Journal of Child Language http://journals.cambridge.org/JCL Additional services for Journal

of Child

Language: Email alerts: Click here Subscriptions: Click here Commercial reprints: Click here Terms of use : Click here

Can bilingual two-year-olds code-switch? Elizabeth Lanza Journal of Child Language / Volume 19 / Issue 03 / October 1992, pp 633 - 658 DOI: 10.1017/S0305000900011600, Published online: 17 February 2009

Link to this article: http://journals.cambridge.org/abstract_S0305000900011600 How to cite this article: Elizabeth Lanza (1992). Can bilingual two-year-olds code-switch?. Journal of Child Language, 19, pp 633-658 doi:10.1017/S0305000900011600 Request Permissions : Click here

Downloaded from http://journals.cambridge.org/JCL, IP address: 138.251.14.35 on 12 May 2015

J. Child Lang. 19 (1992), 633-658. Printed in Great Britain

Can bilingual two-year-olds code-switch?* ELIZABETH LANZA University of Oslo (Received 10 October 1990, Revised 10 September 1991)

ABSTRACT

Sociolinguists have investigated language mixing as code-switching in the speech of bilingual children three years old and older. Language mixing by bilingual two-year-olds, however, has generally been interpreted in the child language literature as a sign of the child's lack of language differentiation. The present study applies perspectives from sociolinguistics to investigate the language mixing of a bilingual twoyear-old acquiring Norwegian and English simultaneously in Norway. Monthly recordings of the child's spontaneous speech in interactions with her parents were made from the age of 2; o to 2; 7. An investigation into the formal aspects of the child's mixing and the context of the mixing reveals that she does differentiate her language use in contextually sensitive ways, hence that she can code-switch. This investigation stresses the need to examine more carefully the roles of dominance and context in the language mixing of young bilingual children.

INTRODUCTION Code-switching, the alternation or mixing of languages within discourse, is well-documented in the sociolinguistic literature on bilingualism. Work in bilingual communities has shown that mixing languages can be a communicative resource in bilingual interactions, revealing the bilingual speaker's sensitivity to both formal and functional aspects of language use (e.g. Poplack, 1981; Grosjean, 1982; Gumperz, 1982; Heller, 1988; Romaine, 1989). Whereas a vast literature has evolved on the code-switching abilities of adults, fewer studies have addressed this issue in children. Those [*] The research for this article was financially supported by the Norwegian Research Council for Science and the Humanities. An earlier version was presented at the Fifth International Congress for the Study of Child Language, Budapest, July 1990. I especially wish to thank Naomi Goodz, Natela Imedadze, Bard Bredrup Knudsen, Hanne Gram Simonsen, Marilyn Vihman, Arnfinn Muruvik Vonen, and two anonymous reviewers for helpful discussion, comments and suggestions. Address for correspondence: Elizabeth Lanza, Department of Linguistics and Philosophy, University of Oslo, Post Box 1102 Blindern, 0317 Oslo 3, Norway (e-mail: [email protected]). 633

CHILD LANGUAGE

studies that have investigated code-switching in children have examined the speech of bilingual children three years of age and older (McClure, 1981 ; Fantini, 1985; Boeschoten & Verhoeven, 1987). Can, however, bilingual two-year-olds code-switch ? In a review of the literature on language mixing in infant bilingualism — the simultaneous acquisition of two languages — one will find that the answer to this question is a more or less implicit, yet firm 'no'. Infant bilinguals are claimed to lack the awareness of dealing with two languages and hence language mixing is interpreted as a sign of linguistic confusion. The study of infant bilingualism, also referred to as bilingual firstlanguage acquisition (Meisel, 1989) has focused on the young child who is exposed to two languages from birth. There has been some disagreement concerning the cut-off point between bilingual first-language acquisition and early second-language acquisition (see McLaughlin, 1984; DeHouwer, 1990). In an investigation of bilingual first-language acquisition, there is a need to focus on the child who has received some input in two languages from infancy. This input, however, may come from within the home or through contact with another language outside the home, in cases in which the home language is not the language of the speech community. A salient characteristic of the language use of these young children who acquire two languages simultaneously is language mixing. The term LANGUAGE MIXING is used in this article as a cover term for any type of linguistic interaction between two languages. This phenomenon is reported on in basically all studies of infant bilingualism although it may not be specifically addressed (for an overview see McLaughlin, 1984; DeHouwer, 1990). In those studies supporting the so-called one-system hypothesis of bilingual development, or as Genesee (1989) labels it 'the unitary language system hypothesis', language mixing by infant bilinguals is interpreted as a sign of the child's lack of awareness of actually dealing with two languages. Hence the child is claimed to form initially one linguistic system out of both languages. This view has received the most support in infant bilingualism research. Scholars holding this view claim that the bilingual child must undergo a process of language differentiation through which two separate linguistic systems are gradually formed (Swain & Wesche, 1975; Volterra & Taeschner, 1978; Redlinger & Park, 1980; Vihman, 1985; Arnberg, 1987). Some of these scholars have referred to the dawning of metalinguistic or bilingual awareness as an impetus to this language differentiation process. Implicit in these studies is the notion that an underlying bilingual awareness is only manifested in FORMAL separation of the two languages during the child's early years. There are a few studies on language mixing that have espoused the socalled two-system hypothesis, claiming that the language mixing in bilingual children is a result of the mixing present in the linguistic input (Bergman, 634

TWO-YEAR-OLDS AND CODE-SWITCHING

1976; Lindholm & Padilla, 1978). These studies focus on the bilingual development of Spanish-English-speaking children in the Southwest of the United States, an area known for its code-switching behaviour. Both bilingual first-language acquisition and second-language acquisition are, however, confounded in these studies. Despite this focus on language mixing in infant bilingualism, apart from Vihman (1985), no serious comparison has been made with sociolinguistic studies of code-switching. Vihman finds the actual mixing patterns of infant bilinguals to be formally different from the code-switching patterns of older bilinguals, this suggesting an underlying difference in bilingual awareness. However, as will be discussed below, apparent differences in the quality of mixing may be an artifact of the nature of young children's utterances. Whereas more mature speakers may embed their mixing in longer utterances, young children's utterances are typically much shorter. In order to take a new look at the question as to whether or not the bilingual two-year-old can code-switch, perspectives from sociolinguistic research into language choice and code-switching will be drawn upon. Poplack (1985: 33) in her work on language contact phenomena in adult bilingual speech (the distinction between code-switching, borrowing, and mixing as a result of language dominance) has stated: Clearly, if we are presented with a sentence of unknown pedigree containing elements from two codes, we cannot be sure of anything. We need to know the community patterns, both monolingual and bilingual, the bilingual abilities of the individual, and whether the context is likely to have produced speech in the code-switching mode or not. These aspects must also be considered in investigating the language mixing of infant bilinguals. Any differentiation process the young child undergoes must be investigated in relation to the community patterns of language use, within the child's process of LANGUAGE SOCIALIZATION (Ochs, 1988). In other words, children learn to differentiate their ways of speaking according to the needs of the social situation (Hymes, 1974). As some children are actually socialized into mixing languages, it is difficult to draw conclusions about the bilingual two-year-old's mixing in language OUTPUT without also investigating the child's language INPUT. The impact of language input, however, has been a neglected issue in studies of infant bilingualism (Genesee, 1989). And in those studies which do invoke input as an explanation for mixing (Bergman, 1976; Lindholm & Padilla, 1978), no operational means are provided for analysing that input. The general discourse structures of parents of bilingual two-year-olds in Australia were investigated by Dopke (1986); however, there is a need to focus on parental strategies toward child language mixing in order to address the issue of language socialization and code-switching. 635

