ed.). Journal of Narrative and Life History. Special Issue: Oral versions of personal experience: Three decades of narrative analysis, Vol 7, 1–4. Bamberg, M. and R. Damrad-Frye, R. (1991). On the ability to provide evaluative comments: Further explorations of children’s narrative competencies. Journal of Child Language, 18 , 3, 689–710. Bamberg, M. and J. S. Reilly (1996). Emotion, narrative and aﬀect. In: D. I. Slobin, J. Gerhardt, A. Kyratzis and J. Guo (eds.), Social interaction, social context and language. Essays in honor of Susan Ervin-Tripp. Norwood, NJ: Lawrence Erlbaum, 329–342. Bellugi, U. and E. Klima (1982). The acquisition of three morphological systems in the American Sign Language. Papers and reports on child language development, 1–34. Berman, R. A. (1988). On the Ability to relate events in narrative. Discourse Processes, 11, 469–497. Berman, R. A. (1993). The development of language use: Expressing perspectives on a scene. In: E. Dromi (ed.), Language and cognition: A developmental perspective Norwood, NJ: Ablex, 172–201. Berman, R. A. (1997). Narrative competence and storytelling performance: How children tell stories from diﬀerent perspectives and in diﬀerent contexts. Journal of Narrative and Life History, Vol 7, 1–4, 235–244. Berman, R. A. and J. Reilly (1995). Evaluative elements in narratives. Paper presented at the meeting of the Stanford Child Language Conference, Stanford, CA.

 Judy Reilly

Berman, R. A. and D. Slobin (eds.) (1994). Relating events in narrative: A cross-linguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Biber, D. and E. Finegan (1989). Styles of stance in English: Lexical and grammatical marking of evidentiality and aﬀect . Text, 9, 1, 93–124. Boyes-Braem, P. (1997). Three kinds of suprasegmental markers in Swiss-German Sign Language. Talk presented at Prosody and Intonation in Signed and Spoken Languages. Haifa. June, 1997. Corina, D. and W. Sandler (1993). On the nature of phonological structure in sign language. Phonology, 10, 165–207. Corina, D., U. Bellugi and J. Reilly (1999). Aﬀective and grammatical facial expression in Deaf signers. Speech and Language, 42. Coulter, G. R. (1993). Phonetics and Phonology: Current Issues in ASL Phonology. San Diego, CA: Academic Press, INC. Emmorey, K. (in press). Sign Language: A window into human language, cognition, and the brain. Hillsdale, NJ: Lawrence Erlbaum. Emmorey, K. and J. Reilly (1998). The development of quotation and reported action: conveying perspective in ASL. In: E. Clark (ed.), Proceedings of the Stanford Child Language Forum. Stanford, CA: CSLI. Engberg-Pedersen, E. (1993). Space in Danish sign language. SIGNUM-Verlag:Hamburg, Germany Engberg-Pedersen, E. (1995). In: K. Emmorey and J. Reilly (eds.) Language, Gesture and Space. Norwood, NJ: Lawrence Erlbaum. Klima, E. and U. Bellugi (1979). The signs of language. Cambridge, MA: Harvard University Press. Labov, W. (1984). Intensity. In: D. Schiﬀrin (ed.), Meaning, form and use in context: Linguistic applications. Georgetown University Round Table on Languages and Linguistics. Washington DC: Georgetown University Press, 43–70. Labov, W. and J. Waletzky (1967). Narrative analysis: oral versions of personal experience. In: J. Helm (ed.), Essays on the verbal and visual arts. Seattle: University of Washington Press, 12–44. Labov, W. and J. Waletzky (1967). Narrative analysis: Oral versions of personal experience. In: J. Helm (ed.), Essays on the verbal and visual arts. Seattle: University of Washington Press. (Reprinted in Narrative Inquiry in 1997) Liddell, S. (1978). Non-manual signals and relative clauses in American Sign Language. In: P. Siple (ed.), Understanding language through Sign Language research. New York: Academic Press, 59–90. Liddell, S. (1980). American Sign Language syntax. The Hague: Mouton. Liddell, S. (1995). Real, surrogate, and token space: Grammatical consequences in ASL. In: K. Emmorey and J. Reilly (eds.), Language, gesture, and space. Hillsdale, NJ: Lawrence Erlbaum, 19–41. Lillo-Martin, D. (1999). Modality eﬀects and modularity in language acquisition: The acquisition of American Sign Language. In: T. K. Bhatia and W. C. Ritchie (eds.), Handbook of second language acquisition. San Diego, CA: Academic Press, 531–567. Loew, R. C., J. A. Kegl and H. Poizner (1997). Fractionation of the components of role play in a right-hemispheric lesioned signer. Aphasiology, 11, 263–281.

