Você está na página 1de 23

International http://ijb.sagepub.

com/ Bilingualism Journal of

Delayed acquisition of irregular inflectional morphology in Hebrew in early sequential bilingualism


Mila Schwartz, Ely Kozminsky and Mark Leikin International Journal of Bilingualism 2009 13: 501 DOI: 10.1177/1367006909353238 The online version of this article can be found at: http://ijb.sagepub.com/content/13/4/501

Published by:
http://www.sagepublications.com

Additional services and information for International Journal of Bilingualism can be found at: Email Alerts: http://ijb.sagepub.com/cgi/alerts Subscriptions: http://ijb.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://ijb.sagepub.com/content/13/4/501.refs.html

>> Version of Record - Mar 1, 2010 What is This?

Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition International Journal of Bilingualism Volume 13 Number 4 2009, 501522| 501

Delayed acquisition of irregular inflectional morphology in Hebrew in early sequential bilingualism


Mila Schwartz1, Ely Kozminsky2, Mark Leikin3
1

Oranim College of Education and University of Haifa, Israel University, Israel 3 University of Haifa, Israel
2 Ben-Gurion

Abstract

Key words

Acquisition of the irregular forms of inflectional morphology may be a bilinguals L2 challenge for bilingual students because of the possible effect of infrequent acquisition input. Focusing on irregular plural forms of languages such as Hebrew can contribute to better understanding how bilingual children cope with irregular noun anomalous morphological forms. The present study compares Russian Hebrew-speaking sequential bilingual children with Hebrew-speaking plurals monolingual children in their command on four measures of irregular forms of Hebrew plural nouns at two data collection points: the beginning multi-interval study of the second grade and the beginning of the third grade, at a time when the acquisition of these forms is still going on. Although results show that the bilingual children continued to be less accurate than their monolingual peers in producing the irregular forms at the second point of data collection, the medium-size effect (Cohen, 1992) was obtained only on one of four measures. Furthermore, the finding attests to the significant improvement of both groups in the course of one school year on all categories of irregular plural forms. It was also found that both groups acquire the irregular forms of the Hebrew plural noun system in a similar way and exhibit related patterns of developmental errors.

1 Introduction

The objective of the present multi-interval study is to shed light on how young Russian heritage speakers (sequential RussianHebrew bilinguals) cope with irregular noun pluralizing in Hebrew as a second language (L2), compared with native Hebrew-speaking monolingual children. Primary focus is on the link between the more salient and frequent input of irregular plural forms in Hebrew (L2), which seems to be provided by formal literacy and grammar acquisition in L2, and the bilinguals growing command of these forms.

Address for correspondence

Dr Mila Schwartz, Oranim College of Education; Edmond J. Safra Brain Research Center for the Study of Learning Disabilities, University of Haifa, Israel. [email: milasch@bgu.ac.il]
The International Journal of Bilingualism
Copyright 2009 the Author/s 2009, ISSN; Vol 13 (4): 501522; ID no 353238; DOI; 10.1177/1367006909353238 http://Ijb.sagepub.com

Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

502

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

Two main theoretical models that have been proposed to explain the acquisition of inflectional morphology, the single-route and the dual-route models, generally agree on the critical role of the frequency of input of irregular forms. The models are based primarily on past tense construction in English among monolingual children. Under the single-route approach, both regular and irregular aspects of inflection are derived in the same way from a single, integrated mechanism, and their accessibility is subject only to the frequency effects (Bybee, 1995, 2001; McClelland & Patterson, 2002). Within the framework of the single-route (usage-based) model, Bybee (1995, 2001) maintained that both regular and irregular verbs of the past tense enter the childs lexicon with a certain strength, based on their token frequency. Following this approach, McClelland and Patterson (2002) suggested that the two inflectional types emerge from a parallel distributed processing or connectionist framework in which cognitive processes are viewed as graded, probabilistic, interactive, context-sensitive and domain-general. The dual-route model (Pinker, 1991; Pinker & Ullman, 2002) maintains that regular and irregular forms are treated differently and that the regularirregular division is an epiphenomenon of the design of the human language faculty, which distinguishes between two systems, lexical and grammatical (Pinker & Ullman, 2002). Whereas irregular forms behave like words in the lexicon, and are acquired and stored like other words (but with grammatical features similar to those that the past tense contributed to lexical entries), regular forms are part of a grammatical system created by a symbolic rule. As a result, an irregular form that cannot be generated by a rule must be learned as part of the lexicon. Although the two approaches account differently for the presentation of regular inflectional forms in cognition, both stress that irregular forms are subject to frequency effects as they are stored in associative memory, and that their memorization depends on opportunities for learning trials. In addition, both approaches make similar predictions for cases in which children do not remember the target irregular form and therefore overregularize the -ed form because it is the most common pattern encountered (single-route model) or because it represents an over-generalized use of the regular rule (dual-route model). (The single-route approach predicts that children can also irregularize, e.g. produce brang instead of brought, by analogy with sing/sang based on the relatively high frequency of that irregular pattern). Several researchers have noted that the production of irregular forms of inflectional morphology in languages rich in irregular forms seems to be especially challenging for bilingual children during L2 acquisition (Gathercole, 2006; Nicoladis, Palmer, & Marentette, 2007; Paradis, Nicoladis, & Crago, 2007; Schelletter, 2007). Because these forms cannot be acquired by generalized rule and must be learned one by one as separate lexical units, their mastering may be related to how often the child hears them. It has been suggested, therefore, that initial delays may appear in bilingual childrens development of certain structures because their exposure in each language is only a portion of that of monolingual children (Gathercole, 2006; Nicoladis et al., 2007; Paradis et al., 2007). Gathercole (2006) suggested that as the children were gaining a critical mass of linguistic information, the gap between bilingual and monolingual children eventually narrows. This claim found support in a study that focused, among others, on the acquisition of grammatical gender in children speaking at home only Welsh, both Welsh and English, and only English. The children were aged 5, 7, and 9, and were educated either
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

