International Journal of Speech-Language Pathology

ISSN: 1754-9507 (Print) 1754-9515 (Online) Journal homepage: http://www.tandfonline.com/loi/iasl20

The relationship between bilingual exposure and morphosyntactic development Elin Thordardottir To cite this article: Elin Thordardottir (2015) The relationship between bilingual exposure and morphosyntactic development, International Journal of Speech-Language Pathology, 17:2, 97-114, DOI: 10.3109/17549507.2014.923509 To link to this article: http://dx.doi.org/10.3109/17549507.2014.923509

Published online: 16 Jul 2014.

Submit your article to this journal

Article views: 191

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iasl20 Download by: [Florida Atlantic University]

Date: 06 November 2015, At: 02:07

International Journal of Speech-Language Pathology, 2015; 17(2): 97–114

The relationship between bilingual exposure and morphosyntactic development

ELIN THORDARDOTTIR

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

McGill University, School of Communication Sciences and Disorders, Montréal, Quebec, Canada

Abstract Purpose: The study examined the effect of bilingual input on the grammatical development of bilingual children in comparison to monolingual peers. Method: Spontaneous language samples were collected in English and French from typically-developing bilingual and monolingual pre-schoolers aged 3 years (n ⫽ 56) and 5 years (n ⫽ 83).Within each age group, children varied in bilingual exposure patterns but were matched on age, non-verbal cognition, maternal education and language status, speaking two majority languages. Measures included mean length of utterance (MLU) in words and morphemes, and accuracy and diversity of morphological use. Result: Grammatical development in each language was strongly influenced by amount of same-language experience. Children with equal exposure to both languages scored comparably to monolingual children in both languages, whereas children with unequal exposure evidenced similarly unequal performance across languages and scored significantly lower than monolinguals in their weaker language. Scoring significantly lower than monolinguals in both languages may, therefore, be a sign of language impairment. Each language followed a strongly language-specific sequence of acquisition and error patterns. Five-year-old children with low exposure to English displayed an optional infinitive pattern, a strong clinical marker for Primary Language Impairment in monolingual English-speaking children. Conclusion: Descriptive normative data are presented that permit more accurate interpretation of bilingual assessment data.

Keywords: Bilingualism, language impairment, assessment

Introduction Language evaluations of bilingual children are conducted both in clinical and research settings, to test for language impairment and to match groups of participants in research studies. The clinical identification of Primary Language Impairment (also termed Specific Language Impairment) in bilingual children is a difficult task and still associated with high rates of erroneous decisions (e.g. Bedore & Peña, 2008). A chief reason is that, because the language knowledge of bilingual children is distributed over the two languages, bilingual children frequently score significantly lower than monolingual counterparts when tested in either language separately. In other words, bilingualism can affect test scores in a similar way as does Primary Language Impairment, although for different reasons. When interpreting language assessment scores of bilingual children, clinicians frequently assume that allowances should be made given that bilingual children should not be expected to score as high as monolingual children in either one language. However, because of the lack of adequate normative data for bilingual children, it is

hard to estimate how much lower they should be expected to score and still not be considered to have a language impairment. One potential way around this problem that has received considerable recent research attention is the use of measures such as non-word repetition, which attempt to tap into the ability to process language rather than accumulated language knowledge (e.g. Elin Thordardottir & Brandeker, 2013; Girbau & Schwartz, 2008). A better understanding of the precise effect of bilingual experience on the development of each language and the development of bilingual norms would provide another way to increase diagnostic accuracy. This would have the added advantage of providing information not only on the presence or absence of Primary Language Impairment, but would also help establish therapy goals and evaluate the severity of the disorder and its impact on communication. This would contribute to a better understanding of the manifestation of Primary Language Impairment in bilingual children. The development of bilingual normative data is, however, complicated by the tremendous heterogeneity of bilingual populations. One of

Correspondence: Professor Elin Thordardottir, McGill University, School of Communication Sciences and Disorders, Beatty Hall, 1266 Pine Avenue West, Montréal, Quebec, H3G1A8 Canada. Email: [email protected] ISSN 1754-9507 print/ISSN 1754-9515 online © 2014 The Speech Pathology Association of Australia Limited Published by Informa UK, Ltd. DOI: 10.3109/17549507.2014.923509

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

98

E. Thordardottir

the principal factors on which bilingual children vary is the amount of exposure they have had to each of their languages. Previous studies have shown that language input exerts a significant influence on the development of vocabulary, for both monolingual and bilingual children (e.g. Huttenlocher, Haight, Bryk, Seltzer, & Lyons, 1991; Pearson, 2007). A recent study involving the same 5-year-old children who participated in the present study characterized this relationship in greater detail, showing that gradual increases in the relative amount of exposure to either language produced systematic and gradual increases in the vocabulary performance in that language, albeit with a non-linearity such that, beyond an exposure rate of around 50%, further increases in input resulted in smaller increases in performance. A background history of markedly unequal exposure to the languages was mirrored by a pattern of unequal proficiency. Interestingly, children having spent as little as 50% of their time with speakers of either language scored within normal limits compared to monolingual children in vocabulary. Thus, most children scored within normal limits in at least one language, and many in both. The study provided means and standard deviations on vocabulary scores for groups of children with different types of exposure patterns; these can be used as a preliminary normative reference for bilingual children with different exposure patterns that permits the examiner to set clear expectations of performance for individual bilingual children based on their background. The relationship with bilingual language exposure is much less well understood for grammatical development than it is for vocabulary. Considerable research is available on the grammatical development of young bilingual children. Longitudinal case studies, mostly of children raised following the one parent–one language principle, and receiving comparable amounts of input in both languages, have revealed that the two languages develop largely separately and follow a similar developmental sequence as that observed in monolingual children (De Houwer, 2005). However, the typical rate of bilingual grammatical development (or how fast it proceeds) in relation to amount of input or in comparison to monolingual development cannot be inferred from case studies of individual children. Further, because most of the cases involved children with roughly equal exposure to both languages, the typical grammatical patterns of bilingual children receiving unequal amounts of exposure have not been well documented. De Houwer stated that the Separate Development Hypothesis predicts that, since the two grammars of bilingual children are distinct, unequal levels of grammatical development in the two languages should be possible, but that little empirical evidence was available on children receiving such input.

An overview of review chapters on bilingual development reveals a common statement to the effect that the rate of grammatical development in bilingual children is, unlike vocabulary development, comparable to that of monolingual children (e.g. Bathia & Ritchie, 1999; Genesee & Nicoladis, 2007; Nicoladis & Genesee, 1997). These statements are based on observations of major milestones such as first words and first word combinations being attained at similar ages by bilingual and monolingual children (Genesee & Nicoladis, 2007) and by MLU levels of bilingual children being situated in the (low) normal monolingual range (Paradis & Genesee, 1996). In some cases, no particular reason is advanced. However, a lesser dependence of grammatical development on input has been attributed by some to the operation of innate triggering mechanisms that render the effect of quantity of input irrelevant as long as it is greater than zero (cf. Pearson, Fernandez, Lewedeg, & Oller, 1997). Chomsky (2011) has described the role of Universal Grammar (UG) in early language development as allowing children to “instantly and reflexively extricate language-relevant data from the blooming, buzzing confusion …” (p. 269). Pinker (1991) used the emergence of the English regular past tense to illustrate the existence of a modular rule system that is non-associative (insensitive to frequency and similarity effects) and, therefore, containing information that could not have been learned, but had to be given. Irregular inflections, in contrast, are seen as relying on memorization. The regular rule system is seen as having a genetic basis, with rules becoming available based on a pre-determined schedule, uninfluenced by environmental input (see also Berent, Pinker, & Shimron, 2002). In contrast, nativist accounts assume that vocabulary is acquired through learning from input without guidance from UG. Therefore, a finding that grammatical development requires significantly less input than vocabulary could be seen as an indication that these two domains of language are sub-served by clearly different mechanisms. Note, however, that not all nativist accounts assume triggering mechanisms, and some allow for a greater influence of input (Legate, & Yang, 2007). Therefore, a finding of no difference in learning rate across vocabulary and grammar does not rule out innate parameters. Paradis and Genesee (1996) studied language samples of three bilingual French– English toddlers and found that their grammatical development (Mean Length of Utterance) did not show appreciable delays compared to monolingual children. They concluded that their results were consistent with the existence of innate triggering mechanisms, because the bilinguals performed similarly to the monolingual children in spite of markedly less input in each language. Differential effects of input on vocabulary and grammar can also be consistent with accounts that do not involve innate parameters. Even if both

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Bilingual exposure and morphosyntax

vocabulary and grammar are assumed to be learned from input, vocabulary learning requires item-byitem learning, whereas grammatical development involves learning systematic principles or rules (cf. Oller, Pearson, & Cobo-Lewis, 2007). But, how are the rules learned? In monolingual children, a number of studies have shown the emergence of the regular past tense rule and syntactic complexity, as indexed by the appearance of past tense overgeneralization errors and use of increasingly complex sentences, respectively, to depend on the storage of a critical mass of lexical items from which regularities can be computed (e.g. Caselli et al., 1999; Elin Thordardottir, Ellis Weismer, & Evans, 2002; Marchman & Bates, 1994). Similar effects have been shown in bilingual children (Conboy & Thal, 2006; Marchman, Martínez-Sussman, & Dale, 2004). In addition, these bilingual studies showed that grammatical development (length of utterance, sentence complexity as measured by the MacArthur-Bates Communicative Development Inventory (CDI)) in either language (English or Spanish) was more strongly related to same-language vocabulary size than to overall vocabulary size combined across the languages or to grammatical development in the other language. This suggests that grammatical development in each language depends on a critical mass of stored vocabulary in that same language. Pearson (2007) proposed that the relationship between amount of exposure and language performance is not always direct, but rather involves threshold amounts resulting from critical mass effects. Indeed, during the period that the critical mass is being acquired but has not reached the critical size, grammatical progress is not observed (Marchman et al., 2004). A certain level of accumulated information is needed for a rule to emerge; beyond this point, more input is irrelevant for that particular rule. However, a substantial amount of exposure may be required to reach the critical mass. Gathercole (2007) reported critical mass effects in the grammatical development of school-age children learning English and Spanish, and English and Welsh, such that mastery of certain grammatical rules as assessed by targeted experimental tasks was related to certain input levels. Yet another particularity of bilingual development that has been proposed to be able to influence the rate of bilingual grammatical development is related to how the languages compare in the transparency of certain grammatical structures. Gathercole (2007) observed that the opacity of particular grammatical rules in a given language influenced how much input was required for the rule to be learned by bilingual children. Such effects have been observed as well in cross-linguistic research which have shown that the growth trajectories of particular grammatical structures differ across languages (e.g., for a comparison of French and English morphology, Elin Thordardottir, 2005). In the case of bilingual children, it has been