Evaluation in spoken English and ASL narratives 

McIntire, M. L. and J. S. Reilly (1996). Searching for frogs in the narrative stream: a crosslinguistic and cross-modal study of maternal narratives. Journal of Narrative and Life History, 6, 1, 65–86. Peterson, C., and E. McCabe (1983). Developmental psycholinguistics: Three ways of looking at a child’s narrative. New York: Plenum Press. Poizner, H., E. S. Klima and U. Bellugi (1987). What the hands reveal about the brain. Cambridge: MIT Press. Provine, K., and J. Reilly (1992). The expression of aﬀect in signed and spoken stories. Paper presented at the Fourth Annual Meeting of the American Psychological Society, San Diego, CA. Reilly, J. S. (1992). How to tell a good story: The intersection of language and aﬀect in children’s narratives. Journal of Narrative and Life History, 2 , 4, 355–77. Reilly, J. S. (2000). Bringing aﬀective expression into the service of language: Acquiring perspective marking in narratives. In: K. Emmorey and H. Lane (eds.), The Signs of Language Revisited: An Anthology in Honor of Ursula Bellugi and Edward Klima, 415–432. Reilly, J. S. and U. Bellugi (1996). Competition on the face: Motherese in ASL. Journal of Child Language, 23, 219–239. Reilly, J. S. and L. Seibert (in press) . Language and emotion. In: R. Davidson, K. Scherer and H. Goldsmith (eds.), Handbook of Aﬀective Sciences. Academic Press. Reilly, J. S., M. L. McIntire and D. Anderson (1994). Look who’s talking! Point of view and character reference in mothers’ and children’s ASL narratives. Paper presented at the Boston Child Language Conference. Boston, MA. November 1994. Reilly, J. S., M. L. McIntire and U. Bellugi (1990). FACES: The relationship between language and aﬀect. In: V. Volterra and C. Erting (eds.), From gesture to language in hearing and deaf children. Springer-Verlag, 128–141. Reilly, J. S., M. L. McIntire and U. Bellugi (1991). BABYFACE: A new perspective on universals in language acquisition. In: P. Siple (ed.), Theoretical issues in Sign Language Research: Psycholinguistics. Chicago, IL.: University of Chicago Press, 9–23. Reilly, J. S., M. L. McIntire and H. Seago (1992). Aﬀective prosody in American Sign Language. Sign Language Studies, 75, 113–128. Reilly, J. S., E. S. Klima and U. Bellugi (1990). Once more with feeling: Aﬀect and language in children from atypical populations. Development and Psychopatholog , 2, 4, 367–392. Rossini, P., J. Reilly, D. Fabretti and V. Volterra (1998). Non-manual behaviors in Italian Sign Language. Italian Sign Langague Conference, Genoa. Scherer, K. (1986). Vocal aﬀect expression: A review and a model for future research. Psychological Bulletin, 99, 143–165. Stokoe, W., D. Casterline and C. Croneberg (1965). A Dictionary of American Sign Language. Silver Spring, Md.: Linstok Press. Supalla, T., and E. Newport (1978). How many seats in a chair? The derivation of nouns and verbs in American Sign Language. In: P. Siple (ed.), Understanding language through sign language research. New York: Academic Press. Wilbur, R. (1997). Stress in ASL: What we know, how we know it, and typological implications thereof. Paper presented at Prosody and Intonation in Signed and Spoken Languages. Haifa. June 1997.

Chapter 16

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

It is a privilege to contribute a concluding chapter to this volume. In the ﬁrst place, it gives me a chance to extend the history of cooperation I have enjoyed with the two distinguished scholars who edited the book. Besides, the studies collected here are of obvious interest to anyone concerned with learning about language acquisition and narrative development, from the important perspective of situations of ‘‘Languages in Contact’’ (Weinreich 1952). The chapters in this book deal with children from early preschool through adolescence and on to adulthood, they cover some two dozen languages from English, Norwegian, and Spanish to Hebrew, Papiamentu, and American Sign Language. In methodology, they include some longitudinal case-studies, with most chapters relying on larger, cross-sectional investigations. The studies involve adult-child conversational interactions at one extreme and more structured picturebookbased narrative elicitations at the other. And they consider topics ranging from mastery of linguistic structures such as connectivity markers, verb tense, and relative clauses via discourse functions such as reference, temporality, and evaluation snd on to false belief reasoning and command of global narrative structure and discourse organization. Perhaps most importantly, many of these studies go beyond questions of cognitive and linguistic development to address socially relevant aspects of multilingualism. Thus, more than half the chapters in the book devolve around the important demographic changes that have taken place in western Europe over the past few decades, by investigating narrative discourse produced by children of culturally as well as linguistically diverse immigrant backgrounds in Norway, Sweden, Germany, France, and the Netherlands. Further, a number of studies deal with the ‘‘same’’ language in diﬀerent contact situations: Turkish as