503

in Welsh or WelshEnglish settings. Although irregular inflectional forms are not a characteristic of grammatical gender in Welsh, some singular feminine and masculine nouns show morpho-phonological mutations, feminine nouns change after the definite article, and adjectives and masculine nouns change after a possessive reference to antecedent nouns. For all these reasons gender marking in Welsh is complex and opaque. Gathercole showed that all the children were near ceiling on gender production of nonmutated forms, but on mutated forms, as expected, children from only-Welsh homes outperformed those from WelshEnglish and only-English homes. But the gap between the groups diminished as the children grew older and gradually gained the critical mass of linguistic information necessary for mastering these complex structures. Although in recent years there has been a growing interest in the acquisition of irregular inflectional forms in L2 among young bilinguals, relatively few studies have addressed this topic systematically and directly. Some studies focused on irregular verb production in English and French (Nicoladis et al., 2007; Paradis et al., 2007; Schelletter, 2007). Paradis et al. (2007) suggested that forms with non-transparent (opaque) structure create inflectional islands and predicted that both monolingual and bilingual children acquire them later than they do regular forms, bilinguals doing so less accurately. Paradis et al. compared 14 young English-dominant and 11 French-dominant bilinguals with their 12 French-speaking monolingual peers (ages 4 to 5.5) and the norming sample used for English past tense probe on past tense elicitation task in both English and French as L1/L2. The results showed that all groups were more accurate with regular than irregular verbs in both target languages. Overall, no delay was observed between bilinguals and monolinguals in either English or French on past tense usage. At the same time, English-dominant bilinguals outperformed French dominant bilinguals on irregular verbs, but their scores were lower than those of monolingual children. In French, as in English, significantly fewer errors were observed in all groups in regular than in irregular verb production. But unlike in English, there was no interaction with language dominance and no significant differences were found between the groups on irregular forms. The researchers suggested that the differences between English and French may be attributed to cross-linguistic dissimilarity between the languages in the token frequencies of regular and irregular verbs and their type (i.e. the number of different lexemes with which they have been used). This assumption received support from the analysis of longitudinal data for Frenchand English-speaking children acquired by Suppes, Smith, and Lveill (1973) and Brown (1973). The analysis was conducted by Nicoladis et al. (2007), who stressed that in French the difference between regular and irregular verb forms is less clear than in English because irregular forms are comprised of a small number of families and are characterized by higher type frequencies than regular forms are, although the latter constitute the majority of verbs. By contrast, in English, the data indicate that regular verbs make up a relatively small proportion of tokens in child-directed speech, with high type frequencies. As a result, in French the acquisition of irregular forms by bilinguals may be less dependent on frequency of exposure. Nicoladis et al. tested the role of type and token frequency in the acquisition of past tense morphology (regular and irregular forms) in English and French. The study compared young FrenchEnglish bilingual with English- and French-speaking monolingual children. The authors constructed two monolingual groups (French, n = 10; English, n = 10) and a FrenchEnglish simultaneous
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

504

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

bilingual group (n = 10). The children, aged 4 to 6, were asked to tell back the story of a cartoon they watched in English, French, or in both target languages, depending on group. The results showed that when speaking French, bilingual children produced a higher token number of regular than irregular verbs, similarly to French-speaking monolinguals. When speaking English, the same children elicited a higher token number of irregular than regular verbs, similarly to English-speaking monolinguals. This pattern is consistent with dissimilarities between French and English in type and token frequencies of regular and irregular verbs found in the corpus analysis. Moreover, FrenchEnglish bilinguals were significantly less accurate in producing past tense than were English-speaking monolinguals (for both regular and irregular verbs) and French-speaking monolinguals (only for irregular verbs). Thus, the delay in irregular verb production has been found in both languages of bilingual children. These findings were attributed to the fact that bilingual childrens exposure to either language was lower than that of monolingual children (limited input approach). The study also emphasized the need to investigate the nature of the relationship between the bilingual childrens knowledge of vocabulary and the degree of accuracy in their use of past tense morphemes in the target language. But because the researchers did not control for the vocabulary knowledge of bilinguals in the target languages, it is difficult to reach a conclusion about the link between the lower level of accuracy in the bilinguals production of past tense verbs in either French or English and their limited vocabulary in these languages. Schelletter (2007) also examined whether the acquisition of irregular morphology among bilingual children (EnglishGerman) was subject to a frequency effect or to the degree of marking in the two target languages. The study compared the use of regular and irregular inflections (regular/irregular plural nouns and past tense verbs) in English and German by bilingual and monolingual English-speaking children. Eight bilingual and 10 monolingual children aged 3 to 5 participated in the study. All bilingual children had one German parent. The data showed that, contrary to the usage-based model (Bybee, 1995, 2001), bilingual children were less accurate than their monolingual peers in producing both regularly and irregularly inflected verb forms, but only slightly less accurate in producing noun forms. Similarly, bilingual children showed a pattern of minor delay, and their performance on plural nouns and past tense verbs did not differ significantly from that of 4-year-old monolinguals. Both groups also showed high levels of over-regularization in producing irregular plural nouns in English. All children showed 90% accuracy in regular noun and verb morphology. To summarize, the empirical data derived from elicitation tasks show that even English-dominant bilinguals (Paradis et al., 2007) scored lower on irregular verb forms in L2 than did age-matched monolinguals, but in French these differences were not observed. Furthermore, both monolingual and bilingual groups showed great similarity in the over-regularization pattern of errors and the tendency to master regular forms before irregular ones. Note, however, that several theoretical and methodological weaknesses in the above studies limit the conclusions that can be drawn from them. First, the samples were relatively small (up to 14 children in each group) and included only children in pre-school and of similar age. Second, the studies focused on the acquisition of past tense morphology by
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

505

bilingual and monolingual children in two languages with similar structural properties (EnglishFrench and EnglishGerman), where cross-linguistic transfer can affect the mastering of past tense morphology by bilingual children in each language. Therefore, studies involving typologically different languages can enrich the field and yield new theoretical and empirical insights. Third, it has already been noted that English-speaking children tend to learn more irregular verbs than regular ones when they first start using verbs marked for the past tense (Bybee, 2001; Marcus , Pinker, Ullman, Hollander, Rosen & Xu, 1992; Tomasello, 2003). Because the regular form (-ed) has a high type frequency (enhancing rule formation), whereas some irregular forms have a high token frequency (favoring rote learning), constructing models of morphological acquisition by bilingual children in L2 focusing mainly on the English past tense can limit the generalizability of the results. Finally, a longitudinal design is needed to approach developmental issues in irregular morphology acquisition by young bilinguals to verify assumptions about delays in bilingual children relative to their monolingual peers. The complex Hebrew (L2) system of irregular pluralization constitutes a productive context for investigating bilingual childrens acquisition from the developmental perspective because data collected on monolingual children indicate (a) delayed acquisition with low token frequency in early childhood (i.e. in child-directed speech and in child speech) at the onset of the acquisition of the Hebrew plural noun system; and (b) a dramatic increase in output with formal exposure to literacy and grammar (Berman, 1981a; Nir-Sagiv & Ravid, 2006; Ravid, 2007; Ravid & Schiff, 2009).