99

suggested that the two languages might interact in such a way that structures that are more frequent or more transparent in one language of the bilingual child might be learned more easily in the child’s other language than they are by monolingual children. Roeper (2011) proposed that if a phenomenon appears in one grammar of a bilingual child, it can trigger it in the other grammar, and goes on to propose that this does not implicate transfer between the languages but an interface at a more abstract level. The exact structures that could benefit from such language interaction would vary depending on the particular language combination. Several researchers have examined potential interaction effects in individual children or small groups, with varying results (e.g. Döpke, 2000; Liceras, Fernandez Fuertes, & Alba de la Fuente, 2011; Paradis & Navarro, 2003; Silva-Corvalán & Montanari, 2008; Yip & Matthews, 2000). A clear example of a potential acceleration effect of this kind in the case of French and English is the much richer use and earlier development of finiteness in French than in English, which has been documented in monolingual children (Elin Thordardottir, 2005). Paradis and Genesee (1996), however, concluded that the development of finiteness in English in their three participants was not accelerated by the fact that finiteness was more developed in their French. Although language structures are found in the speech of bilingual children that could result from interaction between the languages, the overall effect of these on the rate of grammatical development has not been evaluated in group studies. In contrast to the view that the grammatical development of bilingual children is overall within the normal monolingual range, several recent studies have started to advance evidence that amount of bilingual exposure does significantly influence grammatical development. Nicoladis and Marchak (2011) showed that bilingual French–English pre-school children were less accurate than monolingual peers in the use of gender marking and Nicoladis, Palmer, and Marentette (2007) showed bilingual French– English children to be less accurate than monolinguals in the use of past tenses. The choice of grammatical structures targeted in these studies was based on the assumption that gender and tense are both structures that are relatively highly dependent on input for learning. Significant effects of input were also found by Paradis, Nicoladis, Crago, and Genesee (2011) for production of past tenses in an elicitation task, on which bilingual children were less accurate than monolingual children. Several recent studies have targeted the effect of input more directly. Elin Thordardottir, Rothenberg, Rivard, and Naves (2006) reported significant differences on vocabulary and grammatical measures for a group of French–English bilingual pre-schoolers having had equal exposure to both languages in comparison to monolingual children of the same age. Grammatical

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

100

E. Thordardottir

measures included a receptive (RDSL, Reynell Developmental Language Scales, Reynell, 1969) and expressive measure (MacArthur-Bates CDI, Fenson et al., 1993). Working with a similar agegroup of Spanish–English bilingual children and using the CDI in both languages, Hoff, Core, Place, Rumiche, Senor, and Parra (2012) showed clear differences between monolingual children and some groups of bilingual children on both vocabulary and grammatical CDI sub-sections based on exposure patterns. Blom (2010) documented differences in Mean Length of Utterance in words (MLUw) and vocabulary diversity of four Dutch–Turkish bilingual children, revealing that children with a clear dominant vs weaker language environment scored significantly differently than monolinguals, whereas the difference between bilinguals and monolinguals was less clear for children having received more equal exposure. These studies indicate that early grammatical development is sensitive to the amount of input received. However, the exact nature of this relationship cannot be discerned from these studies— whether the effect is gradual as for vocabulary or involves a clearer threshold effect, what the magnitude of this effect is and to what extent it is different for children with different language exposure histories, such as equal vs unequal exposure to each language. A more refined understanding of the typical performance of bilingual children in relation to their language exposure background will contribute importantly to a better ability to distinguish between typical and impaired patterns of development. Research questions and study design The study examines the grammatical development in French and English of bilingual pre-schoolers exposed to these languages to varying relative degrees and compares their development in each language to that of monolingual children. It was hypothesized that increased exposure to either language would lead to a higher grammatical level in that language. Language samples were obtained for a total of 139 children in two age groups, 3 years and 5 years. Within each age group, children with different language exposure histories were equivalent in age, socio-economic status (SES), non-verbal cognition and language status (majority vs minority). These variables are frequently confounded in bilingual populations, but could be controlled due to the unique context in Montreal, where both French and English are majority languages and a comparable range of SES levels is found in both language groups. Language samples were analysed for Mean Length of Utterance (MLU) in words and morphemes, morphological diversity and morphological accuracy in both languages. The performance of the bilingual children was first examined with amount of exposure treated as a continuous variable. Group com-

parisons were then made in each language between monolingual children and bilingual children with more exposure to English, equal exposure to English and French and more exposure to French. Descriptive data were obtained on the particular grammatical morphemes used productively by children in the different exposure groups. Participants Participants included two age groups of pre-schoolers: Younger (3 years, n ⫽ 56) and Older (5 years, n ⫽ 83). The upper range of 5 years was set to enable recruitment of children before school entry, as the language environment of many Montreal children changes significantly at that point (most notably with many children from English homes entering French-immersion programs) and also because, in Quebec, formal evaluation aiming to diagnose Primary Language Impairment (or dysphasie) is frequently conducted around age 5 years. All of the children were considered to be simultaneous bilinguals, having had regular bilingual exposure from before age 3 years. For the 3-year-olds, the mean age of onset of bilingualism was 5.23 months (SD ⫽ 6.65, range ⫽ 0–24). Fifty percent of the 3-year-olds had been exposed to bilingualism from birth, 87% from age 12 months or younger and 100% from age 24 months or younger. For the 5-year-olds, the mean age of onset was 12.5 months (SD ⫽ 11.45, range ⫽ 0–36). Thirty-one percent of the 5-year-olds had been exposed to bilingualism from birth, 57% from age 12 months or younger, 85.7% from age 24 months or younger and 100% from age 36 months or younger. (A previous study on the vocabulary scores of the 5-year-olds revealed that when children were equated on total amount of exposure to each language, early vs later onset (before 6 months vs after 21 months) produced no difference in scores.) Exposure to French and English over the children’s lifetime was assessed by a detailed parent questionnaire (previously used in Elin Thordardottir et al., 2006; Elin Thordardottir, 2011; Elin Thordardottir & Brandeker, 2013). The questionnaire documents, for each year of the child’s life, the percentage of waking hours spent in French vs English (or bilingual) situations, by asking parents factual information about time spent at home, daycare and other relevant situations (dates of daycare attendance, number of hours per week, communicative partners, languages used in each environment). The calculation considered situations involving potential interactions with people (excluding TV watching) and yielded a single number for each child ,representing the percentage of their time since birth spent exposed to English, with the remainder being spent in French. This percentage can be viewed as an estimate of time on task in each language. It represents the relative amount rather than absolute amount of waking hours giving the child opportunity for English vs French interactions. More information

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Bilingual exposure and morphosyntax

on the quality of bilingual interactions might reveal further effects of input. However, the procedure used here is similar to the treatment of amount of exposure in monolingual norms, which typically consider the number of months or years that have passed since the child’s birth and that were, therefore, spent in the monolingual environment in question, without further consideration of how much interaction actually occurred. Based on the exposure questionnaire, children were divided into five exposure groups: Monolingual French (MF, 5% or less English), more French (FF, 6–39% English), equal French and English (EF, 40–60% English), more English (EE, 61– 94% English) and monolingual English (ME, 95% or more English). The 3-year-old (Younger) group included 56 children with a mean age of 34.34 months (SD ⫽ 3.51), including 14 MF, seven FF, 13 EF, 10 EE and 12 ME (26 monolingual and 30 bilingual). One-way analysis of variance using these five sub-groups revealed, for the Younger group, no significant difference across groups for age (p ⫽ 0.414), non-verbal cognition (p ⫽ 0.379) or SES (p ⫽ 0.799). A significant group effect was found between groups for exposure to English (F(4,51) ⫽ 367.372, p ⫽ 0.000) with post-hoc tests (Tukey HSD) revealing that every group differed from every other group. The 5-year-old (Older) group included 83 children with a mean age of 58.34 months (SD ⫽ 3.93), with the following distribution into exposure groups: 19 MF, 20 FF, 13 EF, 16 EE and 15 ME (34 monolingual and 59 bilingual). Oneway analysis of variance revealed no significant difference between bilingual exposure groups in age (p ⫽ 0.062), maternal education (p ⫽ 0.166) or non-verbal cognition (p ⫽ 0.697), but a significant group difference for amount of exposure to English (F(4,78) ⫽ 571.621, p ⫽ 0.000). Post-hoc testing revealed that all groups differed significantly from all other groups in this respect. Background variables are reported in Table I, showing means and SDs for each exposure group. The exposure groups

were formed to facilitate analyses and do not imply that these groups are seen as being qualitatively different types of bilinguals. All children included in the samples demonstrated typical development in language and in other developmental areas as per parent report (no developmental diagnoses or significant worries, no major illness or hospitalizations). The children were recruited through various daycare centres and preschools in Montreal. They varied as to whether they were exposed to both languages at home, to one language at home and the other at daycare, both languages at home and one in daycare or to both languages at both home and daycare. Detailed information on this aspect for the older group was reported in Elin Thordardottir (2011). One third of the bilingual 3-year-olds and 44.8% of the bilingual 5-yearolds were from monolingual homes. Within each age group, variability in age was deliberately kept small. Procedure Children were tested individually by trained research assistants; monolingual children in one session and bilingual children in two sessions, with French and English measures administered on separate days and by separate native-speaking examiners who gave no indication of speaking any other language. The order of testing of French and English was counterbalanced across participants. The non-verbal cognitive test was administered in either session, usually the first (using the brief IQ scale of the Leiter International Performance Scale- Revised (Roid & Miller, 1997) which involves no use of language in the instructions or test items). Spontaneous language samples were collected in a conversational interaction with the examiner using age-appropriate standard sets of toys (for 3-year-olds, a Fisher price house and a set of food and household items; for 5-year-olds Playmobil and Polly Pocket toys). Language sampling was attempted in both languages for all children reported to have had more than 5%