 Ruth A. Berman

a home language in Germany, in France, and in the Netherlands; Dutch as the school language in the Netherlands on the one hand, and on the island of Curaçao on the other; Hebrew as the home language in the United States and as the school language in Israel; and South American Spanish as a home language compared with Hebrew in Israel and Central American Spanish compared with English in the United States. These varied situations provide a rich basis for comparing not only how the same languages are acquired in diﬀerent situations of contact, they also make it possible to consider the impact of the relative social status of the home (typically minority) language compared with the school (mainly majority) language. The chapters comprising this volume are thus rich and complex in both the topics which they address and the perspectives which they adopt. In the remarks which follow, I will not attempt to recapitulate major themes or to summarize the results of the research reported in the book. Rather, I consider the signiﬁcance of these studies against the background of ideas which have emerged in related work of my colleagues and myself (several of whom are represented in this volume) in crosslinguistic studies of monolingual ﬁrst language acquisition and developing text production abilities. The ﬁrst of these concerns what is involved in ‘‘narrative development’’. Research on children´s narrative in monolingual contexts suggests that it is helpful to distinguish between the general cognitive underpinnings of narrative text production, on the one hand, and the ability to make appropriate use of linguistic devices in the verbal expression of this ability, on the other. The ﬁrst refers to the internalized narrative schema that underlies the ability to understand, to recall, and to produce diﬀerent types of stories: personal anecodotes and accounts of past experiences, imaginary tales, traditional fables, fairy stories, adventure stories, etc. In western cultures, at least, this involves knowledge of how information is organized in temporal sequence, structured in terms of an initial orientation or setting, followed by the events or episodes which constitute the backbone of the plot, leading up to a resolution, often ending in an explicit coda in the form of a moral or other generalization relating the events to the current or a more universal state of aﬀairs. The second type of knowledge involves the ability to make use of linguistic forms and devices — bound morphology, closed class grammatical items, lexical expressions, and syntactic constructions — to express such narrative functions as making reference (by means of nominal phrases, proper and common nouns, pronouns, and null arguments), expressing temporal relations (by means of grammatical tense/aspect distinctions and by adverbial and subordi-

A cross-linguistic perspective 

nating markers of temporal sequencing), and creating interclausal connectivity through use of varied lexico-syntactic means of clause packaging. The distinction between these two types of knowledge, couched in terms of ‘‘narrative competence’’ compared with ‘‘storytelling performance’’ (Berman 1995a; Reilly 1992) corresponds to two interconnected facets of narrative development: general cognitive underpinnings compared with linguistically speciﬁc knowledge, on the one hand, and the interrelation between topdown, global discourse abilities and bottom-up, more localized linguistic expression, on the other. The studies in this book, undertaken as they were with diﬀerent motivations and from diﬀerent perspectives, provide strong evidence for the importance of such distinctions. Children of bilingual or less educated home backgrounds may display less proﬁcient usage in the morpho-syntactic constructions and lexical devices which they deploy in telling a story. But we can expect them to manifest similar abilities in overall text construction and in distinguishing between narrative and other types of discourse. This is demonstrated by studies which form the background to some of the chapters in this book (most particularly, Akinci 1999; Kupersmitt 1999) as well as by crossgenre and cross-linguistic investigations of monolingual narrative development (Berman, in press; Slobin 1996). Children with varied native languages, from immigrant and established families, of bilingual and monolingual backgrounds, rely on largely similar strategies for global discourse production in the conceptualization, planning, and organization of their narrations. Across the populations investigated in this book, this in fact appears to be the case for children of very diﬀerent linguistic and even socio-cultural backgrounds, even when in strictly linguistic terms, they are more proﬁcient and hence do better at telling a story in one language than in another. A second theme likewise points to shared underpinnings between monolingual, ﬁrst language narrative development and the kind of diﬃculties encountered by non-native learners or children of bilingual backgrounds. Work comparing the discourse productions of Hebrew-English university-level bilinguals engaged in two distinct tasks — writing a university essay and telling a children´s story (Berman 1995b; Berman 1999 respectively) — suggests that language proﬁciency can be characterized along a cline corresponding to diﬀerent levels of ‘‘non-nativeness’’ Analysis of these bilingual texts yielded a four-tiered hierarchy of distance from native-like (and/or adultly mature) norms of grammatical knowledge and language use. At the most basic level are ’’core’’ grammatical elements. These are obligatory features of simple clause structure including: basic word order; markers of case relations between the