CHILD LANGUAGE

Context has also been a neglected dimension in studies of infant bilingualism (Genesee, 1989). As Ferguson (1983:42) has pointed out, the notion of language differentiation has inevitably been 'adult-centric and overly concerned with languages as formal systems', without attention to situational variation. An operational notion of bilingual awareness that is proposed in this article is knowledge of when it is appropriate to keep both languages separate and when it is appropriate to mix languages, all dependent on the context of language use. The term CONTEXT has traditionally served as a cover term for a cluster of extralinguistic factors including setting, topic and the social characteristics of the participants in the study such as age, social class, ethnicity and education. However, context is not merely a static variable. Talk continually creates a context for more talk. A dynamic notion of context reveals that it is 'an achieved and negotiable part of any social interaction' (Cook-Gumperz & Gumperz, 1978: 12). This dynamic notion of context forms a theoretical basis for analyses of conversational discourse (Gumperz, 1982). The child's language use within the context of conversation provides an important anchoring point for investigating the child's bilingual awareness. Clark (1978: 34) has pointed out that 'adjusting one's speech to the age and status (and language) of the listener' is one of the earliest signs of the ability to reflect on language. And as Ochs (1988) has stressed, the organization of everyday conversational discourse can reveal important aspects concerning the child's psychological development. The term LANGUAGE MIXING has been used in studies of infant bilingualism to refer to various types of linguistic alternation. Some studies have examined syntactic mixing (e.g. Volterra & Taeschner, 1978; Taeschner, 1983) while others have looked at lexical mixing (e.g. Swain & Wesche, 1975 ; Redlinger & Park, 1980; Vihman, 1985). The unit of analysis has been the utterance, with a focus on MIXED UTTERANCES, utterances containing elements of both languages. In addition to mixed utterances, some scholars have also examined single-word utterances and multi-word utterances in one language used in the context for use of the other language. Genesee (1989) has pointed out that most proponents of the one-system hypothesis do not present or analyse their data by context; moreover, for those who do examine mixing as a function of interlocutor or context, their analyses are 'incomplete or questionable' (Genesee 1989: 166). Hence there is a need for a careful examination of language mixing within the context of conversational discourse. CODE SWITCHING, which occurs in bilingual discourse, involves the mixing of languages within and across utterances or sentences. A diversified terminology has evolved in the code-switching literature and controversy prevails concerning the actual data to be investigated for postulating constraints on intra-sentential code-switching (for an overview see Appel & Muysken, 1987; Romaine, 1989). Some scholars, for example, have called for a distinction between code-switching and borrowing, a distinction which

636

TWO-YEAR-OLDS AND CODE-SWITCHING

hinges on whether or not the foreign item is morphologically and syntactically integrated; phonology is not always a reliable criterion given individual variability and/or similarities in the phonological systems of a bilingual's two languages (Poplack, Wheeler, & Westwood, 1989). Other scholars have employed the term code-switching more broadly to cover various types of bilingual behaviour. As the very young bilingual child is in the process of developing the grammatical systems of both languages, the distinction between code-switching and borrowing is difficult to draw. In a discussion of infant bilingualism, a crucial distinction should be drawn between the type of linguistic behaviour that occurs in bilingual contexts as opposed to monolingual contexts. It is the child's inappropriate use of language mixing that must be examined as an indicator of the child's lack of bilingual awareness. Language mixing per se is not a valid measure for determining a lack of bilingual awareness. Hence in this article the question as to whether or not bilingual two-year-olds can code-switch is essentially a question as to whether or not the child exhibits pragmatic competence by mixing languages appropriately or keeping them separate, given the context of language use. This article presents the results of a case study on the language mixing of a bilingual two-year-old and evaluates alternative interpretations for the recorded mixing. The issue of language mixing is addressed through the investigation of (1) the formal aspects of the child's language mixing, as Vihman (1985) has claimed that the mixing patterns in infant bilingualism and more mature bilingualism are qualitatively different; and (2) the context of the child's language mixing, since code-switching is constrained by contextual variables as has been shown in sociolinguistic studies of bilingualism. An examination of these aspects may provide an alternative basis for evaluating the ' pedigree' of different types of language mixing in infant bilingualism (see Poplack, 1985, cited above).

METHOD

Data The data in this article come from a longitudinal investigation of the simultaneous acquisition of English and Norwegian in Norway (Lanza, 1990). In the family of the child reported on (Siri), the mother is American and the father Norwegian; the parents practise a one person-one language strategy of interaction with their daughter. Both parents are bilingual, with English as their main medium of communication in the home. At the time of data collection, Siri was an only child. The primary database consists of audiotape recordings of the child's spontaneous speech in separate father-child and mother-child interactions as 637

CHILD LANGUAGE

well as in family interactions. The recordings were made by the respective parent with whom the child was interacting; for example, the mother in the case of mother-child interactions. The parents were not given any instructions on how to act and, moreover, the parent and child determined the activity for the taping session. The mother-child and father-child interactions consist of two main activities: free play and book-reading. The family interactions were recorded at mealtimes. Secondary data consist of diary notes by the mother, interviews, and observations of the child by the author. As the study was a developmental one, recorded samples of speech were collected at approximately one-month intervals. As the samples consisted of recordings made over several days, a one-month interval was deemed suitable for differentiating the individual samples. The periods of investigation cover a developmental span of seven months beginning just prior to Siri's second birthday. Hence the approximate age for the first sample of speech (Siri I) is 2;o; the second sample (Siri II) is 2; i, and so on such that Siri in the final sample (Siri VIII) was 2;7. During this period, Siri advanced from one- and two-word utterances in Siri I to multi-word utterances with languagespecific word order in Siri VIII. The tapes were transcribed orthographically by a bilingual Norwegian-English speaker who had basic training in phonetics and linguistics. All of the transcriptions were checked by the author. Phonetic transcription was used where necessary.

Analysis Two basic types of analysis were performed on the data. In order to address the formal aspects of Siri's mixing, the unit of analysis was the MIXED UTTERANCE. An utterance is a word or a group of words with a single intonation contour. A mixed utterance consists of a co-occurrence of both languages either within one word (e.g. pigen 'the pig') or a group of words (e.g.jeg money ' I money') (Bold face is used to indicate the mixed item). In order to address the mixing of languages in the context of conversation, the unit of analysis was the CONVERSATIONAL TURN at talk. A turn at talk comprises an utterance or group of consecutive utterances bounded by a pause, or by an utterance of another participant in the conversation. Siri's turns at talk consisted of (1) an utterance of one word either in Norwegian or English; (2) a mixed utterance; (3) a multi-word utterance in either Norwegian or English; and (4) a group of utterances, typically a repetition of the same utterance, as in example (4) below (all utterances in the examples are demarcated by a slanted line - / ) . There are two important reasons for my choice of the conversational turn at talk. First, it is an interactive unit of analysis with a focus on the dynamics of ongoing communication. Secondly, it provides a theoretical basis for 638

TWO-YEAR-OLDS AND CODE-SWITCH I NG

including certain forms of repetition in the analysis. Repetition was dismissed in earlier studies of infant bilingual language mixing because children often use the same utterances repeatedly in various situations. Although repetition within a turn does not figure in the analysis, repetition across turns is of importance. For example, a child may repeat a mixed utterance across turns after a parent attempts to repair the utterance through a request for clarification. This repetition provides evidence that the mixed utterance was not considered by the child to be the trouble spot in the conversation. This aspect of conversation will be further discussed below.