2 The Hebrew plural system

The most characteristically Semitic feature of Hebrew is its derivational morphology (Berman, 1985). Hebrew nouns include two components: the consonantal root and the word pattern. Words are formed by inserting a root into a word pattern. The root is the semantic core of a word and consists of a sequence of three-four consonants. The entire Hebrew lexicon is based on approximately 2000 roots. Word patterns include placeholders for the root consonants, and they provide the vowels and affixes. To illustrate, verb patterns derived from the triconsonantal root / = klt/ include: / = kalat/, kalat, he grasped, / = niklat/, niklat, was grasped/absorbed, and / = hiklit/, hiklit, he recorded (Share & Levin, 1999). Nominal word patterns are called mishkalim (singular: mishkal). For instance, the words / , kelev/, kelev, dog, and / , begged/, beged, garment, are formed by / klv/ and / bgd/ into the word pattern CeCeC. inserting their root consonants Similarly, the words / , miXtav/, mihtav, letter, and / ,mispa/, mispar, number, which are members of the word pattern miCCaC, are created by inserting their root consonants / Xtv/ and / spr/ into this pattern. Hebrew nouns are further marked for gender (masculine or feminine) and for number. Pluralizing Hebrew nouns is a linear process of stem suffixation. Plural suffixes incorporate information about both number and gender. According to the basic rule, the -im suffix is added to masculine nouns (e.g., (1) / , / tik/tikim/, tik / tik-im, (2) , / bag/s; / / , / panas/panasim/panas/panas-im, flashlight/s, and the -ot suffix to feminine nouns (e.g. (1) / / tipa/tipot/, tipa / tip-ot, drop/s; (2) / , /tmuna/tmunot/, tmunot / tmun-ot, picture/s. Native Hebrew speakers acquire plural
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

506

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

noun morphology very early, in the second-third years (Berman, 1985; Levy, 1980; Ravid, 1995a). Plural suffixation always shifts the noun stress to the final syllable created by the conjunction of the suffix and the final consonant of the stem. Some plurals, however, take irregular suffixes. Masculine plurals can take the feminine -ot suffix (e.g. (1) / , / ki/kiot/, kir / kir-ot, wall/s; and (2) / , / eh ov/ eh ovot/, rehov /rehov-ot, street/s, instead of the expected regular (1) kir-im and (2) rehov-im). Similarly, feminine nouns can be inflected with the masculine suffix -im (e.g. (1) / / , / bejca/bejcim/, beytsa /beyts-im, egg/s, and / , / mila/milim/, mila / mil-im, word/s, instead of the expected (1) beyts-ot and (2) mil-ot). According to Ben Or (1967), modern Hebrew contains over 200 masculine nouns that take the -ot suffix and about 50 feminine nouns that take the -im suffix, compared with tens of thousands of nouns with gender-linked plurals. Regularization of morphological exceptions has been consistently observed among Hebrew-speaking preschoolers, broadly described within the context of the dual-mechanism model cited earlier, so that when children do not remember the target irregular form they tend to default to the regular one. Another case of over-regularization in plural formation is associated with morpho-phonological alterations of the stem (see Berman, 1985; Ravid, 1995b). These changes typically result in the deletion of vowels and sometimes in vowel epenthesis. For example, feminine nouns with the pattern CiCCa and masculine nouns with the form CaCaC require the regular suffixes (-ot or -im), and lose their initial vowels (e.g. the vowel /i/ in feminine noun / , / simla/smallot/, simla / smal-ot, dress/es, and the vowel /a/ in masculine noun / , / afan/fanim/, shafan / shfan-im, rabbit/s). The stem changes are a function of vowel and consonant length, and of phonological contrasts that existed in Tiberian Hebrew but have been lost in Modern Hebrew (Bolozky, 1978). Because members of a given word pattern share their vowels and affixes they are rather similar phonologically. The similarity also helps predict the morpho-phonological structure of the plural form when the stem changes. For instance, all members of the CeCeC word pattern (e.g. / , begged/, beged, clothes) are inflected as CCaC-suffix -im (e.g. / , bgadim/bgad-im, clothes). Schwarzwald (1991) used the term minor rules (regularized irregularity) to designate all semantic and morphological generalizations that apply to groups of exceptions. In sum, in the majority of the cases the morphological structure of the singular nouns provides definite clues about the plurals. Ravid (1995b) distinguished three additional types of morpho-phonological stem changes in pluralizing: (1) vowel change (e.g. /, / h ec/h ecim/, hets /hits-im, arrow/s); (2) dropping of the final t in feminine nouns (e.g. /, / h anut/h anujot/, hanut /hanuy-ot, shop/s); and (3) stop / spirant alternation (spirants changing into stops, e.g. / , / af/apim/, af /ap-im, nose/s and spirantization, e.g., / ,/ / zikaon/ ziXonot/, zikaron /zihron-ot, memory/memories). What is possibly the most complex exceptional category of plural formation combines a full stem change and unpredictable suffixation, presenting a case of radical irregularity, which is completely lexicalized. For example, in the case of the word , /ia/, isha, woman, the plural form has an irregular masculine instead of a feminine suffix and a full stem alteration: / , naim/, nash-im, women. This category contains a relatively small number of items, and according to Schwarzwald (1991) needs to be
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

507

memorized as words in the lexicon, as no sub-rule (regularized irregularity) can capture its idiosyncrasies.

3 Delayed acquisition and the role of literacy


It has been noted that opaque and complex structures are generally more difficult to learn and acquired later than are more transparent structures (e.g., number marking in Arabic, gender in Welsh). Given the opacity of some plural forms, and the relatively restricted number of nouns with irregular pluralization, it is quite clear why young native speakers of Hebrew first acquire plurals that are handled by rule. The developmental data indicate that the different forms of irregular pluralization are mastered considerably later than are the regular ones. For example, Berman (1981c) found that 3-year-old Hebrew-speaking children avoided stem changes in CiCCa feminine ( / , simla/, simla, dress / , smallot/, small-ot, dresses) and CaCaC masculine ( / ,afan/ shafan, rabbit / , fanim/, shfanim, rabbits) forms. Ravid (2007) noted also that there is a strong connection between developmental growth of irregular plural usage in child speech on one hand, and child-directed speech and childrens literature on the other. As children grow older, the frequency of irregular plural input and output increases. Analysis of childrens speech (n = 17), aged 34, using a semi-structured test, showed that children produced two words with non-changing stems and irregular suffixes, 13 words with changing stems and regular suffixes, and four words with changing stems and irregular suffixes. Children aged 910 (n = 20) showed a significant increase in the use of irregular plurals and produced 21 words with changing stems and regular suffixes, 10 words with non-changing stems and irregular suffixes, and 10 words with changing stems and irregular suffixes. The complete acquisition of irregular forms may be attributed to exposure to literary language, which clearly increases the inputs of irregular nouns. In their longitudinal study, Ravid and Schiff (2009) found that among Hebrew-speaking monolinguals nouns with unchanging stems and regular suffixes were already at ceiling level at age 6. At the same time, the acquisition of the exceptional morpho-phonological categories was still under way in first grade. The results also showed that the childrens performance on pluralizing these categories improved dramatically during the first grade. This progress was attributed to the growing frequency of input as well as to enhanced development of phonological awareness, lexical knowledge, and the resulting consolidation of information about the morphological and phonological structure of words. In sum, progress in the acquisition of irregular plurals in Hebrew may be linked to memorization of these forms in the lexicon as well as to increasing awareness of word pattern combinations, which in turn enhances sensitivity to lexical minor rules (Ravid & Schiff, 2009; Schwarzwald, 1991). No systematic comparison has been performed to date between bilingual and monolingual children with regard to the mastering of irregular plurals in Hebrew. The present multi-interval study has been designed to fill this void by focusing on the intriguing case of RussianHebrew bilingual children with very early exposure to L2.
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