 Ruth A. Berman

verb and its associated arguments and adjuncts; and inﬂections for verb tense/mood/aspect and for agreement in number, gender, and/or person. They are typically acquired in that order (word order, case marking, grammatical inﬂections), and they are mastered in L1 acquisition by around age three years. A second level of diﬃculty for language learners lies in the domain of lexical selection and concerns the degree of speciﬁcity and variety of the vocabulary used by speaker-writers. This kind of knowledge, which involves the precise and appropriate encoding of semantic categories and distinctions, constitutes a later development in L1, since it typically requires a level of vocabulary somewhat beyond the preliterate period of preschool development. The third level concerns rhetorical expressiveness, that is, selection and alternation of appropriate options for expressing a range of discourse functions. This involves making well-formed and discursively appropriate choices out of all the structural and lexical devices available in the target language in areas such as: connectivity in marking interclausal relations and syntactic packaging; perspective in selecting agent versus patient focus through shifts in voice and valency; and deployment of tense-aspect shifting to alternate between background information and the sequentially ordered foregrounded elements of narrative discourse. In L1, this type of ability may not consolidate until later schoolage, since it requires highly developed cognitive underpinnings in coordinating diﬀerent facets of text production: preplanning of what is going to be said, online processing of monologic verbal output, and selection of appropriate linguistic devices for expressing relevant thematic content. Finally, what might be considered the highest level of language proﬁciency entails register appropriateness, that is, the ability to vary language usage skillfully by observing cultural norms and genre conventions in according with diﬀerent social settings and discourse conditions. This requires that speakers (and writers) be able to select and maintain a consistent level of colloquialness or formality in both grammar and the lexicon, and that they can alternate their discourse to suit the particular communicative context: telling a story to a young child, relating an anecdote to a friend, or providing a narrative illustration in an academic lecture in speech, or a personal letter compared with a journalistic report or a formal account relating a particular event in writing. This level of expression develops latest of all (and in some instances may never be achieved), since it depends on a high standard of literacy and considerable experience with diﬀerent kinds of texts, both narrative and non-narrative. These four levels of language proﬁciency constitute a continuum, so that there are no clear cut-oﬀ points between one category and another. For

A cross-linguistic perspective 

example, errors in the use of modal verbs in English, of grammaticized aspectual distinctions in Spanish, or of derivational verb-pattern morphology in Hebrew may be indicative of grammatical and/or lexical and/or rhetorical diﬃculties. That is, these diﬀerent levels of linguistic knowledge are typically interlinked in actual language use, while there is a powerful interaction between lexicon and syntax in describing events or presenting a point of view. Moreover, in psycholinguistic terms, diﬀerent levels on the hierarchy correlate with degree of accessibility to monitoring by speakers and to analysis by researchers, in the following sense. Any layman will detect deviations from core-grammar morphology and syntax, and proﬁcient bilinguals will be able to self-correct such errors if required. Lexical violations are harder both to identify and to correct, while rhetorical infelicities and inconsistencies of register may be totally opaque and inaccessible to analysis by speakers and, perhaps, by researchers too. These ‘‘high-level’’ deviations are the most resistant both to introspection and to change, and they are rarely if ever eradicated. That is, the proposed continuum reﬂects a hierarchy which is intended to capture several interrelated facets of non-native usage: extent of deviance from native norms; accessibility to monitoring and to linguistic description; and resistance to change or eradication. This line of thinking is borne out by the results of many of the studies in the present volume, and it suggests that what is at issue here is acquiring not only linguistic structures, but also language usage in a range of contexts. Second language pedagogy and research alike have traditionally paid considerable attention to the order in which linguistic systems and subsystems are presented to or acquired by learners. But there has been little concern for learner progression in diﬀerent communicative contexts and in diﬀerent discourse genres. This book thus constitutes a milestone in multilingual studies by going beyond isolated linguistic structures and interactive conversational discourse to both co-constructed and monologic narrative text production. As such, it aﬀords rich potential for further directions in crosslinguistic and multilingual research. Developments in monolingual ﬁrst language discourse abilities may also help in characterizing the range of discourse contexts in which bilinguals are expected or required to function in their ‘‘other’’ language. In their chapter on ‘‘Sociocultural aspects of bilingual narrative development’’, Boyd and Nauclér cite Anward’s (1983) characterization of a hierarchy of discourse activity types ranged along increasing degrees of speaker autonomy, from directed dialogue (e.g., adult-child conversations, classroom interactions) to free dialogue

 Ruth A. Berman

(conversations between peers), thence to directed monologue (e.g., meetings, interviews), and on to free monologue (speeches, radio programs, written texts). This proposal is strongly supported by research on narrative development among monolingual children. Children’s earliest narrative productions have their antecedents in interactive conversations, richly scaﬀolded by adult caretaker input (Eisenberg 1985; Hudson 1993; Miller and Sperry 1988; Sachs 1983; Veneziano and Sinclair 1995). And even when required by a particular research paradigm to ‘‘tell a story’’, young preschoolers tend to treat the task as interactive and to require considerable scaﬀolding and input from the experimenter. On the other hand, the ‘‘directed dialogue’’ of second language classrooms, like interviews with immigrants or members of a minority group by representatives of linguistically, socially, or institutionally more privileged groupa, might in fact fail to promote the interactive and communicatively motivated type of narrative discourse typical of parent-child conversations in western cultures. And the type of ‘‘free dialogue’’ which helps in fostering skilled narrative productions in young preschool children in nursery school settings Nicolopolou (1996) diﬀers markedly from typical second language classroom situations. Relatedly, in contrast to interactive, communicatively motivated contexts for narrative production, more structured tasks such as elicitations based on picture-series or picturebooks of the kind favored for both clinical and research purposes yield well-structured narratives only at relatively late preschool age or beyond (Berman and Slobin 1994; Hickmann 1991; Katzenberger 1994). This suggests that, as noted by several chapters in this volume, too, we need to consider a hierarchy of diﬃculties within the apparently unitary notion of ‘‘narrative abilities’’, ranging from appropriately non-threatening conversational interactions to less contextualized, more autonomous settings of text construction. In fact, research into the development of monolingual text production abilities in diﬀerent languages shows that while grade-school children are proﬁcient storytellers in both speech and writing, it takes until later, high school age for youngsters to achieve two important aspects of literacy-based activities (Aisenman 1999; Aparici, et al., 2000; Berman, 2000; Strömqvist, 2000; Tolchinsky, 2000). The ﬁrst concerns construction of expository type discourse, where subjects were asked to discuss ideas and talk about abstract topics of social relevance such as violence in schools and other situations of interpersonal conﬂict. The ability to produce expository, non-narrative prose such as giving talks in front of the class or writing essays expressing one’s ideas on or knowledge about a given topic is essential for school success. Yet ﬂuent