RESULTS AND DISCUSSION

Formal aspects of language mixing Let us first address the claim that the actual mixing patterns of infant bilinguals are qualitatively different from the code-switching of older bilingual children and adults. For this analysis, the focus will be on the utterance. Vihman (1985) makes an important observation: in the speech of infant bilinguals, elements of the category of functors (grammatical words) prove to be the most often mixed as single items in mixed utterances, accounting for a larger percentage of both types and tokens in the mixed utterances of these young bilingual children. In the code-switching of older bilingual speakers, however, functors are rarely switched (as single items), the category of nouns being the most often mixed as a single category (see, however, Romaine, 1989). This differential distribution of mixing is interpreted as evidence for a qualitative difference between language mixing in infant bilingualism and code-switching among the older bilingual population. The data in Vihman's study are from her son who is bilingual in English and Estonian and cover the age span of 1 ;8 to 2;o; however, data from Redlinger & Park's (1980) study of the bilingual development of four bilingual two-year-olds are also used to support this claim. As these two studies focused on WORDS that are mixed, it is important to perform a similar analysis for comparison. Hence the analysis is implicitly limited to multi-word mixed utterances. A comparable examination of Siri's language mixing (focusing on words) yields similar results: Siri mixed more functors than content words in both types and tokens. See Table 1, which concerns tokens. (This table is based on Redlinger & Park's (1980) Table 3. The functors listed by Redlinger & Park in their table are the following: adverb, article, pronoun, preposition, conjunction. The slight changes made in this table were made to accommodate the types of mixes in Siri's data.) Table 1 indicates that four-fifths of the single item mixes in Siri's mixed utterances belonged to the class of functors, similar to the prediction made by Vihman. Examples of mixed utterances here include Mama stay ut ' Mama stay out'; jeg lean over ' I lean over'; ikke rain now ' not rain now'; 639

CHILD LANGUAGE

T A B L E I . Siri: distribution of single mixed words by word-category in all multi-word mixed utterances c

N Contentives Noun Verb Adjective Subtotal Functors Adverb Determiner Pronoun Preposition Conjunction/infinitive marker Modal auxiliary/copula Subtotal Total

26

7 -

15-5 4'1

33

196

38

22-6 8-9 43'5

IS

73 1

8 13s 168

o-6 4-8 804 1000

i et lite house 'in a little house'. These data on Siri's mixing patterns in mixed utterances, however, may be interpreted differently when analysed through a different approach. Let us first examine a more comprehensive analysis of the distribution of lexical and grammatical morphemes in Siri's mixed utterances, including multi-word utterances as well as single-word utterances with at least two morphemes. By lexical morpheme is meant open class items/contentives (nouns, verbs, adjectives) whereas by grammatical morphemes is meant both bound grammatical morphemes and function words/closed class items. The following general pattern is revealed: (1) English lexical morphemes co-occur with Norwegian and English grammatical morphemes, e.g. looker, looks; jeg eat, I eat. (2a) Norwegian lexical morphemes only co-occur with Norwegian grammatical morphemes, e.g. husker ('swing(s)'); jeg spiser ('I eat'). This entails the following: {zb) Norwegian lexical morphemes do NOT co-occur with English grammatical morphemes, e.g. *husks ('swings'); * I spiser. Hence Norwegian grammatical morphemes (bound morphemes as well as function words) co-occurred with Norwegian and English lexical morphemes while English grammatical morphemes only co-occurred with English lexical morphemes. Similar to Petersen (1988) who proposes the 'dominant language hypothesis', I interpret this DIRECTIONALITY OF MIXING as an indication of Siri's DOMINANCE in Norwegian, the majority language in her environment. In 640

TWO-YEAR-OLDS AND CODE-SWITCH I NG

many cases, Siri had acquired the equivalent English grammatical morphemes; however, she tended to use the Norwegian ones. Hence dominance must be differentiated from a state which Meisel (1989: 37) has noted in the literature and labelled FUSION, 'the alleged inability to separate two grammatical systems'. Dominance is not a necessary correlate of bilingual first-language acquisition. Some children may in fact be able from an early age to keep both languages in balance (Meisel, 1989; DeHouwer, 1990). Moreover, dominance may also be evidenced in adult code-switching patterns (Petersen, 1988). Furthermore, dominance is not static and may change if there are any changes in the linguistic environment (Leopold, 1949; Berman, 1979). Hence, mixing as a result of dominance cannot be invoked as evidence for the child's lack of language separation, that is, as a developmental stage to be overcome. The dominance of one language over another has been the object of testing in adult bilinguals (see Baetens Beardsmore, 1986). Berman (1979: 169) in her study of a young child suggests that dominance is affected by three interrelated aspects: ' quantity of situational exposure and variety of contexts of use; linguistic knowledge and proficiency; and cognitive processing and the nature of bilingual strategies'. Siri exhibited dominance in Norwegian through each one of these aspects. Her only daily English-speaking interlocutor was her mother whereas she interacted daily with many Norwegian speakers including her father, and had frequent contact with her Norwegianspeaking grandparents. Grosjean (1982) points out the main reason for dominance in one language is that the child has greater exposure to that language and needs it in order to communicate with more people. As for linguistic proficiency, Siri's morpho-syntactic development was more diversified in her Norwegian than her English; for example, languagespecific intra-sentential negation developed only in her Norwegian and not in her English during the period of study. Furthermore, her pronominal system was more developed in her Norwegian than in her English. The bilingual child will, of course, use what is available to express herself adequately. For example, Siri relied on a Norwegian pronoun in the mixed utterance 'han sick' during a period in which she had not yet acquired the English equivalent he. However, the fact that she would rely on Norwegian grammatical structure in communicating in English but never English grammatical structure (in cases in which she had both) when communicating in Norwegian must be interpreted as evidence for her dominance in Norwegian. The influence of dominance on the development of two morpho-syntactic systems is a rich field of inquiry with regard to the child's processing strategies in language development. However, for the focus of this article, the noted directionality of mixing is merely invoked as evidence for Siri's dominance, an example of Berman's third factor - the nature of her bilingual strategies. Swain & Wesche (1975) also noted that their informant's English 641