508

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

4 The present study

We based the present research on three assumptions. The first assumption was that young bilingual children typically use and hear less of the second language than their monolingual peers do because of limited L2 input in the home (Cobo-Lewis, Pearson, Eilers, & Umbel, 2002; De Houwer, 1995; Pearson, Fernndez, Lewedag, & Oller, 1997). The limited L2 input seems to affect both the size of their vocabulary (type frequency of irregular plural forms) and the frequency of exposure to particular words (token frequency of irregular plurals). If the childrens progress in the acquisition of irregular plural forms depends on the amount of active and passive contact needed to store these forms in the lexicon and memorize them, bilingual children are likely to perform below their monolingual peers in the correct use of these forms and in vocabulary knowledge in L2 Hebrew. Based on the data cited earlier, we predicted that bilingual children would experience patterns of difficulty (developmental mistakes) in producing exceptional plurals, similar to the difficulties experienced by their monolingual peers but with a higher level of over-regularization and more persistently. Thus, consistent with Gathercoles (2006) suggestion, bilingual children seem to require more time to acquire the critical mass of input needed to master irregular pluralization. The second assumption was that bilingual children were heterogeneous in their language dominance and preferences, the sequence of exposure to the two languages, and the type of educational system in which they were enrolled. In the present study we focused on the second generation of RussianHebrew-speaking (RHS) immigrants in Israel, who are typically exposed to L2 Hebrew soon after birth (following enrollment in Hebrew daycares), almost simultaneously with home language acquisition. Members of this research population have been recently defined as almost simultaneous or very early successive bilinguals, that is, children exposed to L2 near the onset of L1 (Meisel, 2008). As soon as a bilingual child enters nursery school, Hebrew inevitably appears to be the socially and educationally dominant language (Kopeliovich, 2006; Schwartz, 2006). Only a slight gap, if any, can be expected between young bilinguals and monolinguals in their knowledge of Hebrew vocabulary. As noted earlier, the development of Hebrew plural noun categories is still under way during the first school years and it is enhanced by massive exposure to written language (Berman, 1985; Ravid & Schiff, in press). These factors seem to affect both bilingual and monolingual children in a similar way, so that both groups should exhibit comparable patterns of difficulty in producing irregular plurals. Whereas Ravid and Schiff (in press) measured the production of irregular forms at the beginning and end of the first grade, while children acquire basic literacy skills in Hebrew, we used two data collection points that capture the next advance in schooling: The beginning of second and third grades. Accordingly, our third assumption was that bilinguals close the gap with their monolingual peers possibly owning to exposure to written texts and increased L2 input. We made three predictions. First, based on the agreement that irregular forms must be memorized and on the assumption that on average bilingual children probably hear irregular Hebrew nouns less frequently than their monolingual peers do, we predicted that bilinguals would be less accurate in the production of these forms. Second, given
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

509

that acquisition of irregular forms of Hebrew plural nouns is still under way during the first school year, we expected both groups to improve their performance on measures of irregular forms between the two data collection points. Both data collection points were past the first grade. Finally, we predicted that despite possible quantitative differences between the groups in the production of irregular plural forms, both groups would show a resemblance in the pattern of difficulty they faced in pluralizing Hebrew nouns.

5 Method
Participants

A sample of 7- to 8-year-old Russian heritage speakers (RussianHebrew-speaking bilinguals) and monolinguals participated in the study. Participants were tested twice on the same receptive vocabulary knowledge and plural tasks, at two data collection points: The beginning of the second grade, in October, after completing the acquisition of basic literacy skills in Hebrew (Time I), and at the beginning of third grade (Time II). We selected 7-year-old children (Time I) because this age allows testing the way in which bilingual children cope with irregular forms as opposed to regular ones, when acquisition of pluralizing is still incomplete (Berman, 1985; Ravid, 1995a). As noted earlier, the choice of the beginning of third grade as the second data collection point is linked to the childrens intensive exposure to written texts and textbooks, and to explicit instruction about target morphological systems in the second grade. At Time I, 125 children (70 bilinguals and 55 monolinguals) participated, with a mean age of 7, 2 (years, months). At Time II, six children (five bilinguals and one monolingual) were not tested because they had changed their place of residence, leaving 119 participants for Time II. The groups did not differ significantly in gender distribution (Time I: c2 (1) = 0.07, p > 0.05; Time II: c2 = 0.38, p > 0.05), and non-verbal intelligence (Time I: F(1, (1) 123) = 0.03, p > 0.05; Time II: F(1, 117) = 0.19, p > 0.05 ). Participants were selected from 12 elementary schools (24 classrooms) in the northern region of Israel, including the city of Haifa and its suburbs, with a population of predominantly middle-lower socio-economic status. The relatively large number of schools and classrooms enabled us to minimize school and teacher effects. In all the schools, Hebrew was the sole language of formal instruction and communication. Note that both bilingual and monolingual children were chosen from each classroom. The percentage of Russian heritage speakers in these classrooms ranged from 28% to 42%. Participant selection was conducted in two stages. First, we obtained parental consent of approximately 70% of the students in these classes by communicating directly with both bilingual (RussianHebrew) and monolingual (Hebrew) parents during parent teacher conferences at the beginning of the school year. The consent form for bilingual parents included questions about their date of arrival in Israel and their rating of language dominance at home. Parents were asked to state which language they used with spouses and children: only Russian, only Hebrew, or both Russian and Hebrew. The reports show that parents preferred to use mainly Russian for communication with family members. Next, based on the information provided by parents and teachers, we selected our groups of bilingual and monolingual participants. The bilingual participants, children of RussianJewish immigrants, met the following inclusion criteria: (a) Speaking Russian as the first and dominant language at home; (b) Israeli-born (n = 65) or having immigrated
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

510

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

to Israel at least 6 years before the start of their schooling (n = 4), more specifically before the age of 9 months. All bilingual children were sequential bilinguals who had been exposed to L2 Hebrew in Hebrew-speaking preschool programs. The immigrant families residence in Israel was relatively long (M = 11.6; SD = 3.93). To collect data on language use in the bilingual group we administered a short questionnaire to bilingual children, requesting information about their linguistic choice, use, and preference by domains and by speech acts: Russian only, Hebrew only, or both Russian and Hebrew with father, mother, grandparents, siblings, and Russian-speaking friends. Two additional items provided data about the childrens best-known and favorite language. The questionnaires showed that although children reported frequent use of Russian (L1) with their parents and grandparents, some of them were more likely to communicate with their parents in Hebrew only. In addition, as expected, the children commonly used Hebrew with their siblings. At the time the study was conducted, 21% of the children had no siblings and 27% had older siblings in the family, which seems to have affected the childrens preference of Hebrew over Russian in between-sibling interactions. Hebrew was also the preferred language of communication with Russian-speaking friends. Finally, the questionnaire revealed that most bilingual children considered Hebrew to be their strongest and preferred language. These findings are consistent with our expectation concerning the dominance of Russian as a heritage language in parentchild communication and of Hebrew in social interactions. We also collected information on the socio-economic index of each school that participated in the study. This index is calculated by the Israel Ministry of Education based on parents reports on their income, vocation, and occupation, and on a ranking of the families residential areas. The SE index is measured on a 10-point scale, ranging from 1 to 10, with higher scores indicating lower SES. In the present study, the SE index for the schools ranged from 4.25 to 7.40 (M = 5.3; SD = 0.93) and can be characterized as average and slightly below average. The monolingual children were matched with their bilingual peers for background measures: Gender and non-verbal IQ.
Design Measures