A cross-linguistic perspective 

and well-formed construction of such kinds of texts constitutes a major cognitive burden even for proﬁcient native speakers well into adolescence (Katzenberger, in press). The second diﬃculty encountered by younger schoolage subjects in this crosslinguistic study was in adopting the norms and registers of language usage appropriate to the conventions of written language in their cultures. It takes through to as late as high school age for nativespeaking children of well-educated, middle-class backgrounds to achieve proﬁciency in producing written language not only as a notational system, but as a particular style of language usage (Gayraud, 2000; Ravid and Tolchinsky, 2000; Strömqvist, Ahlsén and Wengelin 1999; Strömqvist, Nordqvist and Wengelin, in press). These ﬁndings for ﬁrst language discourse development and text construction abilities across diﬀerent languages and in diﬀerent contexts point to relevant orientations for research and practice in multilingual situations. As noted, an analogous kind of progression might usefully be adopted, proceeding from from highly scaﬀolded interactional situations, with rich linguistic and thematic structuring of both input and output, to freer, more monologic storytellings, from personal experience accounts to picturebook recountings. Young schoolchildren in multilingual, multicultural situations cannot be expected to perform well in contexts which appear to lie beyond the capacities of their monolingual peers from literate, well-educated backgrounds: producing well-formed expository texts, on the one hand, and deploying written language as a specialized style of expression, on the other. Yet these situations could constitute valuable directions for bilingual research beyond the earlier levels of basic proﬁciency to the more academic usage necessary for school success. Crosslinguistic research on monolingual narrative development using the ‘‘frogstory’’ booklet has provided a basis for methodologically controlled and carefully comparable studies in multingual contexts, as reported in many of the chapters in this volume. The way now lies open for extending crosslinguistic research designed for eliciting both narrative and non-narrative discourse, in both speech and writing, to varied contexts of ‘‘languages in contact’’. This in turn could contribute to ensuring that the populations studied in this volume are empowered to achieve high levels not only of conversational and narrative linguistic and communicative proﬁciency, but of literate and academic language use as well. My remarks so far have addressed a number of distinctions emerging from studies in this volume combined with research on monolingual language acquisition, which appear relevant to ‘‘narrative development in a multilingual

 Ruth A. Berman

context’’. One is the assumption that children from diﬀerent linguistic and socio-cultural backgrounds will share similar cognitive, conceptual, and developmental abilities underlying narrative competence, although they will diﬀer in command of linguistic expression required for storytelling performance. Another relates to diﬀerent levels of linguistic proﬁciency, proceeding from core grammatical abilities achieved by early preschool age on to discourseappropriate use of rhetorical devices and socio-culturally sensitive register distinctions. The latter are mastered at much later phases of development, and they tend to typify highly proﬁcient, mature native speaker-writers. A third distinction concerns the relative cognitive load as well as linguistic diﬃculty entailed by diﬀerent contexts of use and types of discourse, with appropriately scaﬀolded interactive conversation at one end of the continuum, followed by personal experience narration and on to the highly complex and demanding task of abstract, expository text production. Current research further points to the need to distinguish between oral language skills and mastery of written language as a special type of rhetoric required for monologic text construction in academic and other more formal contexts. A further distinction which seems both extremely important and very hard to characterize is that between socio-cultural predispositions and more strictly linguistic types of knowledge and norms of usage. As noted in the introduction, this volume makes an important contribution by considering contact situations between speakers of typologically very diﬀerent languages. From a quite diﬀerent perspective, anthropologically motivated studies focus on the content of narratives produced by children from diﬀerent backgrounds (e.g., Ben Ezer 1992; de Fina, 2000; John-Steiner and Panofsky 1992). Between these two extremes are studies which suggest that what constitutes a good story from the point of view of narrative discourse structure, and not only of content, may diﬀer from one culture to another. An example of this approach is a study of story re-tellings in English as a common language of instruction written by gradeschool children in the United States compared with in Papua New Guinea (Invernezzi and Abouzeid 1995). Other as yet largely unpublished studies concern diﬀerent cultural norms of narrative discourse structuring among indigenous populations in Central America and in the Australian bush (Strömqvist and Verhoeven, in press). The work contained in this volume dealing with narrative development in multilingual contexts constitutes an important point of reference for research aimed at teasing apart the relative contribution of linguistic abilities and cultural conceptions to children´s narrative development in varied situations of ‘‘languages in contact’’.