Table I. Participant background characteristics, means (SD). 3-year-olds Age MF FF EF EE ME

35.8 33.4 34.3 34.3 33.3

(2.8) (3.5) (4.2) (4.4) (2.4)

Leiter 114.4 120.5 107.1 113.2 106.9

(13.4) (15.6) (15.5) (18.9) (19.5)

5-year-olds

Mat. Ed. 16.7 16.9 17.0 18.3 17.6

(2.3) (4.7) (2.9) (3.6) (3.7)

% Engl. 1.1 20.7 50.8 75.6 95.2

(2.1)* (12.4)* (6.4)* (10.2)* (1.6)*

101

Age 59.0 58.0 60.5 56.4 58.1

(4.6) (3.8) (4.2) (2.6) (3.3)

Leiter 105.6 108.3 101.3 104.2 103.8

(12.3) (12.8) (16.8) (15.8) (11.1)

Mat.Ed. 16.3 18.2 17.0 16.9 18.0

(2.3) (2.7) (2.3) (2.7) (3.3)

% Engl. 1.0 21.1 51.2 76.3 98.1

(2.2)* (8.3)* (7.5)* (9.9)* (2.5)*

MF, Monolingual French exposure; FF, More French than English exposure; EF, Equal French and English exposure; EE, More English than French exposure; ME, Monolingual English. Age in months. Leiter: Brief IQ standard score (the test has a mean SS of 100 and SD of 15). Mat. Ed., maternal education in years; % Engl., percentage of waking hours since birth spent in English environments (remainder spent in French environments). * significant difference between all groups.

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

102

E. Thordardottir

exposure to their weaker language (children with up to 5% exposure were classified as monolingual). Not all children, however, produced full usable samples in both languages, in particular those with very low exposure levels to one language. Some of these children produced samples that were entirely codeswitched relative to the language used by the examiner (consistently responded in the other language throughout). Such samples were not included in the analyses. In rare cases, technical difficulties, time constraints or other reasons precluded the collection of both samples. For the Older children, a total of 56 usable English samples and 57 French usable samples were obtained. Samples were obtained in both languages for 31 children (eight in the EF group, 10 in the EE group and 13 in the FF group). Among the Younger children, a total of 42 English samples and 34 French samples were obtained, with samples in both languages for 18 children (nine in the EF group, five in the EE group and four in the FF group). It should be noted that, even for those children who produced a sample only in one language, one sample is informative on the language level in that language and the failure to produce a sample in the other language advances our understanding of the amount of exposure required to be able to produce spontaneous speech in two languages. Availability of samples in both languages is useful for within-subject analyses of grammatical performance across languages. Language samples included 100 utterances, excluding imitations. Utterances with unintelligible words were included. Partially code-switched utterances (typically involving a lexical borrowing) were included and contributed to MLU. However, only the target language of the sample was coded for morphology in such utterances. These utterances were included because utterances with minor unintelligible or code-switched segments are easily transcribed and analysed. Excluding them would have diminished the available data and might have led to significant biases in that children may tend to produce more unintelligible segments in their more complex utterances. Entirely code-switched utterances were excluded (for example, a series of fully French utterances within an English sample). The samples were transcribed in English according to SALT conventions (Miller, Andriacchi, & Nockerts, 2011) and in French using the French adaptation of SALT conventions (Elin Thordardottir, 2005). This type of analysis is widely used clinically as well as in research, and captures developmental increases in MLU and morphological diversity and accuracy within the age range covered here (Leadholm & Miller, 1992; Miller et al., 2011). The French adaptation is based on parallel principles but focuses on the more complex morphology of French. The French coding procedure has been shown to be developmentally sensitive for monolingual children from 20 and up to at least 66 months (Elin Thordardottir, 2005; Elin

Thordardottir, Kehayia, Lessard, Sutton, & Trudeau, 2010). The grammatical morphology targeted in the English coding included Brown’s morphemes (progressive –ing, plural –s, third person –s, regular past tense –ed, and various contracted verb forms (copulas, auxiliaries and modals). The irregular past tense which is not coded in SALT analysis (because it is viewed as being used by rote by very young children) was coded in this study to enable tracking of both types of past tenses. The coding of French morphemes targeted verb person, tense (when other than the present) and mood (when other than the indicative), plural marking of nouns, adjectives and pronouns, and the gender marking of adjectives and pronouns. The coding of English and French morphology was not meant to be identical or directly comparable across the two languages but rather to target developing structures in each language. To ensure the reliability of transcription and coding, the procedure proposed by Heilmann, Miller, Iglesias, Fabiano-Smith, Nockerts, & Andriacchi (2008) was followed, whereby each sample is reviewed by an independent scorer who views the videotaped interaction and the original transcription, making changes as needed. The proposed changes are then reviewed by the first coder, and final decisions are made by consensus, or by the first coder if consensus is not reached. All samples were subjected to this reliability procedure, resulting in inter-judge agreement exceeding 90%. The procedure ensures that the final version of each coded sample has been fully reviewed and corrected by two coders. Data analyses The following measures were calculated from each language sample: (1) Mean Length of Utterance in words (MLUw) and in morphemes (MLUm). (2) Accuracy of grammatical morphology (percentage correct use of grammatical morphemes in obligatory contexts—obligatory contexts were calculated by adding all instances coded as correct use plus all instances coded as incorrect use, including omission and substitution errors. Therefore, children who had no contexts that required the use of a particular morpheme did not contribute to this analysis). In English, accuracy is reported for contracted verb forms, tense (regular and irregular past), 3rd person –s, progressive –ing, and noun inflections (plural and possessive –s). In French, accuracy is reported for verb person marking, verb tense, verb mood, gender (adjective and pronoun agreement with referent) and plural (plurals of nouns, adjectives and pronouns other than personal pronouns). (3) Diversity of grammatical morphology (number of different types of grammatical morphemes used). The diversity of morphological use further considered whether there was evidence of productive use of the morpheme following criteria used previously by Bassano, Maillochon, Klampfer, & Dressler (2001) and Elin

103

Bilingual exposure and morphosyntax

Thordardottir (2005): sporadic use (appearing in the sample of at least one but less than half of the members of the group), consistent use (appearing in samples of half or more of the members of the group) and productive use (appearing with at least two different words in samples of half or more of the members of the group).

groups in the same language. Results are reported for Older and Younger groups separately, first for English and second for French. Mean length of utterance and morphological diversity as a function of percentage exposure to English By way of example of the curve estimation analysis, Figure 1 illustrates visually the relationship between percentage exposure to English (x-axis) and English and French MLUm, respectively, for the Older children.

We first present results on MLU and morphological diversity data in French and English as a function of language exposure. This analysis uses curve estimation and treats the percentage of previous exposure to English as a continuous variable (the remainder of the time, 100% – percentage of exposure to English, represents the percentage of exposure to French). These analyses include the entire groups of Older and Younger children; however, analysis of French excluded the monolingual English group and analysis of English excluded the monolingual French group. Comparable curve estimation analyses were reported for the Older group previously for receptive and expressive vocabulary (Elin Thordardottir, 2011) and sentence imitation and non-word repetition (Elin Thordardottir & Brandeker, 2013). For further examination of statistical differences between children having received different amounts of exposure to each language, group comparisons were made for MLU as well as morphological accuracy and diversity using the exposure level sub-groups (ME, EE, EF, FF and MF, see Table I). For French measures, this comparison involved those four exposure groups tested in French, excluding the monolingual English children. For English, this comparison involved all groups except the monolingual French speakers. For each of the measures, Analysis of Variance was used to compare the performance of different exposure

English samples. For the Older group, a significant association was found between exposure to English and MLUm (linear fit: r ⫽ 0.489, p ⫽ 0.000; quadratic fit: r ⫽ 0.511; p ⫽ 0.002) and for MLUw (linear fit: r ⫽ 0.483, p ⫽ 0.000; quadratic fit: r ⫽ 0.493, p ⫽ 0.001). For morphological diversity, both linear and quadratic fits were significant (linear: r ⫽ 0.327, p ⫽ 0.014; quadratic: r ⫽ 0.446, p ⫽ 0.003); however, no significant fit was found for the diversity of morphology used productively. For the Younger group, a significant linear fit was obtained for MLUm (r ⫽ 0.349, p ⫽ 0.022; the non-linear fit approached significance, p ⫽ 0.072), but no significant fit was obtained for MLUw (p ⫽ 0.089). Significant fits were found for diversity of morphology (linear: r ⫽ 0.400, p ⫽ 0.009; quadratic: r ⫽ 0.520, p ⫽ 0.002). Similarly, for diversity of morphology used productively, both linear and quadratic fits were significant at the 0.05 level (linear: r ⫽ 0.345, p ⫽ 0.012; quadratic: r ⫽ 0.415, p ⫽ 0.025). French samples. For the Older Group, no significant linear fit was found for MLUm; however, the nonlinear fit approached significance (linear: p ⫽ 0.434; non-linear: r ⫽ 0.347, p ⫽ 0.059). Significant fits were

9

7

8

6 MLU in morphemes in French

MLU in morphemes in English

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Results

5 4 3 2 1

7 6 5 4 3 2 1

0 0

10

20

30

40

50

60

70

80

Percent exposure to English

90 100

0 0

10

20

30

40

50

60

70

80

90

100

Percent exposure to English

Figure 1. Mean length of utterance in morphemes in English samples (left panel) and in French samples of the Older group (5-year-olds) as a function of percentage of time spent in English environments since birth.