General cognitive ability Ravens Colored Matrices (Raven, Raven, & Court, 1976, sets A, B, and C). This is a multiple-choice test of abstract reasoning designed for young children. In each test item children are asked to identify the missing segment required to complete a larger pattern by matching one of six graphic patterns with a visual array. Most items are presented on a colored background to make the test visually stimulating for participants. The maximum score is 36. Receptive vocabulary knowledge Peabody Picture Vocabulary Test (PPVT-R) (Nevo, 1979, Hebrew version; after Dunn, 1965). The test consists of plates containing four pictures each. PPVT-R requires children to indicate which of the four pictures matches a spoken word pronounced by the examiner. Items are arranged in increasing order of difficulty. The maximum score is 50. Plurals Task (Ravid & Yagev, 2003). Children are shown 24 pairs of pictures, each pair represented single and plural forms of the same object (e.g., a picture of a, / , bejca/, beytsa, egg and one of / , alo bejcim/, shalosh beytsim, three eggs). Children
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

511

are asked to name the plural form of the object. The test was devised to reflect all four morpho-phonological categories in the Hebrew plural system: one regular and three exceptional ones. The regular category of a non-changing stem with a regular suffix was represented by two words. The three exceptional categories included forms with an irregular suffix or a changing stem or both: (a) Non-changing stem with irregular suffix, represented by six words; (b) changing stem with regular suffix, represented by 11 words; and (c) changing stem with irregular suffix, represented by five words. Because of existing data on the early acquisition of regular plural inflexions, the number of items in the regular category was limited to two words. Each category contained both masculine and feminine nouns. The 24 plural items were selected from a preliminary list of 50 items in consultation with the participants teachers to ensure that all words were familiar to both groups at the beginning of the second grade (Time I). The teachers were asked to rank the frequency of the 50 items on a scale of 15, from low to high frequency. Only words ranked 4 and above were chosen for this task. Table 1 shows the structure of the task and examples of the items illustrating the target morpho-phonological categories in the Hebrew plural system. Internal consistency (Cronbachs ) was .70 for Time I and .64 for Time II.
Scoring the Plural Task

Following the model proposed by Ravid & Schiff (2009), the stem and the suffix of each word were scored on a separate scale shown below. The detailed ranking enabled us to conduct a fine-grained analysis of morphological knowledge of stems and suffixes. The stem was scored on a 4-point scale (03): (a) repeating the stimulus word or no response (0 points); (b) producing an irrelevant linguistic operation (e.g., / , jeladot/, yelad-ot, girls, for / ,banot/, ban-ot, daughters, from / , bat/, bat, daughter), and leaving the stem unchanged where change is required (e.g. /cadim/, cad-im for , /cdadim/, tsdad-im, sides, from / , cad/, tsad, side) (1 point); (c) partial stem change where change is required (e.g. / atihim/, shatih-im for , / tihim/, shtih-im, carpets, from / , atiah /, shatiah, carpet) (2 points); (d) correct stem (3 points). The suffix was scored on a 3-point scale (02), with 0 points allocated for repetition of the stimulus item, 1 point for an incorrect suffix, and 2 points for a correct suffix.
Procedure

Testing of both groups at each data collection point (Time I and Time II) was completed in one session lasting approximately 30 minutes. Each child was assessed individually in a quiet room. A native Hebrew speaker administered the tasks. Instructions were given in Hebrew. The questionnaire for bilingual children was administered in Russian by a native Russian speaker at Time I, and took approximately 10 minutes to complete.

6 Results

To test the effect of time and group on all measures of irregular Hebrew plural forms, we performed four separate ANOVAs (2 2) with repeated measures on the type of
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

512

Table 1
Changing stem Masculine Feminine

Structure of the plural tasks and examples

Stem type Feminine

Non-changing stem

Stem gender

Masculine

The International Journal of Bilingualism

Regular suffix / , tik/, tik, bag / ,begged/, beged, clothes / ,bgadim/, bgad-im, clothes / ,cad/, tsad, side / , cdadim/,
tsdad-im, sides

/ , tmuna/, tmuna, picture / , tmunot/, tmun-ot, pictures

/ , bat/, bat, daughter / , banot/, ban-ot, daughters / ,ah ot/, ahot, sister , / , ahajot/,
ahay-ot, sisters

/ ,tikim/, tik-im, bags

Irregular suffix

/ , ki/, kir, wall

/ , mila/, mila, word

/ , lev/, lev, heart / , lev/, levav-ot, hearts / , kaf/, kaf, soupspoon / , kapot/, kap-ot, soupspoons

/ , ia/, isha, woman / , naim/, nash-im, women / , i/, ir, town/city / , aim/, ar-im, towns/cities

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

/ , ki ot/, kir-ot, wall / , milim/, mil-im, words / , bejca/, beytsa, egg / , bejcim/, beyts-im, eggs

/ , sulam/, sulam, ladder ,

/ , sulamot/, sulam-ot, ladders

Schwartz et al.: Delayed acquisition

513

irregular pluralization (suffix, stem, or both), and with one within-group variable (time) and one between-groups variable (bilinguals vs. monolinguals). Because we did not find significant differences between the groups on regular measures, including exceptional morpho-phonological categories (see Table 2), we conducted repeated measures only on irregular plural forms in each category (i.e., on irregular suffixes (first category), on changing stems (second category), and on changing stems and irregular suffixes (third category). As predicted, the analysis revealed the significant main effect of time, with both groups improving on all four measures of irregular plurals: Irregular suffixes: F(1, 117) = 9.72, p < 0.05; changing stems: F(1, 117) = 5. 57, p < 0.05; changing stem in the changing stems with irregular suffixes category: F(1, 117) = 5. 57, p < 0.05; and irregular suffixes in the changing stems with irregular suffixes category: F(1, 117) = 6.07, p < 0.05. In addition, as expected, the main effect of the group has been found on three of four measures of irregular forms: Irregular suffixes within the first category (F(1, 117) = 17.26, p < 0.001); changing stems within the second category (F(1, 117) = 8. 13, p < 0.05); and changing stems within the category of changing stems with irregular suffixes (F(1, 117) = 7. 87, p < 0.05). Finally, we did not find significant interactions between time and group on any one of the target measures. Table 2 summarizes the childrens performance on the three categories of plural noun measures at the two data collection points. The table also presents p-values and Cohens d values. Cohens d was used as a measure of effect size (.20 being small, .50 medium, and .80 large; Cohen, 1992). Because all participants correctly produced the regular plural forms at both data collection points, as expected, these findings have not been included in the table. The findings show that at both data collection points the two groups performed similarly on regular measures, including the first and second morpho-phonological categories, and reached near ceiling levels. We also found a clear tendency for bilinguals to close a gap with the monolinguals on all measures of irregular plural forms. Moreover, in the case of changing stems (second category), although the main effect of the group was significant, with monolinguals outperforming the bilinguals, at Time II the p-value revealed only a marginal difference between the groups (F(1, 117) = 2.22, p = 0.05) We also investigated the effect of time and group on receptive vocabulary by performing 2 2 ANOVA tests with repeated measures on one within-group variable (time) and one between-groups variable (bilinguals vs. monolinguals). Similarly to the findings reported above, the results show the significant main effects of time (F(1, 117) = 56.57, p < 0.001) and group (F(1, 117) = 14.30, p < 0.001), with monolingual children outperforming their bilingual peers (for bilinguals: M = 59.2; SD = 10.32; for monolinguals: M = 64.9; SD = 7.48). Note also that there was not evidence of closing of the gap between the groups (Table 2). Finally, no interaction effects were detected between time and group. Pearson productmoment correlations were computed separately for bilingual and monolingual children to examine the relations between the receptive vocabulary measure and all four measures of exceptional plural categories in Hebrew at Time I and Time II. As shown in Table 3, which presents the results of correlation analysis for the bilingual group, there were significant correlations between the target measures at
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