A cross-linguistic perspective 

References Aisenman, R. (ed.). (1999, February). Developing literacy across genres, modalities, and languages. Working Papers, Volume I. Tel Aviv University: International Literacy Project. Anward, J. (1983) Språkbruk och språkutveckling i skolan. Lund: Liber. Akinci, M-A. (1999). Développement des compétences narratives des enfants bilingues turcfrançais en France àges de 5 à 10 ans. Thèse de doctorate nouveau régime en sciences du language, Université Lumière, Lyon 2. Aparici, M., N. Argerich, J. Perera, E. Rosado, and L. Tolchinsky (eds.) (2000). Working Papers, Volume III. University of Barcelona: International Literacy Project. Ben-Ezer, G. (1992). As light in a clay pot: Migration and absorption of Ethiopian Jews in Israel. Jerusalem: Ruben Mas Publishers [in Hebrew]. Berman, R. A. (1995a). Narrative competence and storytelling performance: How children tell stories from diﬀerent perspectives and in diﬀerent contexts. Journal of Narrative and Life History, 5, 4, 285–313. Berman, R. A. (1995b). Aspects of Hebrew/English contrastive rhetoric. In: H. Shyldkort and L. Kupferman (eds.), Tendances Récentes en Linguistique Francaise et Générales. Amsterdam: Benjamins, 41–54. Berman, R. A. (1999). Bilingual proﬁciency/proﬁcient bilingualism: Insights from HebrewEnglish narratives. In: G. Extra and L. Verhoeven (eds.), Bilingualism and migration. Berlin: Mouton de Gruyter, 187–208. Berman, R. A. (2000). Developing literacy in diﬀerent contexts and in diﬀerent languages. Final Report submitted to the Spencer Foundation, Chicago. Berman, R. A. (in press). The role of context in developing narrative abilities: the frogstory ﬁndings in light of other narrative genres and elicitation settings. In: Strömqvist and Verhoeven (eds.). Berman, R. A. and D. I. Slobin (eds.) (1994). Relating events in narrative: A crosslinguistic developmental study. Hillsdale, NJ: Lawrence Erlbaum. Eisenberg, A. (1985). Learning to describe past experiences in conversation. Discourse Processes, 8, 177–204. de Fina, A. (2000). Orientation in immigrant narratives: the role of ethnicity in the identiﬁcation of narratives. Discourse Studies, 2, 131–157. Gayraud, F. (2000). Développement de la diﬀérentiation oral/écrit vu a travers le lexique, Unpublished doctoral dissertation, Université Lumière, Lyons 2. Hickmann, M. (1991). The development of discourse cohesion: Functional and crosslinguistic issues. In: G. Pieraut-Le Bonniec and M. Dolitsky (eds.), Language bases … Discourse bases: Aspects of contemporary French language psycholinguistics research. Amsterdam: John Benjamins, 158–185. Hudson, J. A. (1993). Reminiscing with mothers and others: Autobiographical memory in young 2-year-olds. Journal of Narrative and Life History, 3, 1–31. Invernizzi, M. A. and M. P. Abouzeid (1995). One story map does not ﬁt all: A crosscultural analysis of children’s written story retellings. Journal Narrative and Life History, 5, 1–19.

 Ruth A. Berman

John-Steiner, V. and C. Panofsky (1992). Narrative competence: cross-cultural comparisons. Journal of Narrative and Life History, 2, 219–233. Katzenberger, I. (1994). Cognitive and linguistic abilities in picture-series narration. Tel Aviv University unpublished doctoral dissertation. [in Hebrew] Katzenberger, I. (in press). Expository text construction. Discourse Studies. Kupersmitt, J. (1999). Development of form/function relations in the narratives of SpanishHebrew bilinguals: What stays and what goes? Tel Aviv University masters’ thesis. Miller, P. J. and L. L. Sperry (1988). Early talk about the past: Origins of conversational stories of personal experience. Journal of Child Language, 15, 293–315. Nicolopolou, A. (1996). Narrative development in social context. In: D. I. Slobin, J. Gerhardt, A. Kyratzis, and J. Guo (eds.), Social interaction, social context, and language. Hillsdale, NJ: Lawrence Erlbaum. Ravid and Tolchinsky (2000). Developing linguistic literacy: A comprehensive model. In: Aparici, et al., (eds.), Working Papers, Volume III, 1–30. [Extended version accepted for publication] Reilly, J. S. (1992). How to tell a good story: The intersection of language and aﬀect in children´s narratives. Journal of Narrative and Life History, 2, 355–377. Sachs, J. (1983). Talking about the there and then: The emergence of displaced reference in parent-child discourse. In: K. Nelson (ed.), Children’s Language, Vol 4, Gardner Press. 1–48. Slobin, D. I. (1996). From ‘‘thought and language’’ to ‘‘thinking for speaking’’. In: J. Gumperz and S. Levinson (eds.), Rethinking linguistic relativity. Cambridge University Press, 70–96. Strömqvist, S. (2000). Panel Convenor: Cognitive processes in text writing: Crosslinguistic and crossmodal perspectives. EARLI Conference on Writing, Verona. Strömqvist, S., E. Ahlsén and Å. Wengelin (1999). The production process in speech and writing. In: R. Aisenman (ed.), Working Papers, Volume 1, 1–20. Strömqvist, S., Å. Nordqvist and Å. Wengelin (in press). Writing the Frog story: developmental and cross-modal perspectives. In: Strömqvist and Verhoeven (eds.). Strömqvist, S., and L. Verhoeven (eds.) (in press). Relating events in narrative: variation across languages, cultures, and genres. Mahwah, NJ: Lawrence Erlbaum. Tolchinsky, L. (2000). Panel Convenor: research on the development of text production in crosslinguistic perspective. EARLI Conference on Writing, Verona. Veneziano, E. and H. Sinclair (1995). Functional changes in early child language: Appearance of references to the past and of explanations. Journal of Child Language, 22, 557–582. Weinreich, U. (1952). Languages in Contact. The Hague: Mouton.