104

E. Thordardottir

found for diversity of morphology (linear: r ⫽ 0.391, p ⫽ 0.003; quadratic: r ⫽ 0.506, p ⫽ 0.000). However, for diversity of morphemes used productively, only the quadratic fit was significant (quadratic: r ⫽ 0.532, p ⫽ 0.000; linear: p ⫽ 0.43). For the Younger Group, no significant fit was obtained for MLUm (linear: p ⫽ 0.086; quadratic: p ⫽ 0.096) or MLUw (linear: p ⫽ 0.181; quadratic: p ⫽ 0.193). Significant linear and non-linear fits were found for diversity of morphology (linear: r ⫽ 0.377, p ⫽ 0.031; quadratic: r ⫽ 0.534, p ⫽ 0.007). The diversity of morphology used productively was not significantly related to percentage of previous exposure (linear: p ⫽ 0.080; quadratic: p ⫽ 0.124).

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

MLU in morphemes and in words Group means for MLU for the five exposure groups are reported for both age groups in Table II. English samples. In both age groups, visual inspection shows that MLU increased systematically with increased exposure to English with the order FF ⬍ EF ⬍ EE ⬍ ME (ranging from 3.17–4.54 for the Older group and from 2.31–3.22 for the Younger group). For the Older group, a significant exposure group difference emerged for both MLU measures (MLUw: F(3,49) ⫽ 4.673, p ⫽ 0.006; MLUm: F(3,49) ⫽ 5.106, p ⫽ .004). Post-hoc tests (Tukey HSD) revealed the same pattern for both MLUw and MLUm: the FF group performed significantly lower than the ME and EE groups, but not significantly differently from the EF group. No significant group differences were found on either measure for the Younger group (MLUw: p ⫽ 0.120; MLUm: p ⫽ 0.064). French samples. For both age groups, MLUm was lowest in the EE group and highest in the MF group. For the Older group, MLUm ranged from 4.55–5.82; with the order among the bilingual groups EE ⬍ EF ⬍ FF. However, the EF and FF groups both had higher means than the MF group. For the Younger group, MLUm ranged from 2.99–4.12 and increased with increased exposure to French, although the EF

group had a higher mean than the FF group. ANOVA analysis revealed no significant group difference for either MLU measure for the Older group (MLUw: p ⫽ 0.393; MLU: p ⫽ 0.148) or the Younger group (MLUw: p ⫽ 0.398; MLU: p ⫽ 0.187). Accuracy of grammatical morphology Group means for accuracy of grammatical morphemes for the Older and Younger groups are displayed in Figure 2. English samples. For the Older children, accuracy of production of grammatical morphemes varied between exposure groups from 70–98% for contracted verb forms, progressive –ing, past tense and noun inflection, but from 44–90% for the third person –s, with the FF group having the lowest group mean. A significant exposure group difference was found for the contracted verb forms (F(3,49) ⫽ 3.402, p ⫽ 0.025, η2 ⫽ 0.17), but posthoc tests (Tukey HSD) revealed no significant pairwise differences between individual groups although the difference between the FF and EF groups and the FF and ME groups approached significance (p ⫽ 0.055 and 0.087, respectively). A significant group difference was also found for plurals (F(3,47) ⫽ 3.197, p ⫽ 0.032, η2 ⫽ 0.17), with posthoc tests revealing that the FF group performed significantly lower than the ME group. The difference between the ME and FF groups and EE and FF groups approached significance (p ⫽ 0.051 and 0.090, respectively) with the FF group performing lower in each case. Finally, a significant group effect was found for the third person –s (F(3,39) ⫽ 4.703, p ⫽ 0.007, η2 ⫽ 0.27), with post-hoc testing revealing a significant difference between the FF and ME groups (p ⫽ 0.016), and that the difference between the FF and EE groups approached significance (p ⫽ .090), with the FF group in each case performing lower than the other group. No significant group differences were found for tense (p ⫽ 0.152) or the progressive –ing (p ⫽ 0.125). For the Younger group, performance varied markedly between morphemes and between groups.

Table II. Group means and (SD) for mean length of utterance in words and morphemes for 3-year-olds and 5-year-olds. 3-year-olds English MLUw MF FF EF EE ME

2.28 2.46 2.97 2.95

(0.63) (0.68) (0.95) (0.77)

French

MLUm

2.31 2.66 3.25 3.22

5-year-olds

MLUw

English

MLUm

3.26 (0.66) 4.12 (0.64) 2.74 (0.76) 3.35 (0.84) 3.18 (0.95) 3.98 (1.02) 2.69 (0.74) 2.99 (0.89)

MLUw*

(0.98) (0.95) 2.95 (0.86)* (1.32) 3.77 (0.99) (0.88) 3.85 (0.89) 4.15 (1.16)

French

MLUm*

3.17 4.16 4.29 4.54

MLUw

MLUm

4.19 (0.97) 5.45 (0.99)* 4.54 (0.93) 5.85 (1.04) 4.41 (1.14) 5.82 (1.01) 3.90 (1.30) 4.55 (1.16)

(1.33) (1.21) (1.2) (1.60)

* Significant group difference: group FF differed significantly from groups EE and ME, but not from group EF.

ENGLISH SAMPLES

Bilingual exposure and morphosyntax 1

1

0.8

0.8

EE EF

0.4 0.2

EF 0.4

FF

0.2 0 contr v tense 3rd -s

ing

nouns

contr v tense

3rd -s

ing

nouns

1

1 FRENCH SAMPLES

EE

0.6

FF

0

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

ME

ME 0.6

105

0.8

0.8

MF FF

0.6

MF FF

0.6

EF

EF 0.4

EE

0.4

EE

0.2

0.2 0

0 person tense mood

gender plurals

YOUNGER

person tense

mood gender plurals OLDER

Figure 2. Accuracy of morphological production (percentage correct in obligaory contexts) in English (top panels) and French (bottom panels) for the Younger group (left panels) and Older groups (right panels), for each of the exposure groups.

The ME group achieved accuracy ranging from 72–90% on contracted verb forms, past tense, progressive –ing and noun inflection, but was at 46% for the third person –s. Groups with less English exposure had lower scores in general, although with some variability, with performance as low as 8% for past tense use by the FF group. In spite of this variability, no significant exposure group differences were found for the use of the progressive –ing (p ⫽ 0.104), the third person –s (p ⫽ 0.618), contracted verb forms (p ⫽ 0.149) or noun morphemes (p ⫽ 0.259). A significant group difference emerged for tense marking (F(3,22) ⫽ 3.836, p ⫽ 0.024, η2 ⫽ 0.34), with post-hoc tests showing that the FF group performed more poorly than group ME (p ⫽ 0.024).

use of mood other than the indicative were too few to permit statistical analyses—mood was used correctly where it occurred. For the Younger group, means for correct use of verb person and verb tense varied across groups only from 94–100%. Group means for plurals varied from 78–90% and gender marking from 67–94%. The lowest means were those of the EE group. No significant effect of exposure group was found for any morpheme (tense: p ⫽ 0.748, person: p ⫽ 0.271, noun plural: p ⫽ 0.088, gender: p ⫽ 0.143). Occurrences of mood other than the indicative were too few to compare groups (accuracy was 100% where the use of mood occurred).

French samples. For the Older group, group mean accuracy ranged from 96–98% and 91–98%, respectively, for verb person and tense. Accuracy of plurals ranged from 80–96% and that of gender marking from 75–94%. Significant differences between exposure groups were found for gender marking (F(3,51) ⫽ 6.227, p ⫽ 0.001, η2 ⫽ 0.27), with post-hoc tests revealing the EE group performed significantly more poorly than each of the other groups (p ⫽ 0.001), with no other group differences. The exposure group effect approached significance for tense (p ⫽ 0.051). No significant group differences were found for plurals (p ⫽ 0.106) or verb person marking (p ⫽ 0.483). Instances of

Group means for diversity for each of the exposure groups of Older and Younger groups in English and French are displayed in Table III, for morphemes appearing at least once (DivM) and for morphemes with evidence of productive use (DivPM).