514

Table 2

Means and standard deviations of plural noun measures in Hebrew (by time of testing, morpho-phonological category, and receptive vocabulary) for the bilingual and monolingual groups (percentages of the maximum variable score)

Number of category/ variables Time I Monolinguals (n = 54) Total p 0.062 (6.04) 0.000 (10.29) 0.009 (7.50) 0.436 (3.69) 0.012 .45 75.1 (12.26) 0.030 0.001 .39 .56 88.6 (9.98) 62.1 (9.74) 98.5 .47 87.7 (7.91) 90.2 (5.35) 99.7 (1.73) 80.7 (10.73) 90.4 (9.71) 67.9 (7.43) .64 85.8 91.1 (5.76) 93.2 94.6 93.9 (5.90) 88.4 (9.10) 88.9 (6.42) 99.1 (2.71) 77.9 (11.49) 89.5 (9.84) 65.0 (8.58) 0.001 0.336 0.025 0.261 0.046 0.003 d 92.1 (6.21) 84.7 (9.75) 86.8 (10.04) 97.6 (4.59) 73.8 (11.95) 86.4 (11.54) 59.1 (9.22) Total p 0.206 93.1 (6.50) 88.0 (9.86) 89.3 (8.57) 98.0 (3.58) 76.6 (11.96) 88.7 (11.23) 61.9 (7.53) Bilinguals (n = 65) Monolinguals (n = 54) Time II

Bilinguals (n = 65)

The International Journal of Bilingualism

stem

91.0

1. Non-changing stem Irregular suffix

(5.93)

suffix

81.4

.55

(9.64)

2. Changing stem Regular suffix

stem

84.3

(11.51)

suffix

97.3

.40

(5.61)

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

3. Changing stem Irregular suffix

stem

71.1

(11.94)

suffix

84.1

.63

(11.86)

Receptive vocabulary

56.3

(10.91)

Schwartz et al.: Delayed acquisition

515

Time I and II, except for irregular suffixes (first category at Time I and third category at Time II). The strongest correlations (medium-size effect) were obtained between the receptive vocabulary and changing stem measures (second category) (Time I: r = .55, p < 0.01; Time II: r = .45, p < 0.01). Table 3
Correlation between receptive vocabulary and measures of irregular plural forms for the bilingual group at two data collection points

Number of category

Variable

Receptive vocabulary Time I Time II .35** .45** .29* .16

1 2 3 3
*p < .05, ** p < .001

Irregular suffix Changing stem Changing stem Irregular suffix

.15 .55** .39** .30**

We found only one significant correlation, with a medium-size effect, for the monolingual group, between the receptive vocabulary and changing stem measures (second category) (r = .49, p < 0.01) at Time I. This pattern may be attributable to the fact that the results of monolingual children were close to ceiling levels. An additional set of analyses employed multiple regression procedures on the total sample to distinguish the contribution of receptive vocabulary to the acquisition of irregular plural forms at the two data collection points separately. To this end, two composite measures of irregular plural forms (each for one data collection point) were created by extracting the first principal component from the set of four measures. This first component accounted for a majority of the variance in this set (Time I: 57%; Time II: 54%), with substantial weights for each of the four variables (Time I: .600, .722, .872, and .802; Time II: .690, .586, .881, and .800). Using hierarchical regression analysis, we examined the unique contributions of receptive vocabulary to irregular plural forms after controlling for background variables (non-verbal IQ (Raven) and gender). When we entered non-verbal IQ and gender at Step 1, we found that neither of these variables contributed to significant amount of difference in the production of irregular plural forms at the two data collection points. At Step 2, after entering non-verbal IQ and gender, receptive vocabulary accounted for a substantial amount of variance in the production of irregular plural forms (Time I: F(1, 117) = 46.76, p < 0.001, R 2 = .28; Time II: F(1, 117) = 25.57, p < 0.001, R 2 = .18). These outcomes suggest the importance of vocabulary knowledge in the acquisition of irregular plurals in Hebrew.
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

516

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

7 Discussion

The objective of the present study was to shed light on the acquisition of irregular noun pluralization in Hebrew (L2) by very early successive bilingual children. The study produced several important findings. First, the results support our hypothesis about the link between the bilingual childrens possibly limited input of Hebrew (L2) and their slight delay in the acquisition of irregular noun plurals relative to their monolingual counterparts. Note that a direct measurement of the amount of L2 input was not within the scope of the present study. The construct validity of our assumption concerning the reduced amount of Hebrew spoken in the home can be enhanced if multiple factors are considered together (Mackey & Gass, 2005): the parents and childrens reports of L1 dominance in the parentchild interactions and the significantly lower scores achieved by bilingual children on receptive vocabulary measure in Hebrew (L2). Whereas at Time II monolingual children achieved close to ceiling scores, their bilingual peers were behind by significant differences on irregular suffixes (first category) and changing stems with irregular suffixes (third category). In the case of irregular suffixes, bilingual children showed a greater tendency than their monolingual peers to apply canonical rules of morphology, a pattern attributable to the pragmatic factor of lower familiarity resulting from the reduced frequency of target plural forms encountered and used by bilingual children. It appears, therefore, that bilingual children relied less on the phonological properties of the single stem governing inflectional irregularity with irregular suffixes, and were thus more dependent on memorizing similar words in the lexicon (Berman, 1985; Schwarzwald, 1991).