Index age related features 107–24 anaphora 3, 5, 53, 60–79, 190, 234 aspect see temporality bilingual development 1, 2, 277, 310 balanced 87, 352, 361 immersion 322 sequential 153 simultaneous 16 unbalanced 88 CHILDES archive 137, 258, 380 clause 20, 97, 285, 289, 300, 302, 361, 374, 377, 383 combining 97 connectors 97, 359 coordinated clause structures 287 deictic markers 360 noun clauses 387, 389, 391, 393 null-subjects 280 relative clauses 280, 288, 302–3, 309 co-construction 129, 153, 182 symmetric vs asymmetric 144 cognition 375, 377, 389 cognitive development 319, 373 cognitive underpinnings 420, 421 metacognitive knowledge 379 conversation 155–6 strategies 158 cross linguistic research 3, 130, 288, 425 comparison 212, 324, 352, 369, 412 discourse devices 5 dialogue 132 monologue 132 planned vs unplanned discourse 56 elicitation procedures 56, 311 error analysis 192, 202, 303–8, 394 developmental errors 281, 304, 306 false beliefs 378–9, 382, 386, 389 fossilization 281, 304 gender 193, 202, 361–2 grammatical errors 304, 310 interference 193 standard deviations 307 evaluation 399, 404 ﬁltering principle 4 ﬁrst language acquisition 1

frog story 4, 15, 22, 129–49, 156, 192, 212, 235, 258, 260, 273, 277–8, 282, 291, 302, 323, 343, 345–50, 364, 373, 375–6, 381, 392, 393, 405, 407–8, 425 key components 22 interaction strategies 147–9, 185 interlocutor 154, 160, 184 language: attrition 320, 326, 334, 337 code mixing 328 cultural loans 181 dominance 17, 27, 158, 320, 344, 366 input 88 interlanguage 168, 358 majority 420 minority 17, 420 mixing 161–5 morphology 168 proﬁciency 257, 421–2 socialization 131 source 51 structure 87 target language 51 transfer 1, 30, 45, 64, 88, 234, 281, 304–5, 364–6 languages: ASL 399–417 Creole 256 Dutch 209–31, 233–54, 255–75 English 15–50, 319–40, 341–71, 399–417 Finnish 87–128 French 189–208 German 153–88 Hebrew 277–317, 341–71, 373–98 Moroccan Arabic 233–54 Norwegian 15–50, 51–85 Papiamento 255–75 Spanish 277–317, 341–71, 373–98 Swedish 87–128, 129–51 Turkish 51–85, 129–51, 153–88, 189–208, 209–31 lexical development 102 evaluation 400, 405, 409 lexical gaps 165, 333 non-standard forms 167–80, 183, 185 semantic overextension 102 linguistic development 319, 373 devices 5, 16, 285, 292 constraints 326

 Index

linguistic development (cont.) paralinguistic evaluation 400, 405, 407, 409–12 socialization 320, 321 structures 196, 197, 277, 331–2, 337, 423 compounds 332–3 coordination 196–7, 273 state of mind 196–7 structural variations 198–201 subordination 196–7, 273 linguistic threshold 124 literacy orientation 145 logic 379, 385 metalinguistic comment 157 mind 373 narrative structure 26, 91–102, 278, 344, 364 coherence 189 cohesion 189, 274 components 193–5 connectives 189 episodic structure 190 experience 204 organisation 210, 284, 290 relative structure 190 sequential markers 26 story types 56 narrative development 2, 319 home 134 (non)literate cultures 135 production 257 school 134 text length 289 overgeneralization 125, 247–8 parent input 18 one person, one language strategy 343 pronominalization 52 reference to actions 329 reference to entities 329 referential devices 52 referent introduction 57 referent maintenance 53, 57, 60–2, 66, 76, 267–9 referent switch 57, 69, 270–2 zero-marking 52, 76 propositions 95 regions: Curaçao 255–75 France 189–208 Germany (Berlin) 153–88 Israel 277–317, 341–71, 373–98 Morocco 233–54