Diversity of correctly used grammatical morphology

English samples. For the Older group, a significant group difference was found for basic diversity of use of morphology (DivM) (F(3,52) ⫽ 4.243, p ⫽ 0.009, η2 ⫽ 0.20). Post-hoc tests revealed that the FF group performed significantly lower than both the ME and EF groups. The difference between the FF and MF groups and the FF and EE groups approached significance (0.064 and

106

E. Thordardottir Table III. Group means and (SD) for diversity of morphemes used at least once by a member of the group (DivM) and diversity of morphemes used productively (DivPM), appearing with two or more different words in samples of half or more of the children in the group. 3-year-olds English DivM

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

MF FF EF EE ME

3.17 4.85 7.50 6.15

(2.11) (2.91) (2.46) (2.11)

French

DivPM

1.33 1.92 3.30 3.00

5-year-olds

DivM

English

DivPM

8.00 (2.04) 5.57 (2.16) 6.60 (1.14) 3.20 (1.75) 8.11 (3.10) 5.22 (1.77) 4.40 (2.41) 3.00 (1.73)

DivM

DivPM

(2.62) (1.64) 6.21 (2.76) 2.95 (2.33) 9.09 (1.36) 4.27 (2.00) 8.38 (2.25) 4.15 8.46 (2.90) 4.53

0.078, respectively), with the FF group in each case performing worse. No significant group difference was found for productive use of morphology (DivPM) for the Older group (0.255). The Younger group also had a significant group difference for basic diversity of morphemes (F(3,38) ⫽ 3.470, p ⫽ 0.025, η2 ⫽ 0.22), with post-hoc tests showing that the FF group performed less well than the EE group. The group effect was not significant for productive use of English morphology in the Younger group (p ⫽ 0.094). French samples. For Older children, a significant group effect was found for the basic diversity of use of morphology (DivM) (F(3.53) ⫽ 4.364, p ⫽ 0.008, η2 ⫽ 0.20). Post-hoc tests revealed that the EE group used significantly fewer different morphemes than each of the other groups. A significant exposure group effect was also found for the diversity of productive morphology (DivPM) (F(3,53) ⫽ 4.548, p ⫽ 0.007, η2 ⫽ 0.20), with post-hoc tests showing that the EE group used fewer morphemes productively than each of the other bilingual groups (EF and FF) but did not differ significantly from the MF group. For the younger children, a significant group effect was found for basic diversity of French morphemes (F(2,29) ⫽ 3.475, p ⫽ 0.029, η2 ⫽ 0.26), with post-hoc tests revealing that the EE group produced fewer different morphemes than the MF and EF groups. No group difference was evident for productive use of French morphemes in the Younger group (p ⫽ 0.094). Cross-language comparisons Within-subject cross-language comparisons of the bilingual children’s performance in English and French could not be submitted to statistical testing as they did not involve identical coding procedures given structural differences between the languages. The descriptive data show that MLUm was overall higher in French than English, such that the monolingual French children produced a higher MLUm than the monolingual English children and the EF group had a substantially higher MLUm in French

French DivM

DivPM

9.83 (2.20) 7.44 (2.22) 10.00 (1.37) 8.00 (1.10) 9.80 (1.48) 8.40 (2.03) 7.45 (2.84) 8.00 (3.62)

(1.69) (1.43) (1.43) (1.33)

than in English. This was expected based on previous normative data on monolingual French-speaking and English-speaking children (Elin Thordardottir, 2005) and reflects both the fact that French is more highly inflected than English and that the coding procedure targets this greater morphological richness. However, this difference does not mean that these children were ‘better’ at French than English. Similarly, the data show that the accuracy of use of grammatical morphology was overall greater in French (Figure 2), as was morphological diversity (Figures 3 and 4). These inherent structural differences between the languages preclude direct comparisons of MLU, accuracy or diversity scores across the languages. A first question that can be asked regarding how the bilingual children are affected by this difference in structure is how their developmental pattern in each language compares to that of monolingual children of that language. The use of particular English morphemes by the different exposure groups is shown in Figure 3 and the use of French morphemes in Figure 4. In these figures, progressively darker shades represent a higher degree of mastery of the morpheme in question by the group: (1) not used at all (white), (2) used at least once by at least one child in the group (lighter grey), (3) used at least once by half or more of the children in the group (darker grey) and (4) used with two or more different lexical items by half or more of the children in the group (black). In English, the morphemes mastered the earliest are the contracted verb form –s, progressive –ing and plural –s. Past tenses, first the irregular, then the regular appear, but neither is used productively by any of the Younger or Older groups. In the Older group, the third person –s is used productively by all exposure groups except the FF group, appearing in less than half of the samples of that group. Among the Younger children, only the EE group used the third person –s productively; however, more than half of children in all other Younger exposure groups did use it at least once. Later appearing morphemes are the possessive and plural –s and contracted verb form (-‘ve). Although bilingual children, in particular those with low exposure to English (FF) lag behind other exposure groups,

Bilingual exposure and morphosyntax

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Morpheme Contracted verb –s Progressive –ing Plural –s Third person –s Regular past –ed Irregular past Contracted verb ‘re Contracted verb ‘m Contracted verb ‘ll Negation ‘t Negation ‘nt Possessive –s Possessive+plural -s Contracted verb ‘ve

ME Younger

Older

EE Younger

Legend: No use of morpheme Used at least once Used by half or more of group Used with 2 words by half or more of group

Age 3

Older

EF Younger

Older

FF Younger

107

Older

Age 5

Figure 3. Types of morphemes used in English samples of the 3- and 5-year olds.

were used by more than half of the children in some of the 3-year-old groups. Again, the sequence of acquisition of French is similar in monolingual and bilingual children. This descriptive analysis, therefore, indicates that the sequence of acquisition in each language is similar across monolingual and bilingual children.Within each language, however, the rate of acquisition is influenced by how much time has been spent in that language. This was evidenced in the curve estimation analyses for morphological diversity as a function of previous exposure. In the descriptive analyses in Figures 3 and 4, this is particularly evident for the FF and EE groups, who do more poorly than other groups in English and French, respectively (the language to which they have been exposed less). The

the sequence of acquisition of English morphemes appears to be similar across monolingual and bilingual children. In the French samples, the morphemes mastered by the greatest number of groups were gender marking of adjectives and pronouns, verb person and plurals. Other morphemes mastered by many groups included the past tenses passé composé (compound past) and imparfait (imperfect), and futur proche (periphrastic future). Other tenses that appeared sporadically (with use by more than half of the children in some Older exposure groups) were the futur simple (simple future), passé simple (simple past) and plus-que-parfait (pluperfect). Some use of the subjunctive and conditional moods began before use of these later appearing tenses and they MF 3 years

5 years

FF 3 years

5 years

EF 3 years

5 years

EE 3 years

5 years

Gender of adj. Verb person Plural of noun Passé composé Gender of pron. Plural of adj. Imparfait Futur proche Imperative Plural of pron. Subjunctive Conditional Futur simple Passé simple Plus-que-parfait Legend: No use of morpheme Used at least once Used by half or more of group Used with 2 words by half or more of group

Age 3 Age 5

Figure 4. Types of morphemes used in French samples of the 3- and-5-year-olds (darker colours denote a greater degree of productivity).

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

108

E. Thordardottir

three bilingual groups (EE, EF and FF) appear fairly similar amongst themselves in terms of which morphemes they use in each language. For both French and English, the Younger children with the least exposure did the most poorly and the Older children with relatively more exposure did the best. Overall, therefore, the conclusion is that the bilingual children acquired the grammatical morphology of each language in a sequence comparable to that of monolingual learners of each language, but that the pace of development of each language was associated with how much time they had spent with speakers of that language. How does learning two languages that differ in structure manifest in individual bilingual children? Again, direct comparisons are hard to make across the languages. However, we can match children in terms of how much exposure they have had to each language and how they compare to monolingual children of each language and make a qualitative comparison of their grammatical proficiency. Children in the Older EF group have had equal exposure to English and French. They do not differ significantly from monolingual peers in either language in terms of MLUw, MLUm, morphological diversity or morphological accuracy (see previous section) and can, thus, be said to be highly (and relatively equally) proficient for their age in French and English. Qualitative examination of the morpheme diversity of this group (Figures 3 and 4) shows that this group has not reached productive use of the regular or irregular past tense in English. However, this group is using several French tenses in addition to the present (passé composé, imparfait, futur proche). Therefore, even though these children can be said to be equally proficient in French and English, they differ sharply in the kinds of grammatical structures they use when speaking French or when speaking English. Children in the Older EE group have received significantly more exposure to English than French. They score comparably to monolingual English speakers in the diversity of English morphology used, but significantly lower than monolingual French-speakers in the use of French morphology. However, similar to the children with equal exposure to both languages (EF), these children do not, as a group, show productive use of the regular or irregular English past tenses (neither do the monolingual English children), but do use the French passé composé and futur proche productively. Therefore, they use a greater diversity of tenses in the language to which they have been exposed less—a finding that is counter-intuitive and easily misinterpreted if the appropriate normative data on each language are not considered. The greater use of morphology in French than English within the same bilingual children is even more striking in the Older children with greater exposure to French than English (FF group in Figures 3 and 4). This group of children scores significantly more poorly in English than monolingual

English speakers, but does not differ significantly from monolingual French speakers. Qualitative analysis reveals that this group of children does not display productive use of the English third person –s, regular or irregular past tenses and uses only some contracted verb forms productively. This group of 5-year-olds, therefore, displays a pattern of verb morphology in their English samples consistent with an Extended Optional Infinitive Stage. When found in monolingual 5-year-old speakers of English, such a pattern is viewed as a strong indication of the presence of Primary Language Impairment (Rice & Wexler, 1996). At the same time as they use verb finiteness only sporadically in English, these same bilingual children use person and tense marking highly accurately and with greater diversity in their French. Example sentences for an individual child in the Older FF group illustrate this. This female participant obtained an MLUm of 5.50 in French and 3.63 in English. In her French sample, she marked verbs consistently correctly for person, and used, in addition to the present tense, the imparfait and futur proche, with sentences such as “Je le savais que ça rentrait” [I knew it that this fit] and “Ah moi, je vais aller puis lui mettre ça” [Oh, me, I’m going to go and put this on her]. In English, she used the irregular past tense once, but never used the regular past tense and used the third person –s and contracted verb forms in some sentences, but not in other sentences, producing sentences such as “She all ready”, “Dora need to go” and “She needs to wear her coat”. Similar patterns of unequal proficiency across the two languages are seen for the Younger children, although they are more clearly evident in the Older group. Individual analyses Inspection of individual data was undertaken to ascertain that group differences, in particular those showing very unequal performance in French and English by bilingual groups, were not an artifact of or accentuated by the data of children who produced a sample in one language only (perhaps because their other language was very weak). This analysis focused on verb inflections because their production differed sharply between the languages. In the Older EF group, three of eight children with samples in both languages produced no regular past tense in English, whereas they all used two different French tenses other than the present productively. In the Older EE group, two children never used the third person –s, whereas these same children consistently used person marking in French and no child in this group did not use person marking in French. Three children in this group never used the irregular past in English, but used two or more different tenses in addition to the present in French. Among 13 children in the Older FF group with samples in both languages, eight children used no third person –s

Bilingual exposure and morphosyntax

marking in their English, whereas each of them used person marking productively in French. Six children in this group used no regular past tense in English and five children used no irregular past (with three children using neither). In their French samples all of these children used multiple tenses other than the present. Similar patterns were observed among the Younger children.