CHILD LANGUAGE

was filled in with French functors, a situation they interpreted as a reflection of the difference in the level of development of the child's English and French, French being his dominant language. Furthermore, Vihman (1985) reports that the majority of her son's mixed utterances involved the use of English function words with Estonian nouns. Vihman's subject, however, is claimed to be dominant in Estonian. Unfortunately, no criteria for this assessment of dominance are provided other than that Estonian was the home language. However, as noted by Genesee (1989), an important aspect of Vihman's study is that the home, or Estonian context, was apparently not a monolingual but rather a bilingual context. Petersen's (1988) three-year-old informant who was bilingual in English and Danish was claimed to be dominant in English. As Danish and Norwegian are very closely related languages, it is interesting to note that the mixing patterns for Petersen's informant are the mirror-image of Siri's. For example, Petersen listed such forms as washer and liver (an English verb with a Danish suffix) as non-occurring in her data whereas it is precisely this type of mixing that is found in Siri's data. In a corpus of 282 mixed utterances, there are only 3 exceptions to Siri's general pattern of mixing illustrated above. These exceptions, moreover, occurred in conversations in which Siri interacted with both of her parents, each speaking his or her own language. This non-systematic mixing may be attributed to the peculiarities of the situation of having to interact simultaneously in both languages (Swain & Wesche, 1975; Genesee, 1989). Indeed Grosjean (1982) has stressed that even adult bilinguals rarely manage to totally de-activate their one language when speaking in their other language. However, we must point out that the non-systematic mixing is minimal in relation to the systematic mixing. This only emphasizes the need to examine the bilingual child's mixed utterances in relation to her general pattern of mixing. Hence studies of bilingual children's mixed utterances must indicate the REPRESENTATIVITY of cited examples. A type of mixing at the word level that is not included in these analyses of Siri's mixed utterances is Siri's BLENDS, a form of lexical mixing in which a single phonological form is used to express the same concept in either of the two languages. Siri only used a few blends; for example, she once used the form [w D J k a] combining the English wash and the Norwegian vaske. This type of mixing has also been invoked by those advocating the one-system hypothesis as evidence for the young bilingual child's lack of bilingual awareness. Blends of this sort, however, -so-called 'loan blends' (Haugen, 1969)-are also attested in adult bilingual usage as well as in monolingual usage (DeHouwer, 1990) and hence cannot be invoked as evidence for a lack of language differentiation. Before addressing the question as to why functors appear to be so heavily mixed as SINGLE items in infant bilingual speech, as has been claimed by 642

TWO-YEAR-OLDS AND CODE-SWITCHING

Vihman (1985), it is important to point out that not all scholars report a preponderance of functors in mixed utterances. Even in Redlinger & Park (1980), the rate of functor mixing for one child was lower than that of contentive mixing. The distribution of mixing according to word class was not even performed on the mixed utterances of one of the children, due to his low level of lexical differentiation throughout the study. For the other two children, the rate of functor mixing was just over 50 %. An important aspect of the data collection procedures in their study, however, is the fact that the interviewer was a native speaker of English, also fluent in German and Spanish, who ' spoke only in German with the French/German subjects and primarily in German with the other subjects' (Redlinger & Park, 1980: 338-9). We cannot rule out the possibility that the bilingual identity of the interviewer may have had some impact on the child's contentive mixing. The mixing of functors as single items in young bilingual speech may be salient because of the fact that these children's utterances are, relatively speaking, simple compared to the utterances of adult bilinguals; for example, there are many child utterances consisting of only two words. The following is a list of some mixed utterances from Siri's data: (3) Some two-word mixed utterances: mer paper ('more paper'); jeg careful ('I careful'); meg finish ( ' m e finish'); den full ('it full'); bygge cow ('build cow'); net kitty ('no kitty'). Consider the utterance mer paper which could potentially be filled in either as 'Jeg vil ha mer paper', or the equivalent ' I want mer paper'. In other words, we may ask what the mix actually is in a two-word utterance. Is it the functor or is it the contentive ? Invariably, it is the CONTEXT of the utterance that is a major factor for determining which element in a mixed utterance is the mix. Context In the following, I will first present an analysis of Siri's language choice in interactions with each of her parents, using a static variable of context (i.e. context as a function of the interlocutor in conversation). Previous work on code-switching among children has revealed that the earliest systematic code-switching is a function of the category PARTICIPANT (McClure, 1981). Siri's parents were her most important caregivers, and consequently important participants in her development. And as they claim the one personone language strategy of interaction, the opportunity to investigate Siri's use of both of her languages is available. Subsequently, I will present a followup analysis using a dynamic notion of context. As we are examining language mixing within conversational discourse, the unit of analysis will be the conversational turn at talk, as previously defined. These conversational turns at talk contain single word or multi-word 643

CHILD LANGUAGE

utterances in either language, or mixed utterances. In the following analyses, all conversational turns with utterances containing lexical and grammatical morphemes are counted. As in the previous analyses, 'lexical mixing' will refer to the mixing of contentives (nouns, verbs, adjectives) whereas 'grammatical mixing' will refer to the mixing of grammatical bound morphemes and function words. Only utterances in turns which are interactional are included in the analyses, that is, all utterances in turns which the child addresses to the parent. Hence the child's talk to self, although of potential theoretical interest, does not figure in these analyses. Although every turn at talk is a potential locus for a language mix, my analyses will focus only on those turns at talk actually containing lexical and/or grammatical morphemes. Turns consisting of only unintelligible utterances, yes/no or the Norwegian equivalents ja/nei, or any vocalizations such as oh were excluded from the analyses.

Language mixing in conversation

In the following analyses, 'mixing' refers to Siri's use of Norwegian with her English-speaking mother and the use of English with her Norwegianspeaking father. In interactions with her mother, Siri's general mixing rate (turns with grammatical and/or lexical mixing) dropped between Period I and Period VIII, from 57 to 16 % (see Table A in Appendix). In interactions with her father, her general mixing rate started at Period I at 20 % only to rise at Period III to 4 3 % and finally descend in Period VIII to 14% (see Table B in Appendix). Siri essentially spoke Norwegian with her father and English with her mother, with a greater degree of mixing with her mother, especially at the beginning of the investigation. In conversations with both parents, Siri would essentially switch back and forth between languages in order to single out a particular addressee or to gain one or her parents' attention. Addressee specification as well as switching to attract or retain attention were also among the functions of codeswitching learned early by the youngest group in McClure's (1981) study of the development of code-switching among bilingual Spanish-English children in the US. Let us now examine the contents of Siri's conversational turns in individual parent-child interactions to determine whether the mixing was lexical or grammatical. Table 2 presents the distribution of lexical and grammatical mixing over time in Siri's conversations with her Englishspeaking mother. Apart from Siri I and Siri VIII, there is generally a greater degree of grammatical mixing than lexical mixing. Between Periods I and II, Siri visited the US and received what Arnberg (1987) has referred to as a ' linguistic shock' (resulting from communication failures) which apparently contributes to an increase in the awareness of dealing with two languages. 644

TWO-YEAR-OLDS AND CODE-SW ITCH I NG

T A B L E 2. Siri's lexical and grammatical mixing in interactions with her mother in relation to total number of turns with mixing in each age sample Total turns

Siri I Siri II Siri III Siri IV Siri V Siri VI Siri VII Siri VIII

Lexical mixing

Grammatical mixing

with mixing

59(79%) 5 (9%) 9 (i°%) i (6%) 4 ("%) o (o%) (Not available) n(52%)

16 (21%) 53 (9i %) 77 (90%) •5 (94%) 31 (89%) 22(100%)

75 58

10 (48%)