Netherlands 209–31, 233–54 Norway 15–50, 51–85 Sweden 87–128, 129–51 United States 319–40, 341–71, 399– 417 scaﬀolding 156, 181 sign Language 399–417 manual signs 410 movement 402, 411 prosody 402–4, 412–14 socialization: sociolinguistic context 319, 336 storytelling strategies 18, 26 compensatory strategies 319 performance 421 rhetorical devices 292 story grammar 189 story quality 391 structure 20 subject ellipsis 287, 288–9, 309, 361–3, 367 syntax: syntactic devices 259, 263, 379 syntactic measures 391 temporality 3, 5, 350–8, 366 anchor time 209, 213–14, 235, 238–9, 252, 293–9, 335 conjunctions 35, 250, 259 connectors 31 contrast 31 diverging temporal systems 209 internal temporal relations 3 natural order 233 sequentiality 230, 360 simultaneity 215–20 subordination 35, 178, 360 temporal adverbs 243–6, 253 temporal framework 211 temporal relations 3, 15, 209 tense/aspect 3, 30, 105, 236–7, 240–2, 253, 280, 286, 292, 335 tense preference 352 tense shifts 212, 252–5 time of situation 222, 229 time of utterance 229 topic time 222, 229 tense see temporality topic continuity 3, 5, 260, 265–7, 288 trilingualism 341–69 verbs: converbs 222–4 mental verbs 387 nuclear verbs 102, 104 verb inﬂections 116–24

In the series STUDIES IN BILINGUALISM (SiBil) ISSN 0298-1533 the following titles have been published thus far or are scheduled for publication: 1. FASE, Willem, Koen JASPAERT and Sjaak KROON (eds): Maintenance and Loss of Minority Languages. 1992. 2. BOT, Kees de, Ralph B. GINSBERG and Claire KRAMSCH (eds): Foreign Language Research in Cross-Cultural Perspective. 1991. 3. DÖPKE, Susanne: One Parent - One Language. An interactional approach. 1992. 4. PAULSTON, Christina Bratt: Linguistic Minorities in Multilingual Settings. Implications for language policies.1994. 5. KLEIN, Wolfgang and Clive PERDUE: Utterance Structure. Developing grammars again. 6. SCHREUDER, Robert and Bert WELTENS (eds): The Bilingual Lexicon. 1993. 7. DIETRICH, Rainer, Wolfgang KLEIN and Colette NOYAU: The Acquisition of Temporality in a Second Language. 1995. 8. DAVIS, Kathryn Anne: Language Planning in Multilingual Contexts. Policies, communities, and schools in Luxembourg. Amsterdam/Philadelphia, 1994. 9. FREED, Barbara F. (ed.) Second Language Acquisition in a Study Abroad Context. 1995. 10. BAYLEY, Robert and Dennis R. PRESTON (eds): Second Language Acquisition and Linguistic Variation. 1996. 11. BECKER, Angelika and Mary CARROLL: The Acquisition of Spatial Relations in a Second Language. 1997. 12. HALMARI, Helena: Government and Codeswitching. Explaining American Finnish. 1997. 13. HOLLOWAY, Charles E.: Dialect Death. The case of Brule Spanish. 1997. 14. YOUNG, Richard and Agnes WEIYUN HE (eds): Talking and Testing. Discourse approaches to the assessment of oral proficiency. 1998. 15. PIENEMANN, Manfred: Language Processing and Second Language Development. Processability theory. 1998. 16. HUEBNER, Thom and Kathryn A. DAVIS (eds.): Sociopolitical Perspectives on Language Policy and Planning in the USA. 1999. 17. ELLIS, Rod: Learning a Second Language through Interaction. 1999. 18. PARADIS, Michel: Neurolinguistic Aspects of Bilingualism. n.y.p. 19. AMARA, Muhammad Hasan: Politics and Sociolinguistic Reflexes. Palestinian border villages. 1999. 20. POULISSE, Nanda: Slips of the Tongue. Speech errors in first and second language production. 1999 21. DÖPKE, Susanne (ed.): Cross-Linguistic Structures in Simultaneous Bilingualism. 2000 22. SALABERRY, M. Rafael: The Development of Past Tense Morphology in L2 Spanish. 2000. 23. VERHOEVEN, Ludo and Sven STRÖMQVIST (eds.): Narrative Development in a Multilingual Context. 2001. 24. SCHMID, Monika S.: First Language Attrition, Use and Maintenance. The case of German Jews in anglophone countries. n.y.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)

Managing Development in a Global Context

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)

Beyond Bilingualism: Multilingualism and Multilingual Education (Multilingual Matters)

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

Narrative and Genre (Routledge Studies in Memory and Narrative)

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)

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)

Managing Development in a Global Context

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)

Beyond Bilingualism: Multilingualism and Multilingual Education (Multilingual Matters)

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

Narrative and Genre (Routledge Studies in Memory and Narrative)

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)

Recommend Documents