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Discussion The present study set out to quantify the effect of input on simultaneous grammatical development in a more fine grained manner than has been done previously, by examining the grammatical development of groups of bilingual children that differ in how much they have been exposed to each of their languages and by directly linking the scores obtained on several types of grammatical development to the amount of previous exposure received in each language. A major finding of this study is not only that grammatical development is significantly dependent on input, but, importantly, that the association between amount of input and grammatical development is comparable in strength to that seen for vocabulary. The comparison in skill between bilingual and monolingual children is also comparable to that seen for vocabulary (Elin Thordardottir, 2011): children having spent at least 50% of their time with speakers of a given language did not perform significantly differently than monolingual peers of that language group; those having spent less time did score significantly lower than monolingual speakers. The results show, moreover, that simultaneously bilingual children follow highly language-specific patterns in the acquisition of morphosyntax in their two languages. The sequence of acquisition of grammatical morphemes and error patterns differed across the languages, resembling those of monolingual speakers of each language, whereas the rate of acquisition and frequency of errors was impacted by amount of previous exposure. MLU was significantly associated with amount of input in English for both age groups, when input was considered as a continuous variable—the strength of the relationship was comparable to that previously documented for vocabulary (Elin Thordardottir, 2011). MLU in French, however, demonstrated greater variability such that, even though a fitted curve showed a systematic increase in MLU with increased exposure to French, the association was not statistically significant. When the children were divided into discrete exposure groups, significant differences in MLU were evident for English, but less so for French. In spite of this difference for MLU across the languages, the diversity of morphemes appearing in the samples was significantly associated with previous exposure in both languages and in both age groups. In this respect, the mere use of

109

different types of morphemes was more strongly related to previous exposure than was the diversity of morphemes with evidence of productive use. Similarly to diversity, accuracy of use of grammatical morphology was also sensitive to exposure effects in both languages, as evidenced by comparisons of exposure groups. For accuracy, significant group differences between exposure groups were found for Older children only—in each language affecting the group with the lowest amount of previous exposure to that language (with the EE and FF groups scoring significantly lower than monolingual children in French and English, respectively). A possible reason why significant differences did not emerge for the Younger children for accuracy may be the fact that the sample size was smaller, diminishing statistical power in comparison to the Older group. A second possible contributing factor may be that not only did the Younger children have, overall, a lower morphological diversity than the Older children, it is also reasonable to assume that they used these morphemes in less complex linguistic contexts. The Younger group may, therefore, have used a relatively larger proportion of morphemes by rote, still tied to relatively simple linguistic contexts. These factors may serve to make the Younger children more similar amongst themselves in their use of grammatical morphology compared to the Older children, many of whom were quite advanced in the use of morphology and others significantly less advanced. For diversity of use of grammatical morphology, whether measured as sporadic or productive use, significant group differences were evident even among the Younger children in both languages, such that children with the lowest amount of exposure to either language lagged significantly behind the other groups in that language. It does emerge that, of the measures considered in this study, the diversity of correct use of morphology appears to be the one most sensitive to the effect of amount of input. The descriptive data on the use of morphology appearing in Figures 3 and 4 show that the sequence of acquisition of grammatical morphemes is different across the two languages, but in each language the sequence for bilingual children is similar to that characteristic of monolingual development of that language. This finding indicates that the two languages develop, within the same child, according to separate trajectories, consistent with previous longitudinal studies of simultaneous bilingual children of pre-school age (de Houwer, 2005) and also documented in Paradis and Genesee (1996) and Gutiérrez-Clellen, SimonCereijido, and Wagner (2008). The finding that bilingual grammatical development varies systematically with amount of bilingual input is in line with several recent studies (Blom, 2010; Elin Thordardottir et al., 2006; Hoff et al., 2012), but at odds with many previous reports that have suggested that bilingual children attain global linguistic milestones at comparable times as

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

110

E. Thordardottir

monolingual children and achieve MLUs within the normal monolingual range (e.g. Bathia & Ritchie, 1999; Genesee & Nicoladis, 2007; Paradis & Genesee, 1996). Discrepancies in findings are likely to be related to the types of bilingual participants included in studies—many previous studies focused primarily on children with similar amounts of input to both languages (de Houwer, 2005; Paradis & Genesee, 1996)—and to how closely the previous bilingual exposure patterns were documented and taken into account. Further, the present findings suggest another source of discrepant findings, namely that not all grammatical measures are equally sensitive to the effects of amount of input. MLU is a widely used measure, both in clinical and research settings, including research that has examined the effect of bilingual input, with mixed results (Blom, 2010; Paradis & Genesee, 1996). The present study found MLU to be sensitive to the effects of bilingual input in English, but less so in French. In English MLU increased systematically across exposure groups whereas, for French, the pattern was less systematic with no significant group differences (although the children with least exposure to French did score the lowest). These differences may be related to the greater morphological complexity of French compared to English, and differences in how the coding procedures capture the productive use of morphemes. The very unequal levels of grammatical development in children with unequal levels of exposure to the two languages is a striking finding of this study. De Houwer (2005) had predicted that such unequal grammatical development was possible hypothetically, but that this remained to be verified as most longitudinal case studies providing detailed accounts of morphosyntactic development included children with relatively equal exposure to both languages. The present findings show that such unequal developmental levels are not only possible, but indeed typical in children having received markedly unequal exposure to their two languages. The children in this study showed evidence of earlier mastery of grammatical morphology in French than English, with individual children using verb person marking accurately in French while having high rates of optional verb finiteness marking in English and also using more varied tense marking in French than English. This was true particularly of children having received more exposure to French than English, but also of children having received equal exposure to both languages and even of children having received more exposure to English. As pointed out in earlier sections, this crosslinguistic difference is also seen in the language production of the monolingual children. Importantly from a diagnostic point of view, many of the bilingual 5-year-old children were displaying a pattern in English consistent with an Extended Optional Infinitive stage, a strong clinical marker for Primary Language Impairment in monolingual English speaking

children in this age range (EOI, Rice & Wexler, 1996). And yet, none of the children in this study had previous diagnoses or reported concerns about their language or general development. We return to the theoretical implications of the EOI pattern in the bilingual children in a later section. Theoretical implications The effect of input on grammatical acquisition touches on fundamental questions regarding how children come to know language. No account of language acquisition denies that, in order to acquire language, children must be exposed to language. However, accounts differ in how much input is thought to be required and whether this differs across domains of language. When different amounts of language are thought to be needed across domains, it is assumed that grammatical development might need less input than vocabulary development, not the opposite (cf. Pearson et al., 1997). Nativist accounts that assume triggering mechanisms based on innate parameters assume that only small amounts of input are required for grammatical development because it is guided by innate Universal Grammar (UG) parameters (e.g. Goodluck, 1991; Pinker, 1991). Paradis and Genesee (1996) concluded that the finding that bilingual children obtained within-normalrange MLUs compared to monolinguals in spite of much less exposure supported the operation of innate triggering mechanisms. Our results are in line with theirs in that our group of children with equal exposure to both languages did score within the normal range for monolingual children—presumably, this was the exposure pattern of the children in the Paradis and Genesee study, who were reported to have been raised according to the one parent–one language principle. However, two aspects of our data speak against the interpretation that grammatical development was accelerated by innate mechanisms: first, the children in our study achieved normal range performance with only 50% input not only in grammar, but also in vocabulary (Elin Thordardottir, 2011). Vocabulary is not purported to rely on innate mechanisms by any account of language acquisition. Secondly, our data on children with unequal exposure to each language clearly show that the rate of grammatical development is negatively impacted (slowed down significantly) by low levels of exposure. While the question of the existence and nature of innate triggering mechanisms is an elusive one, a fair conclusion seems to be that if such mechanisms are at work, the amount of input required to activate them is not trivial. The finding that vocabulary and grammatical development are equally strongly affected by input is consistent with the view that both are learned from input in a similar way, consistent with emergentist and usage-based accounts (MacWhinney, 1999; Lieven & Tomasello, 2008). One possibility is that