21

86 16 35 22

Before Period VIII, Siri received increased input in Norwegian due to greater contact with Norwegian-speaking children and exhibited, according to her mother's diary notes, a marked dominance in Norwegian. Siri continued to mix grammatically but began once again to mix lexically in conversations with her mother. Not only did Siri use Norwegian lexical items with her mother, but she also had problems in accessing the equivalent English item (demonstrated through hesitancy) when her mother would issue a request for clarification. Whereas Siri's mixing with her mother was both lexical and grammatical in nature, her mixing with her father was only lexical. In other words, Siri did not mix any English grammatical morphemes in her utterances with her Norwegian-speaking father although she did mix Norwegian grammatical morphemes in her utterances with her English-speaking mother. This analysis supports the claim made above, based on a linguistic analysis of Siri's mixed utterances, that Siri was dominant in Norwegian. In other words, Siri would rely on Norwegian grammatical morphemes in her conversations with her mother even though in many cases she had already acquired the equivalent English morphemes. An example is her use of the Norwegian jeg instead of the English equivalent /, as in the turn containing the utterance jeg full. This analysis of Siri's grammatical mixing points to dominance as an explanation. But what about Siri's lexical mixing? Lexical mixing has been termed a characteristic of more mature bilingual code-switching. Let us now examine Siri's lexical mixing rate with each parent. These rates refer to Siri's use of conversational turns with Norwegian lexical items in interactions with her mother, and with English lexical items in interactions with her father. The lexical mixing rate is calculated on the basis of total number of turns with lexical items. Hence turns consisting of only grammatical items are not included. Fig. 1 illustrates Siri's lexical mixing in interactions with each one 645

CHILD LANGUAGE 100 90 80 70 60 50 40 30, 20 10 0 Siri I

Siri II

Siri III

Siri IV

Siri V

• Mother

Siri VI Siri VII Siri VIII

• Father

Fig. i. Siri in interaction with her mother and father. Percentage of lexical mixing in relation to total number of turns with lexical items.

of her parents (see Table C in Appendix for raw values. Note that there are no data available with the father in Sample II, nor with the mother in Sample VII.) In Fig. i, we can see that, apart from Period I, Siri engages in more lexical mixing with her Norwegian-speaking father than with her English-speaking mother. How can we explain this in light of the fact that Siri is apparently dominant in Norwegian ? In other words, if mixing were only related to dominance one would predict more mixing when Siri speaks English with her mother, and indeed NO lexical mixing with her father. However, Siri's lexical mixing rate with her father was at a rather high level in the early periods investigated (especially Siri III and IV), and remained at a slightly higher level than that with her mother during the remaining periods. Even in the last period (Siri VIII) when Siri's Norwegian was found to be even more dominant in her interactions with her mother, this pattern prevailed. The lack of equivalent vocabulary items in her Norwegian could, furthermore, not be invoked as an explanation for this mixing as in most cases she demonstrated these equivalents. Moreover, the differences found between Siri's mixing to her mother and father could not merely be attributed to the nature of the activity the parent-child dyad was engaged in, since there was lexical mixing in all activities. Negotiation of context In order to attempt an explanation for Siri's lexical mixing, we will now draw on the dynamic notion of context. An examination of the parents' discourse strategies towards their daughter's lexical mixing revealed that the parents implicitly gave different signals or 'contextualization cues' (Gumperz, 1982) in their responses as to the appropriacy of Siri's mixing. In fact, the parents 646

TWO-YEAR-OLDS AND CODE-SWITCHING

demonstrated salient differences as to whether they opened negotiations for a monolingual or bilingual context with their daughter. For example, although Siri's mother as a bilingual understood her child's use of both English and Norwegian, she frequently feigned the role of a monolingual by questioning her child's utterance said in what she saw as the inappropriate language. Note the following exchange between Siri and her Englishspeaking mother. (4) Siri (2 ;2) and her mother are in the kitchen. Siri is drawing and has just asked for more paper. Siri Mother Siri run and find it. yeah/ Mama's standing right here. //Mama lope] ('run') Mama's got to look//after the food]. Mama lope/Mama lope/ Mama lope/ What do you want Mama to do ? run/ Run. Mama run/ Mama run. OK. {Siri's mother goes off to get paper.) In this exchange, Siri initially overlaps with her mother's utterance and responds repeatedly in Norwegian to her mother's request to run and find some paper. In a high rise intonation pattern (as an echo question), the mother replies by questioning what it is Siri wants her to do. Siri then changes the form of her utterance to English. This example of conversational repair on the part of the child illustrates her sensitivity to the ongoing interactional demands of her English-speaking mother. Instead of responding to her mother's request for clarification by a mere repetition, Siri repaired the form of her utterance by using the language expected by her mother. In a discussion of discourse strategies it is important to point out that whereas some strategies constitute a conscious plan for communication on the part of the speaker, not all are always consciously used. This emphasizes the need to examine the child's responses to the parents' discourse strategies, and the discourse the parent and child create together. The child's interpretations of repair cues, for example, can indicate the EXPECTATIONS the child recognizes in discourse (Ochs, 1988). Research on the ability of monolingual children to repair their utterances has indicated that young children (even as young as two) can locate the trouble spot in the conversation and make pragmatically appropriate responses to requests for clarification (see McTear, 1985). Siri's switching languages across utterances as 647

CHILD LANGUAGE

demonstrated in example (4) is, moreover, similar to McClure's (1981) example of code-switching for clarification. This type of switching was prevalent among the three-year-olds in McClure's study. In example (4), Siri's mother through her discourse strategy opened negotiations for a monolingual context with her daughter. We may refer to the mother's use of a PF/i-interrogative as a MINIMAL GRASP STRATEGY (Ochs, 1988) as she leaves it up to the child to reformulate the utterance. The mother conveyed to her child a meta-communicative message, namely, 'this is a context in which to speak English only'. Taeschner (1983) also provides examples of this strategy of feigning a lack of comprehension when the bilingual child speaks in the other language. She refers to this technique as an 'educational strategy' and claims it is the most successful in maintaining bilingualism. A bilingual context can also be negotiated, most obviously when an interlocutor code-switches to another language shared by the partner in conversation. However, bilingual discourse is also negotiated in other ways. Consider the following exchange between Siri and her father: (5) Siri (2 ;4) and her father are reading a book. Siri Father Hva er det for noe ? 'What's that?' name?/ -> En flodhest. 'A hippopotamus.' ja/ 'yes' In this example, Siri responds to her father's query by asking him for the name of the animal in question. The father's continuing of the conversation in example (5) functions as a cue to the child that her English utterance has been understood and that she may, therefore, continue to speak in English. In other words, no negative sanctioning in the form of a repair cue is given in response to Siri's use of an English lexical item. This demonstrates that it is not merely through an interlocutor's code-switching that a bilingual discourse can be created. This MOVE ON STRATEGY is identified as a parental utterance which exhibits comprehension of the child's use of a mix and continues the topic. An excellent example of this strategy is also found in Lindholm & Padilla (1978) in their examination of language mixing by bilingual children between 2; 5 and 6; 2. These scholars noted one motive for the children's switching from one language to the other was to ascertain whether the 'monolingual' experimenters were really monolingual. Scepticism was noted in several of the children about whether the experimenters really did not understand them when they mixed languages. The authors 648