The picture is less clear with regard to the lower scores of bilingual children on changing stems within the most complex category (the one with the most opaque change in the stems and the irregular suffixes). After verifying the robustness of the differences between the target groups by measuring the effect size (Cohen, 1992) at Time II, the gap, although statistically significant, was found to be relatively small (d = .40). The fact that there was only one item on which the scores of bilingual children differed significantly, the plural form for / , kaf/, kaf, soup spoon, seems to confirm that their lower scores have no practical importance. It is unusual, however, that 48% of bilinguals and 18% of monolinguals made a similar mistake in producing the plural form for kaf, using the incorrect form /kapijot/, kapiy-ot, teaspoons. This type of mistake cannot be attributed to either over-regularization (which would have resulted in /kafim/, kaf-im instead of / , kapot/, kap-ot, soup spoons) or to irregularization (which would have produced a stop/spirant alternation of the stem, /kapim/, kap-im, similar to the word , / /daf/dapim/, daf / dap-im, page/s, without irregular suffixation. The production of a form such as kapiy-ot, teaspoons, can more likely be referred to as pragmatic interference, as young children are more likely to use small teaspoons than soupspoons. Note further that this unexpected pattern of mistakes occurred mostly in the context of the most opaque irregular category, which has a relatively small number of members (Schwarzwald, 1991). The findings also bear evidence to the bilingual childrens tendency to close the gap with their monolingual peers. Whereas the scores of monolingual children changed by about 2% as a whole, those of the bilingual children improved significantly by about 4%. A simple explanation for the progress of bilingual children lies in the greater frequency
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

517

of usage of irregular forms. This account is consistent with Ravids (2007) data on the growth in the output of these forms among monolingual early school-age children and with our findings about the progress of bilingual children on the L2 receptive vocabulary measure. A second, complementary explanation is based on the bilingual childrens greater knowledge of word structure. Familiarization with the consonantal-syllabic orthography of Hebrew and insight into the phonemic and morphemic structure of the words enhances sensitivity to word patterns and consequently the ability to predict morphophonemic stem alternations (Berman, 1985; Ravid & Schiff, 2009). The fact that bilingual children nearly closed the gap in the case of the more transparent stem alternations (second category) but have not yet done so in irregular suffixes (first category) supports our assumption that they can predict stem changes based on a developed meta-linguistic awareness of word patterns. For example, higher frequency of exposure and increased use of so-called segolate nouns (with a CeCeC pattern) that change significantly in the plural to CCaC (e.g. / , / sefel/sfalim/, sefel-im, cup/s) can increase the probability of producing a correct plural form by applying the relevant minor rule targeting the single item included in this group (Schwarzwald, 1991). The strongest correlation found for bilinguals was between receptive vocabulary knowledge and changing stems (second category) at both points of data collection. At the same time, the correlation between receptive vocabulary knowledge and irregular suffixes was relatively moderate. These results may be explained by the greater diversity of word patterns requiring morpho-phonological changes in their stems than in their suffixes (Berman, 1985; Ravid, 1995a). It appears, therefore, that the progress of bilingual children in L2 vocabulary may enhance and enrich the representation of such patterns in the lexicon. Finally, based on analysis of our results at Time I and II, we suggest that the bilingual childrens developmental patterns with respect to the pluralization of Hebrew nouns are rather similar to those of their monolingual peers. Specifically, our data indicate that both groups reached a ceiling effect in the easiest category of non-changing stems and regular suffixes, and no reliable differences have been found between the two groups on the regular components (non-changing stems and regular suffixes) in the two exceptional categories. Furthermore, both groups demonstrated the lowest accuracy level in the most challenging type of irregular pluralizing: full morpho-phonological change of the stem in the changing stems and irregular suffixes (third category). These results are also in keeping with earlier research on the acquisition of regular/irregular plural forms by monolingual first graders, reported by Ravid and Schiff (2009). The relatively low performance of both groups in the formation of plurals for words with nontransparent stems demonstrates that these forms emerge later than the more transparent ones (Berman, 1985; Dromi, 1987; Ravid and Schiff, in press). The two groups also showed great similarity in making various types of mistakes. The most frequent error was over-generalization of typical plural forms, as in the case of the masculine noun / , avit/, sharvit, scepter (Time I: 80% among bilinguals, 53% among monolinguals; Time II: 59% among bilinguals, 54% among monolinguals). This singular noun form misled the children by presenting a deceptive phonological marking, a stem ending in -t, similar to the phonological marking of the feminine singular form. The disparity between the masculine gender of the word and its phonological marking of a feminine ending resulted in a strong tendency for over-regularization (producing
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

518

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

/avitot/, sharvit-ot or /avijot/, sharviy-ot instead of / , avitim/, sharvit-im, scepters).

In addition to the high rate of over-regularization, children in both groups engaged in irrelevant linguistic operations and created ill-formed structures at both data collec tion points. For example, six children produced the semantic substitution of , /jeladot/, yelad-ot, girls, for / , banot/, ban-ot, daughters, from the word / , bat/, bat, daughter, which in common speech is used generically for girls (see Ravid, 1995a). The highest rate of ill-formed structures was found in the case of the most opaque irregular type: Changing stems with irregular suffixes (third category). In some cases the erroneous plural construction was based on phonological similarity with other, well-known plural forms, as in the following examples. Five children (three monolinguals and two bilinguals) produced the ill-formed plural form / , levivot/, leviv-ot, potato pancakes, for / , levavot/, levav-ot, hearts, from the word / , lev/, lev, heart; and four children (two bilinguals and two monolinguals) created the ill-formed plural form / , arajot/, arayot, lions, for / , aim/, ar-im, towns/cities, from the word, , /i/, ir, town/city. In sum, the data show considerable similarity between the groups in the pattern of difficulties children faced in pluralizing Hebrew nouns at both points of data collection. The overwhelming amount of developmental errors, such as the overgeneralization of typical plural forms, in both groups, is consistent with the claim made by Genesee, Paradis, and Crago (2004) about the similarity between English (L2) learners and English-speaking monolinguals in the difficulties these populations experiences in acquiring English morphology. Finally, the results of regression analysis underscore the importance of vocabulary knowledge in the acquisition of irregular plurals in Hebrew and support the claim presented earlier that the irregular forms are probably learned by rote, similarly to the way in which words are learned, and that children may memorize these inflectional form as part of the lexicon.
Limitations

The present study faced four methodological limitations. First, the delay in production of irregular L2 forms by bilingual children has been found only by means of the elicitation technique, and therefore future studies must examine the frequency of input of exceptional plural forms by collecting a bilingual corpus. Second, because our assumption about growing sensitivity to the phonemic and morphemic structure of words during Hebrew grammar lessons is based only on personal observation and examination of the school curriculum, future research must investigate directly how this sensitivity actually develops. Third, using a single test to measure the production of irregular plurals may have a restricting effect on our conclusions. Future research should include both structural and non-structural tests, for example, script production. At the same time, the striking resemblance between the patterns of our data about the performance of monolingual children and those reported by Ravid and Schiff (2009) attests to the construct validity of our results. The last limitation is inherent in the use of a relatively small number of test items (24). Note, however, that the items included in the study reflect all four morpho-phonological categories in the Hebrew plural system, covering specifically a great variety of irregular forms (Ravid & Yagev, 2003). Thus, sensitivity
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

519

to the full range of irregular pluralization contributes to the content validity of our measurements.