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Bilingual exposure and morphosyntax

vocabulary and grammatical development each depend on the storage of vocabulary and grammatical items directly from input. Another possibility is that the effect of input on grammatical development is mediated by lexical development or the storage from input of a critical mass of lexical items that allow the extraction of rules (e.g. Marchman et al., 2004). Critical mass effects in bilingual children have been described as involving threshold effects (Gathercole, 2007; Marchman et al., 2004; Pearson, 2007). The findings of the present study are more consistent with gradual increases in input resulting in gradual increases in performance, although the attainment of monolingual performance levels at only ∼ 50% exposure can be seen as a threshold effect; however, one that applies equally to vocabulary and grammar (Elin Thordardottir, 2011). One possible reason for this study finding a gradual effect rather than more abrupt threshold effects may be related to the fact that this study employed overall summary measures of grammatical development. MLU is a global index of language development, influenced by advances in syntax, morphosyntax as well as lexical development. Similarly, the overall diversity and accuracy of use of morphology each encompasses the mastery of a number of different linguistic systematicities over the range of MLU levels covered. In contrast, Gathercole (2007) examined performance on experimental tasks, each targeting the mastery of a specific rule. Another reason for the discrepancy in findings may be the age range of the children and that at least some of the children in Gathercole’s study were sequential rather than simultaneous bilinguals. In sum, the finding that vocabulary and grammar are similarly affected by experience in simultaneous learners up to age 5 years suggests that vocabulary and grammatical development tend to go hand in hand in each language, but do not allow firm conclusions to be drawn as to whether this is because both depend directly on experience or whether grammatical development is mediated by vocabulary learning. Marchman et al. (2004) discuss alternative views to the critical mass view, including grammatical knowledge being built up on a case basis at first and later being encoded in abstract form, as well as the possibility that grammatical learning may influence vocabulary learning. The present findings appear to be compatible with any of these possibilities and do not provide a means to distinguish between them. The striking extent to which the development of morphosyntax by the bilingual children in this study followed language-specific patterns suggests very clearly that the grammatical development of each language of the same bilingual child is dependent on experience in that language specifically, rather than on total language experience or chronological age. A strong separation of morphological development in the two languages is consistent with previous findings such as those of De Houwer (2005); however,

111

this study adds the documentation of groups of children having received unequal exposure to each language and the observation that they exhibit similarly markedly uneven levels of grammatical performance in each language. This is evident across the measures obtained in the lower scores of children with low exposure levels, but perhaps most astonishingly in the Extended Optional Infinitive stage documented in the English production of typically developing 5-year-old children with low exposure to English. The fact that individual children, including those with equal exposure to both languages, as well as those with more exposure to French, demonstrated such very uneven use of finiteness across the two languages does not appear to support the claim that linguistic features that are much more salient and transparent in one language trigger the use of the same structure in the other language (cf. Roeper, 2011). Of course, it is hard to entirely exclude that such effects may have occurred to some extent. However, the consistent pattern across children of low finiteness use in English in children using finiteness highly frequently and correctly in French suggests that, even if some degree of interaction may occur between the languages, it does not appear to have an appreciable impact on the overall pace of morphosyntactic development. In this respect, our study concurs with that of Paradis and Genesee (1996) as well as those of Gutierrez-Clellen et al. (2008), who included monolingual children and second language learners with and without language impairment. The documentation of an EOI stage in 5-year-old bilingual children without language impairment has potential implications for the underlying nature of Primary Language Impairment. Previous studies have shown that clinical markers that distinguish between monolingual children with and without language impairment cannot be assumed to identify language impairment as accurately in bilingual children because typically-developing bilingual children and second language learners have been reported to make errors that resemble those of same-age monolingual children with language impairment, but that differ from same-age monolingual children with typical development (Grüter, 2005; Paradis, 2005). The present findings shed new light on this issue by showing that only some of the bilingual children make errors characteristic of language impairment in monolingual children and that this propensity is predictable from their background, namely amount of exposure to that language. According to the Extended Optional Infinitive Account (Rice & Wexler, 1996), the error patterns of children who remain in this stage for an extended period of time indicate that they have an adequate representation of finiteness marking, but that they consider such marking optional. This conclusion rests on the observation that, when these children mark verbs for finiteness, they do so correctly, however, in a large proportion of obligatory contexts, they simply omit

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

112

E. Thordardottir

the inflection. The EOI account does not offer further explanations as to why children regard finiteness as optional. This study provides a novel insight into why some typically-developing children take longer than others to exit the EOI stage by documenting the children without Primary Language Impairment but with low exposure to English remain in this stage at age 5 years, whereas children with a high level of exposure have exited this stage by age 5 years. This shows that the EOI phenomenon can occur in 5-year-old children who do not have language impairment and is, thus, not necessarily related specifically to a grammatical deficit found only in or caused by language impairment. Strikingly also, the same children who remain in this stage in English at age 5 years are using finiteness correctly in French. Therefore, the problem in their English does not seem to be a generalized lack of ability to view finiteness as obligatory rather than optional at an abstract level. It appears, based on these findings, that the attainment of correct use of English finiteness requires a great amount of exposure to English. It follows that it can be speculated that the EOI phenomenon in English-speaking children with Primary Language Impairment may be the result of low exposure to English in the sense that children with Primary Language Impairment may make poorer use of the input they receive than do children with typical development. This would be consistent with accounts of Primary Language Impairment that associate it with poor processing abilities (see review in Montgomery, Magimairaj, & Finney, 2010). This possibility requires further study to be confirmed or refuted. An alternative explanation is that similar surface patterns can result in children with primary language impairment and in bilingual children, but for different reasons. Further study should also consider the full spectrum of EOI errors, including auxiliaries. Implications for the language assessment of bilingual children To interpret the language assessment scores of bilingual children in a meaningful way, it is essential to establish what range of scores is indicative of typical development vs below the normal range. This is difficult to achieve for bilingual children because of the great diversity in their language learning circumstances, leading to the lack of adequate bilingual normative reference bases. The results of this study provide more detailed information than has been available previously on the level of performance that should be expected of simultaneously bilingual children with different histories of language experience. The descriptive data presented in this study on MLU, morphological accuracy and diversity can be used to more meaningfully interpret the results of individual children, taking into account their particular exposure histories and to set clear expectations of

performance for individual children. In addition, the descriptive data on the productive use of individual grammatical morphemes each language presented in Figures 3 and 4 can be used, as is frequently done clinically for monolingual children, to ascertain whether morphological use is commensurate with MLU in each language, as well as to help select and prioritize treatment goals. The results show that, depending on their previous experiences, bilingual children may score within the same range in grammar as monolingual children in both languages or in one language only (in general, the children in this study did not score below the normal monolingual range in both languages—therefore, such a finding may be indicative of language impairment in children who have had adequate opportunity for interaction). Further, although language scores such as MLU cannot be compared directly across languages, qualitative analyses revealed that the diversity of morphology and particular morphemes mastered at a given point in time by individual children can be expected to vary greatly across the two languages of a bilingual child, even for those with equal exposure to both languages, and more so for those with unequal exposure histories to the point that bilingual children may exhibit patterns consistent with language impairment in monolingual children. This means that bilingual children deemed as equally proficient in both languages should not be expected to have mastered the same linguistic structures in both languages. Further, the structures that are prone to errors vary across the languages. It follows that the manifestation of language impairment should be expected to vary across the two languages as well, given the different manifestation of language impairment in monolingual children speaking English and French, respectively (see, e.g. Elin Thordardottir, & Namazi, 2007). A recent case study of a bilingual child with primary language impairment speaking French and Greek showed different error patterns in the two languages, in each case resembling those of younger monolingual children speaking these languages (Stavrakaki, Chrysomallis, & Petraki, 2011). It is clear that, for bilingual children, whether using quantitative normreferenced measures or qualitative clinical markers, the children’s exposure history must be taken into account. The data presented in this study are directly applicable to clinical settings, allowing clinicians to interpret the scores of individual children with reference to bilingual children with similar exposure histories, and to make sense of error patterns in each language and whether they are indicative of impairment or rather of low levels of exposure. Conclusions This study has documented the grammatical development in spontaneous production in both French and English of bilingual children who vary in how much exposure they have had to each language, but

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

Bilingual exposure and morphosyntax

who are similar in age, SES, non-verbal cognition and language status, in that both languages have majority status. The study documents the production of both languages in a fairly large sample of children, documenting typical patterns of bilingual acquisition involving these two languages. The findings show clearly that amount of input impacts grammatical development and that, in fact, each language is learned in a highly language-specific manner, mirroring the amount of exposure received in this language. The effect of input on grammatical development is similar to that seen previously for vocabulary, suggesting that both vocabulary and grammar depend to a similar extent on input. This study involved a particular language combination—French and English. While this permitted the comparison of two languages with a majority status, allowing the control of many relevant variables, this combination represents two fairly closely related languages being learned in a favourable learning environment. Different results may be found in contexts where one of the languages has a minority status. Another limitation of this study is that the sample of 3-year-old children was limited in size compared to the 5-yearolds. Future studies examining other language combinations are required to establish typical developmental rates in other bilingual populations. It is possible that the relationship between input and performance may be less clear or different in bilingual acquisition involving a minority language, as language status is expected to exert a strong influence as well. Importantly, more research is required documenting typical patterns of bilingual acquisition across groups of children whose background characteristics are well documented.

Acknowledgments This study was supported by a standard research grant from the Social Sciences and Humanities Research Council Of Canada (SSHRC) awarded to the author. Thanks are extended to the children and parents who gave of their time to contribute to the study. Thanks are extended to Marianne Paul and Myrto Brandeker for assistance with the preparation of this article. Declaration of interest: The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper. References Bassano, D., Maillochon, I., Klampfer, S., & Dressler, W. (2001). L’acquisition de la morphologie verbal en français et en allemande autrichien:II. L’épreuve des faits. Enfance, 53, 117–148. Bathia, T., & Ritchie, W. (1999). The bilingual child: Some issues and perspectives. In W. Ritchie, & T. Bathia (Eds.), Handbook of child language acquisition (pp. 569–646). San Diego, CA: Academic Press.