TWO-YEAR-OLDS AND CODE-SWITCH I NG

state (Lindholm & Padilla, 1978: 36) that 'too many times experimenters reinforced the children's doubts by replying to a switch or a mixed utterance'. In other words, the experimenters essentially continued the conversation, responding to the content and not the form of the utterances. This emphasizes the need for examining the discourse context of child language mixing in experimental situations as well. In the Siri data, five basic discourse strategies were isolated as contributing to a negotiation of either a monolingual or bilingual context. These are listed in Table 3. These parental strategies towards mixing may be placed on a

T A B L E 3. Parental discourse strategies towards child language mixing 1. Adult requests clarification: MINIMAL GRASP STRATEGY (Ochs, 1988). 2. Adult requests clarification: EXPRESSED GUESS STRATEGY (Ochs, 1988). 3. Adult REPETITION of the content of the child's utterance, using the other language. 4. MOVE ON STRATEGY : the conversation merely continues. 5. Adult CODE SWITCHES.

continuum as in Fig. 2. This continuum provides an interpretive framework for analysing the discourse context of the young bilingual child's language mixing. The decisive factor for ranking these strategies as contributing either Monolingual Context

Bilingual Context Minimal grasp

Expressed guess

Adult repetition

Move on strategy

Codeswitching

Fig. 2. Parental strategies toward child language mixes.

to a monolingual context in conversation or a bilingual context is the degree to which they compel the child to attempt to adhere to the use of one language in interaction, or the degree to which they open up the opportunity for using both languages. Moreover, it is the degree to which the parent's identity or role as a monolingual or bilingual is highlighted. The notion of ROLES is a traditional theme of social organization: the individual is permitted and obliged to take on one role in one setting and another in another setting, with the role that is highlighted on one occasion being dormant on another. The sociologist Erving Goffman (1959), however, has argued for a recognition of a simultaneous 'multiplicity of selves' so that while the current activity establishes what the individual will mainly be, many of the individual's other identities will simultaneously be given little bits of credit. Goffman adapts the metaphor of the theatrical performance, in which there 649

CHILD LANGUAGE are front stages and back stages, in describing the individual's presentation of self in interaction. Drawing on Goffman's metaphor of the stage, we may say that the mother in her role of a monolingual was alone on the front stage in example (4) since she indicated a lack of comprehension of her daughter's use of Norwegian through her use of a request for clarification. The father's role as a bilingual, however, shared the stage in example (5) as he indicated comprehension of Siri's English and moved on in the conversation. Various analytical categories of clarification requests are found in the developmental literature. The two basic types are ' requests for repetition' and 'requests for confirmation' which correspond to Ochs' (1988) categories of clarification requests, the MINIMAL GRASP STRATEGY and the EXPRESSED

GUESS STRATEGY. Ochs' descriptive labels imply a difference in the degree to which the child is compelled to reformulate his or her utterance, and hence are more appropriate in this analysis. With the minimal grasp strategy, the parent relies primarily on the child to resay the repairable utterance, by using, for example, ' I don't understand', 'Say that again', and Wh interrogatives. With the expressed guess it is the parent who attempts the reformulation of the utterance(s) in a yes-no question form; the child may then confirm or disconfirm, as in the following example: (6) Siri (2;o) and her mother are looking at a book. Siri Mother tiss?/ 'pee' Aw, is he peeing ? -* yeah/ Notice that in employing an expressed guess strategy, the parent indicates comprehensive of the child's use of the other language, which is not the case with the minimal grasp strategy. In other words, it is more difficult to feign the role of a monolingual with the expressed guess strategy. As an alternative to the above strategies, the parent may repeat the meaning of the child's mix, using the other language but in a non-question form - the REPETITION STRATEGY. Similar to requests for clarification, these repetitions may function as repairs (Norrick, 1988). Notice, however, the following example. (7) Siri (2 ;3) and her father are changing her doll. Siri Father sann/ og ny diaper / ' like that / and new diaper' —» Og sa en ny bleie. 'And then a new diaper.' clothes ?/ 650

TWO-YEAR-OLDS AND CODE-SWITCH ING

Siri's father recasts his daughter's utterance with the Norwegian equivalent for 'diaper'. Siri, however, continues the conversation, with a lexical mix. The example illustrates that the role of the monolingual is highlighted less with the repetition strategy than with the minimal grasp and the expressed guess strategies. Requests for clarification call for an answer by the child, which is not the case with the repetition strategy. An important aspect of requests for clarification, as McTear (1985 : 169) notes, is 'their effect on the development of children's linguistic and communicative competence, in that they force children to monitor their language both for the forms they use and the ways in which they use these forms'. In summary, we may consider the repetition strategy less directed towards negotiating a monolingual context than the two types of requests for clarification. Both of Siri's parents claimed to use the one person-one language strategy; however, the mother actively negotiated a monolingual context with her daughter, particularly through the use of requests for clarification towards her daughter's lexical mixing. This strategy was consciously adopted after Period II (as noted in the mother's diary), since the mother feared that Siri would 'lose her English'. Siri's father, however, employed strategies that proposed more of a bilingual context. He actively used the repetition strategy, but also frequently modelled Siri's English words even though he would also supply the Norwegian equivalents, as in example (8) below: (8) Siri (2 ;o) and her father are looking at a picture book. Siri Father hug/ hug/ -» A, bamsen far en kos, eller en hug, ja. Se, der far bamsen en kos, Ja. ' Oh, the teddy bear gets a hug, or a hug, yes. Look, the teddy bear gets a hug there. Yes.' In example (8) Siri had devoiced the final stop in hug pronouncing it as [hAk]. Her father then provided her with a correct adult model of pronunciation in English. This modelling tendency further signalled the father's bilingual identity. Goodz (1989) in her study of bilingual FrenchEnglish children also found that parents claiming the one person-one language strategy in fact modelled their children's mixed utterances, thus signalling the appropriacy of mixing in interactions with that parent. In addition to this tendency to model Siri's English, the father also employed the move-on strategy and even code-switching in the later periods. Siri's mother, however, not only employed a conscious strategy of attempting to negotiate a monolingual context but she also refrained from language mixing either through code-switching or providing a model for 651 24

JCL 19

CHILD LANGUAGE

Siri's Norwegian. We may conclude that Siri's mother negotiated more of a monolingual context with her daughter whereas her father negotiated more of a bilingual context. It is through their responses to language mixing in conversation that the parents provided metalinguistic input as to the appropriacy of such mixing. As Cook-Gumperz (1986:54) has claimed, 'Children's language socialization occurs as part of the continuing history of conversational exchanges that make up daily life'. The analysis of the parents' strategies towards Siri's language mixing in conversation revealed that they developed early on as particular interactional styles towards mixing. In this regard, it is important to refer back to the need to examine the child's responses to the parents' discourse strategies. Parents may, for example, use requests for clarification because of some other aspect of the child's utterance. Compare examples (9), (10) and (11) all from the same sample (Siri III): (9) Sin (2 ;2) is playing with her doll and wants to dress it. Siri Mother Clothes ? yeah on/ clothes on/ clothes on/ Hm? -> baby • clothes on/ You want the baby to put her clothes on? yeah/ (10) Siri (2 ;2) and her mother are looking at a book. Siri Mother Where's the kitty cat? (softly) hop/ Hm? -> hop/ Uhuh. It's hopped up in the window. Jumped up in the window. (11) 5 m (2 ;2) has just clapped her hands. Siri klappe hand/ 'clap' Hm? -> clap hand/

Mother

In the above examples, we note that Siri interprets the same structural cue (' Hm ?') differently. In (9), she interprets it as a cue to expand her utterance, thereby making her reference clearer. In (10), this cue is interpreted as a 652

TWO-YEAR-OLDS AND CODE-SWITCHING

request for repetition as her initial utterance was too quiet. And in (i i), Siri interprets it as a cue to repair her utterance by replacing the Norwegian verb with its English equivalent. Siri's father also sometimes issued similar requests for clarification in the course of his conversations with his daughter, and some of these were in response to her mixing. However, in contrast to her responses with her mother, Siri would generally repeat the mix with her father thus providing evidence that she did not perceive the trouble spot in the conversation to be that of mixing. Consider example (12), taken from the last sample (Siri VIII). (12) Sin (2 ;y) and her father are talking about a visit to her Norwegian grandmother (Bestemor) who has been sick. They will be taking some medicine to her. Siri Father give Bestemor/ Hva sier du ? ' What are you saying ?' -» give Bestemor det/ 'give Bestemor that' Skal vi gi det til Bestemor ? ' Shall we give that to Bestemor ?' ja/ 'yes' Here Siri uses an English verb with a Norwegian suffix [g 1 v s] (the Norwegian equivalent of give is gi [ji:]). After the father's request for clarification, Siri repeats the lexical mix but repairs her utterance by making her reference clearer. In sum, the investigation of the context of Siri's language mixing has indicated that interactions with her mother were essentially monolingual contexts while interactions with her father were contexts that were more likely to produce speech in the 'code-switching mode' (see Poplack, 1985, cited above). This interpretation may explain Siri's lexical mixing patterns as illustrated in Fig. 1. Language mixing, bilingual awareness and code-switching What claim can be made about Siri's bilingual awareness? The notion of bilingual awareness as proposed in this article is knowledge of when it is appropriate to keep both languages separate and when it is appropriate to mix languages, all dependent on the context of language use. Siri engaged in language mixing in interactions with both of her parents in all of the periods investigated. However, a distinction may be drawn between grammatical mixing and lexical mixing. Siri mixed Norwegian grammatical items in her 653 24-2

CHILD LANGUAGE

speech to her English-speaking mother while no such mixing of English items occurred in her speech to her Norwegian-speaking father. Siri's grammatical mixing was interpreted as a sign of her dominance in Norwegian. Grammatical mixing can occur in a monolingual context; lexical mixing, however, is characteristic of a bilingual context. Therefore, it is Siri's lexical mixing that has served as a basis for evaluating her ability to code-switch and hence her bilingual awareness. As for the one-system hypothesis, the implication of this study is that there is no need to postulate a unitary language system in order to explain the child's language mixing. The issue of one system or two, with regard to underlying grammatical systems, is possibly an issue relevant for ALL bilinguals, if dominance is interpreted as a sign of a single system (see Grosjean, 1982). Hence this is not restricted to infant bilinguals as a stage to be overcome in attaining bilingualism. The notion of one system or two is not the relevant issue in searching for explanations for language mixing as evidenced in this paper. Children do learn to differentiate their languages; however, this differentiation process occurs in language socialization through which they learn to differentiate ways of speaking according to the social demands of the situation. Returning to the issue of code-switching, we may point out that although the ability to code-switch assumes an underlying awareness of language, even adult bilinguals are often quite unaware which language they are actually using at any one time since their main concern is with the communicative effect of their utterances (Gumperz, 1982; Romaine, 1989). As for children, McLaughlin (1984: 193) has stated, 'There is an obvious need for an analysis of the DIFFERENT levels of awareness through which the bilingual child passes'. Prior research on bilingual children three years old and older (e.g. McClure, 1981; Fantini, 1985) had revealed that the earliest systematic codeswitching among these children was situational and a function of the participant. The influence of setting and topic on code-switching was secondary to the influence of participant. Extensive use of code-switching as a stylistic device apparently does not appear until the ages of five or six, with the use of particular stylistic code-switching strategies not occurring until several years later. Hence these age differences in the use of code-switching apparently reflect a greater degree of bilingual awareness in older children. The investigation into Siri's language mixing supports this earlier research in that the influence of the participant was an important one in her lexical mixing. Hence we may claim that Siri has attained this early type of bilingual awareness. It is especially from age 253 (Siri IV) on that Siri shows sensitivity to context by (1) refraining from lexical mixing with her mother, who negotiated more of a monolingual context with her, and (2) using lexical mixing with her father, who negotiated more of a bilingual context. By this period, it appears that Siri has been socialized into respective interactional 654

TWO-YEAR-OLDS AND CODE-SWITCHING

patterns with each of her parents. Her lexical mixing in Period V (age 2; 4) with her mother consisted only of items which are phonetically similar in both languages. Siri's language socialization was, furthermore, exemplified by her responses to requests for clarification by her parents. With her mother, Siri switched languages while with her father, she attended to some other aspect of the utterance, apparently not perceiving the language mix as a trouble spot in the conversation. If one is to tease apart language mixing as an effect of dominance and language mixing as a response to context, it is the latter type of Siri's language mixing which should be referred to as code-switching - as a function of the participant. Siri's individual mixes were not constrained by topic or setting nor were the individual mixes stylistically motivated. It was pointed out that the distinction between code-switching and borrowing is difficult to draw in early bilingualism as the child is in the process of developing the grammatical systems of both languages. Hence this analysis has focused on the child's response to context in language choice. An analysis of Siri's language mixing without attention to context would have led to a quite different assessment of her bilingual awareness and communicative competence.

CONCLUSION

In sum, this investigation has indicated that there was no qualitative difference between Siri's language mixing and that of older bilinguals. The observed differences in mixing patterns could be attributed to dominance. An examination of the context of Siri's language mixing indicated that the child did differentiate her language use in contextually sensitive ways. This investigation stresses the need to examine more carefully the roles of dominance and context in the language mixing of young bilinguals. Hence a focus on both form and function has revealed that the bilingual child in this study could and did code-switch at the age of two. My claim, however, is not that the very young bilingual child can codeswitch with the same pragmatic sophistication as an older bilingual, in which individual switches may be pragmatically motivated. There are degrees of bilingual awareness. As the child matures, he or she will be able to resort to more sophisticated code-switching strategies in a display of bilingual identity. A theory of the development of code-switching has yet to be developed. For this, there is a need to bridge the gap between studies of language mixing in infant bilingualism and studies of code-switching among older bilinguals, for such a theory must encompass the language mixing of bilinguals under the age of three. The bilingual individual's use of context in language choice may thus be followed from infancy on into adulthood. The actual case reported on in this article was one of family bilingualism. In cases of societal bilingualism, 655