113

Bedore, L., & Peña, E. (2008). Assessment of bilingual children for identification of language impairment: Current findings and implications for practice. International Journal of Bilingual Education and Bilingualism, 11, 1–29. Berent, I., Pinker, S., & Shimron, J. (2002). The nature of regularity and irregularity: Evidence from Hebrew nominal inflection. Journal of Psycholinguistic Research, 31, 459–502. Blom, E. (2010). Effects of input on the early grammatical development of bilingual children. International Journal of Bilingualism, 14, 422–446. Caselli, C., Casadio, P., & Bates, E. (1999). A comparison of the transition from first words to grammar in Italian. Journal of Child Language, 26, 69–111. Chomsky, N. (2011). Language and other cognitive systems: What is special about language? Language Learning and Development, 7, 263–278. Conboy, B., & Thal, D. (2006). Ties between the lexical and grammar: cross-sectional and longitudinal studies of bilingual toddlers. Child Development, 77, 712–735. De Houwer, A. (2005). Early bilingual acquisition: Focus on morphosyntax and the Separate Development Hypothesis. In J. Kroll, & A. de Groot (Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 30–48). New York: Oxford University Press. Döpke, S. (2000). The interplay between language specific development and cross-linguistic influence. In S. Döpke (Ed.), Crosslinguistic structure in simultaneous bilingualism (pp. 79–103). Amsterdam: John Benjamins Publishing. Elin Thordardottir. (2005). Early lexical and syntactic development in Quebec French and English: Implications for crosslinguistic and bilingual assessment. International Journal of Language and Communication Disorders, 40, 243–278. Elin Thordardottir. (2011). The relationship between amount of bilingual exposure and vocabulary development. International Journal of Bilingualism, 14, 426–445. Elin Thordardottir, & Brandeker, M. (2013). The effect of bilingual exposure versus language impairment on nonword repetition and sentence imitation scores. Journal of Communication Disorders, 46, 1–16. Elin Thordardottir, & Namazi, M. (2007). Specific language impairment in French-speaking children: Beyond grammatical morphology. Journal of Speech, Language, and Hearing Research, 50, 698–715. Elin Thordardottir, Ellis Weismer, S., & Evans, J. (2002). Continuity in lexical and morphological development in Icelandic and English-speaking 2-year-olds. First Language, 22, 3–28. Elin Thordardottir, Kehayia, E., Lessard, N., Sutton, A., & Trudeau, N. (2010). Typical performance on tests of language knowledge and language processing of French-speaking 5-year-olds. Canadian Journal of Speech Language Pathology and Audiology, 34, 5–16. Elin Thordardottir, Rothenberg, A., Rivard, M.-E., & Naves, R. (2006). Bilingual assessment: Can overall proficiency be estimated from separate measurement of two languages? Journal of Multilingual Communication Disorders, 4, 1–21. Fenson, L., Dale, P., Reznick, S., Thal, D., Bates, E., Hartung, J., et al. (1993). MacArthur Communicative Development Inventories: User’s Guide and Technical Manual. San Diego, CA: Singular Publishing. Gathercole, V. (2007). Miami and North Wales; so far and yet so near: A constructivis account of morphosyntactic development in bilingual children. The International Journal of Bilingual Education and Bilingualism, 10, 224–247. Genesee, F., & Nicoladis, E. (2007). Bilingual first language acquisition. In E. Hoff , & M. Shatz (Eds.), Blackwell handbook of language development. Malden, MA: Blackwell Publishing. Girbau, D., & Schwartz, R. G. (2008). Phonological working memory in Spanish-English children with and without specific language impairment. Journal of Communication Disorders, 41, 124–145. Goodluck, H. (1991). Language acquisition: A linguistic introduction. Oxford, UK: Basil Blackwell Ltd.

Downloaded by [Florida Atlantic University] at 02:07 06 November 2015

114

E. Thordardottir

Grüter, T. (2005). Comprehension and production of French object clitics by child second language learners and children with Specific Language Impairment. Applied Psycholinguistics, 26, 363–291. Gutierrez-Clellen, V., Simon-Cereijido, & Wagner, C. (2008). Bilingual children with language impairment: A comparison with monolingual and second language learners. Applied Psycholinguistics, 29, 3–19. Heilmann, J., Miller, J., Iglesias, A., Fabiano-Smith, L., Nockerts, A., & Andriacchi, L. (2008). Narrative transcription accuracy and reliability in two languages. Topics in Language Disorders, 28, 178–188. Hoff , E., Core, C., Place, S., Rumiche, R., Senor, M., & Parra, M. (2012). Dual language exposure and early bilingual development. Journal of Child Language, 39, 1–27. Huttenlocher, J., Haight, W., Bryk, A., Seltzer, M., & Lyons, T. (1991). Early vocabulary growth: Relation to language input and gender. Developmental Psychology, 27, 236–248. Leadholm, B., & Miller, J. (1992). Language sample analysis: The Wisconsin guide. Madison, WI: Wisconsin Department of Public Instruction. Legate, A., & Yang, C. (2007). Morphosyntactic learning and the development of tense. Language Acquisition, 14, 315–344. Liceras, J., Fernándes Fuertes, R., & Alba de la Fuente, A. (2011). Overt subjects and contracted verb form omission in the Spanish and English grammar of English-Spanish bilinguals: On the focus and directionality of interlinguistics influence. First Language, 32, 88–115. Lieven, E., & Tomasello, M. (2008). Children‘s first language acquisition form a usage-based perspective. In P. Robinson, & N. Ellis (Eds.), Handbook of cognitive linguistics and second language acquisition (pp. 216–236). New York: Routledge. MacWhinney, B., (Ed.). (1999). The emergence of language. Mahwah, NJ: Lawrence Erlbaum Associates. Marchman, V., & Bates, E. (1994). Continuity in lexical and morphological development: A test of the critical mass hypothesis. Journal of Child Language, 21, 339–366. Marchman, V., Martinez-Sussmann, C., & Dale, P. S. (2004). The language-specific nature of grammatical development: evidence from bilingual language learners. Developmental Science, 7, 212–224. Miller, J., Andriacchi, K., & Nockerts, A. (2011). Assessing language production using SALT software. Middleton, WI: SALT Software LLC. Montgomery, J., Magimairaj, B., & Finney, M. (2010). Working memory and Specific Language Impairment: An update on the relationship and perspectives on assessment and treatment. American Journal of Speech Language Pathology, 19, 78–94. Nicoladis, E., & Genesee, F. (1997). Language development in preschool bilingual children. The Canadian Journal of Speech Language Pathology and Audiology, 21, 258–270.

Nicoladis, E., & Marchak, K. (2011). La carte blanc or la carte blanche? Bilingual children’s acquisition of French adjective agreement. Language Learning, 61, 734–758. Nicoladis, E., Palmer, A., & Marentette, P. (2007). The role of type and token frequency in using past tense morphemes correctly. Developmental Science, 10, 237–254. Oller, K., Pearson, B., & Cobo-Lewis, A. (2007). Profile effects in early bilingual language and literacy. Applied Psycholinguistics, 28, 191–230. Paradis, J. (2005). Grammatical morphology in children learning English as a second languageÆ: Implications of similarities with Specific Language Impairment. Language, Speech and Hearing Services in Schools, 36, 172–187. Paradis, J., & Genesee, F. (1996). Syntactic acquisition in bilingual children: Autonomous or independent? Seminars in Second Language Acquisition, 18, 1–25. Paradis, J., & Navarro, S. (2003). Subject realization and cross-linguistic interference in the bilingual acquisition of Spanish and English: What is the role of input? Journal of Child Language, 30, 371–393. Paradis, J., Nicoladis, E., Crago, M., & Genesee, F. (2011). Bilingual children’s acquisition of the past tense: A usagebased approach. Journal of Child Language, 38, 554–578. Pearson, B. (2007). Social factors in childhood bilingualism in the United States. Applied Psycholinguistics, 28, 399–410. Pearson, B., Fernández, S., Lewedeg, V., & Oller, K. (1997). The relation of input factors to lexical learning by bilingual infants. Applied Psycholinguistics, 18, 41–58. Pinker, S. (1991). Rules of language. Science, 253, 530–535. Reynell, J. (1969). Reynell Developmental Language Scales (R.D.L.S.). Windsor, England: N.F.E.R. Publishing Company Ltd. Rice, M., & Wexler, K. (1996). Toward tense as a clinical marker of specific language impairment in English-speaking children. Journal of Speech and Hearing Research, 39, 850–863. Roeper, T. (2011). Minimalism and bilingualism how and why bilingualism could benefit children with SLI. Bilingualism: Language and Cognition, 15, 88–101. Roid, G., & Miller, L. (1997). Leiter International Performance Scale-Revised. Wood Dale, IL: Stoelting Co. Silva-Corvalán C., & Montanari, S. (2008). The acquisition of ser, estar (and be) by a Spanish-English bilingual child. Bilingualism, Language and Cognition, 11, 341–360. Stavrakaki, S., Chrysomallis, M.-A., & Petraki, E. (2011). Subject-verb agreement, object clitics and wh-questions in bilingual French-Greek SLI: The case study of a French-Greek speaking child with SLI. Clinical Linguistics and Phonetics, 25, 339–367. Yip, V., & Matthews, S. (2000). Syntactic transfer in a CantoneseEnglish bilingual child. Bilingualism, Language and Cognition, 3, 193–208.

The relationship between bilingual exposure and morphosyntactic development.

The study examined the effect of bilingual input on the grammatical development of bilingual children in comparison to monolingual peers...
1MB Sizes 1 Downloads 3 Views