CHILD LANGUAGE

conversations with other interlocutors outside the home would provide a more comprehensive picture of language mixing within the process of language socialization. REFERENCES Appel, R. & Muysken, P. (1987). Language contact and bilingualism. London: Edward Arnold. Arnberg, L. (1987). Raising children bilingually: the preschool years. Clevedon: Multilingual Matters. Baetens Beardsmore, H. (1986). Bilingualism: basic principles. (2nd edn). Clevedon: Multilingual Matters. Bergman, C. (1976). Interference vs. independent development in infant bilingualism. In G. Keller, R. Teschner & S. Viera (eds), Bilingualism in the bicentennial and beyond. New York: Bilingual Press/Editorial Bilingiie. Berman, R. A. (1979). The re-emergence of a bilingual: a case study of a Hebrew-English speaking child. Working Papers on Bilingualism 18, 157-79. Boeschoten, H. & Verhoeven, L. (1987). Language mixing in children's speech: Dutch language use in Turkish discourse. Language Learning 37, 191—215. Clark, E. (1978). Awareness of language: some evidence from what children say and do. In A. Sinclair, R. J. Jarvella & W. J. M. Levelt (eds), The child's conception of language. Berlin: Springer-Verlag. Cook-Gumperz, J. (1986). Caught in a web of words: some considerations on language socialization and language acquisition. In J. Cook-Gumperz, W. Corsaro & J. Streeck (eds), Children's worlds and children's language. Berlin: Mouton de Gruyter. Cook-Gumperz, J. & Gumperz, J. J. (1978). Context in children's speech. In N. Waterson & C. Snow (eds), The development of communication. New York: John Wiley. DeHouwer, A. (1990). The acquisition of two languages from birth : a case study. Cambridge:

C.U.P. Dopke, S. (1986). Discourse structures in bilingual families. Journal of Multilingual and Multicultural Development 7, 493-507. Fantini, A. (1985). Language acquisition of a bilingual child: a sociolinguistic perspective. Clevedon: Multilingual Matters. Ferguson, C. (1983). 'Auf deutsch, duck': language separation in young bilinguals. Osmania Papers in Linguistics 9, 39-60. Genesee, F. (1989). Early bilingual development: one language or two? Journal of Child Language 6, 161-79. GofTman, E. (1959). The presentation of self in everyday life. Garden City, New York: Anchor Books. Goodz, N. (1989). Parental language mixing in bilingual families. Infant Mental Health Journal 10, 1, 25-44. Grosjean, F. (1982). Life with two languages : an introduction to bilingualism. Cambridge, MA: Harvard University Press. Gumperz, J. J. (1982). Discourse strategies. Cambridge: C.U.P. Haugen, E. (1969). The Norwegian language in America: a study in bilingual behavior. Bloomington: Indiana University Press. Heller, M. (ed.) (1988). Codeswitching: anthropological and sociolinguistic perspectives. Berlin:

Mouton de Gruyter. Hymes, D. (1974). Foundations in sociolinguistics. Philadelphia: University of Pennsylvania Press. Lanza, E. (1990). Language mixing in infant bilingualism: a sociolinguistic perspective. Doctoral dissertation, Georgetown University. Leopold, W. (1949). Speech development of a bilingual child: a linguist's record, Vol. 4: Diary

from age 2. Evanston, 111: Northwestern University Press. Lindholm, K. & Padilla, A. (1978). Child bilingualism: report on language mixing, switching and translations. Linguistics 211, 23-44. 656

TWO-YEAR-OLDS AND CODE-SWITCHING McClure, E. (1981). Formal and functional aspects of the code-switched discourse of bilingual children. In R. Duran (ed.), Latino language and communicative behavior. Norwood, NJ: Ablex. McLaughlin, B. (1984). Second-language acquisition in childhood. Vol. 1: Preschool children. (2nd edn.) Hillsdale, NJ: Erlbaum. McTear, M. (1985). Children's conversation. Oxford: Blackwell. Meisel, J. (1989). Early differentiation of languages in bilingual children. In K. Hyltenstam & L. Obler (eds), Bilingualism across the lifespan. Cambridge: C.U.P. Norrick, N. (1988). Repair in a bilingual family: the preference for other correction. Paper presented at the Eleventh Annual Meeting of the American Association of Applied Linguistics: 'The Acquisition and Assessment of Bilingual Language Ability', New Orleans. Ochs, E. (1988). Culture and language development: language acquisition and language socialization in a Samoan village. Cambridge: C.U.P. Petersen, J. (1988). Word-internal code-switching constraints in a bilingual child's grammar. Linguistics 26, 479-93. Poplack, S. (1981). Syntactic structure and social function of code-switching. In R. Duran (ed.), Latino language and communicative behavior. Norwood, NJ: Ablex. • (1985). Contrasting patterns of code-switching in two communities. In H. J. Warkentyne (ed.), Methods V: Proceedings of the V" international conference in dialectology. Victoria: University of Victoria Press. Poplack, S., Wheeler, S. & Westwood, A. (1989). Distinguishing language contact phenomena: evidence from Finnish-English bilingualism. In K. Hyltenstam & L. Obler (eds), Bilingualism across the lifespan. Cambridge: C.U.P. Redlinger, W. & Park, T. Z. (1980). Language mixing in young bilingual children. Journal of Child Language 7, 337-52. Romaine, S. (1989). Bilingualism. Oxford: Blackwell. Swain, M. & Wesche, M. (1975). Linguistic interaction: case study of a bilingual child. Language Sciences 37, 17-22. Taeschner, T. (1983). The sun is feminine. A study on language acquisition in bilingual children. Berlin: Springer-Verlag. Vihman, M. (1985). Language differentiation by the bilingual infant. Journal of Child Language 12, 297-324. Volterra, V. & Taeschner, T. (1978). The acquisition and development of language by bilingual children. Journal of Child Language 5, 311-26.

TABLE

APPENDIX A. Siri in interactions with her mother. Mixed turns in relation to total number of turns in each age sample

Siri I Siri II Siri III Siri IV Siri V Siri VI Siri VII Siri VIII

Mixed turns

Non-mixed turns

75(57%) 58(45%) 86(32%)

57 72

132

181

16(15%) 35(24%) 22(14%)

Total turns 130

94

267 no

109

144

138

160

"3

134

(Not available) 21(16%)

657

CHILD LANGUAGE TABLE

B. Siri in interactions with her father. Mixed turns in relation to total number of turns in each age sample

Siri I Siri II Siri III Siri IV Siri V Siri VI Siri VII Siri VIII

Mixed turns

Non-mixed turns

Total turns

16(20%)

66

82

20

84

35 no

151

172

(Not available) 15(43%) 26(24%) 21(12%) 12(11%) 12 (9%) 9(i4%)

99

i n

123

135

54

63

C. Siri in interactions with her mother and father. Lexical mixing rate: turns with lexical mixes in relation to total number of turns with lexical items in each age sample. (Basis for Figure 1)

TABLE

Mother

Siri Siri Siri Siri Siri Siri Siri Siri

I II

III IV V VI VII VIII

50%

(59/"7)

(5/106) A 0/ (9/239) 41 /o (1/104) /o 3 % (4/136) 0 % (0/150) (Not available) 1 0 % (11/106) 5%

658

Father 2 7 % (16/59) (Not iavailable) 5 0 % (15/30) 3 3 % (26/80) 18% (21/116)

• 3 % (12/96) 11 % (12/110) 19%

(9/48)

Can bilingual two-year-olds code-switch?

Sociolinguists have investigated language mixing as code-switching in the speech of bilingual children three years old and older. Language mixing by b...
2MB Sizes 0 Downloads 0 Views