8 Conclusion

Our interpretation of the present data is based on the assumption that minor differences between bilingual and monolingual children in exceptional pluralizing are related to differences in Hebrew language input. Among bilingual children, this limitation seems to affect both the L2 receptive vocabulary and the opportunities for learning trials of particular irregular plurals (i.e., token frequency of exposure). At the same time, we expect the slight gap between the groups to almost disappear with further development of phonemic and morphological awareness and greater exposure to oral and written Hebrew. The results can also be interpreted in another way, namely that despite clear differences in frequency of L2 input, the two groups are more similar than one would expect. Following this interpretation, we could speculate that factors other than frequency play a significant role in the progress of bilingual children, for example growing meta-linguistic awareness (sensitivity to morphophonemic stem alternations). Growing awareness of word structures that share common patterns of morphophonological changes in their stems enabled bilingual children to predict the plurals of words exhibiting stem alternations and to close the gap with their monolingual peers. It seems, therefore, that the production of plural forms with more predictable stem changes (second category) depends both on the input frequency of the form and on exploring noun patterns that require this type of change (i.e., minor rules). At the same time, the production of plural forms with irregular suffixes (first category) and unpredictable stem changes (third category) imposes greater demands on the frequency of input of these target plurals and consequently on lexicon memorization. It is possible, therefore, that similarly to their monolingual peers, bilingual children apply three strategies to cope with irregular noun pluralizing in Hebrew: rule, rote, and analogy (Berman, 1981b; MacWhinney, 1975). Note, however, that we did not investigate directly how this sensitivity was actually developed and what type of meta-linguistic activities have been conducted in the target classrooms, and therefore this suggestion must be verified empirically. Within the broader theoretical context of language development by young sequential bilinguals, our data show a need to consider the type of bilingualism and educational background of bilingual children as possible determinants for their L2 grammatical development. Furthermore, the pattern of mistakes in irregular plural forms appears to indicate that sequential bilinguals showed a higher rate of over-regularization than did their monolingual peers, although there was a striking similarity between the two groups. Finally and most important, the findings bear evidence that sequential bilinguals exposed to L2 in early childhood are likely to follow the same developmental trajectory as their monolingual peers in the process of long-term acquisition of irregular pluralization. Consequently, their pattern of L2 inflectional morphology acquisition in the target domain is likely to be closer to simultaneous acquisition, that is to the pattern defined as bilingual (2L1) acquisition (see De Houwer, 1995; Gathercole, 2006; Genesee et al., 2004; Meisel, 2004).
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

520

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

We must use caution in drawing general conclusions about developmental progress in irregular morphology among bilingual children based mostly on English as one of the two languages. Our results highlight the language-specific nature of grammatical development in L2 among young bilinguals (Marchman, Martnez-Sussman, & Dale, 2004). Further research is needed to examine more closely the linguistic diversity in the developmental trajectory of irregular morphology acquisition.

Acknowledgements
This research was supported by the grants awarded to Dr Mila Schwartz by Kreitman Postdoctoral fellowship (Ben-Gurion University, Beer-Sheva, Israel) and by the Edmond J. SAFRA foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We are grateful to Prof. Dorit Ravid, Dr. Ronit Levie and two anonymous reviewers for their useful comments.

References
BEN OR, A. (1967). LaSon Vesignon. Tel-Aviv: Israel. BERMAN, R. A. (1981b). Language development and language knowledge: Evidence from acquisition of Hebrew morphophonology. Journal of Child Language, 8, 609626. BERMAN, R. A. (1981c). Childrens regularization of plural forms in the Hebrew noun system. Papers and Reports on Child Language Development, 20, 3444. BERMAN, R. A. (1985). The acquisition of Hebrew. In D. I. Slobin (Eds.), The crosslinguistic study of language acquisition (pp. 255371). Mahwah, NJ: Erlbaum. BOLOZKY, S. (1978). Some aspects of modern Hebrew phonology. In R. A. Berman (Ed.), Modern Hebrew structure (pp. 1167). Tel-Aviv: Universities Publishing Projects. BROWN, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press. BYBEE, J. L. (1995). Regular morphology and the lexicon. Language and Cognitive Processes, 10, 425455. BYBEE, J. L. (2001). Phonology and language use. Cambridge: Cambridge University Press. COBO-LEWIS, A. B., PEARSON, B. Z., EILERS, R. E., & UMBEL, V. C. (2002). Effects of bilingualism and bilingual education on oral and written Spanish skills: A multifactor study of standardized test outcomes. In D. K. Oller & R. E. Eilers (Eds.), Language and literacy in bilingual children (pp. 98117). Clevedon: Multilingual Matters. COHEN, J. (1992). A power primer. Psychological Bulletin, 112, 155159. DE HOUWER, A. (1995). Bilingual language acquisition. In P. Fletcher & B. McWhinney (Eds.), The handbook of child language (pp. 219250). Oxford: Blackwell. DROMI, E. (1987). Early lexical development. Cambridge: Cambridge University Press. DUNN, L. M. (1965). Peabody picture vocabulary test. Minneapolis, MN: American Guidance Service. GATHERCOLE, V. M. (2006). Miami and North Wales, so far and yet so near: Morpho-syntactic development in bilingual children. Manuscript, University of Wales Bangor. GENESEE, F., PARADIS, J., & CRAGO, M. B. (2004). Dual language development and disorders: A handbook on bilingualism and second language learning. Baltimore, MD: Paul H. Brookes. KOPELIOVICH, S. (2006). Reversing language shift in the immigrant family: A case study of a Russian-speaking community in Israel. Unpublished PhD thesis, Bar Ilan University. LEVY, Y. (1980). The acquisition of gender. Unpublished PhD thesis, [In Hebrew], Hebrew University.
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Schwartz et al.: Delayed acquisition

521

MACKEY, A., & GASS, S. M. (2005). Second language research: Methodology and Design. Mahwah, NJ: Lawrence Erlbaum. MACWHINNEY, B. (1975). Rules, rote, and analogy in morphological formations by Hungarian children. Journal of Child Language, 2, 6578. MARCHMAN, V. A., MARTNEZ-SUSSMAN, C., & DALE, P. S. (2004). The language specific nature of grammatical development: Evidence from bilingual language learners. Developmental Science, 7, 212224. MARCUS, G. F., PINKER, S., ULLMAN, M. T., HOLLANDER, M., ROSE, T. J., & XU, F. (1992). Overregularization in language acquisition. Monographs of the Society for Research in Child Development, 57. MCCLELLAND, J. L., & PATTERSON, K. (2002). Rules or connections in past-tense inflections: What does the evidence rule out? Trends in Cognitive Sciences, 6, 472474. MEISEL, J. M. (2004). The bilingual child. In T. K. Bhatia & W. C. Ritchie (Eds.), The handbook of bilingualism (pp. 91113). Oxford: Blackwell. MEISEL, J. M. (JulyAugust, 2008). Morphological development in early child second language acquisition. Paper presented at the 11th Congress of the International Association for the Study of Child Language, Edinburgh, UK. NEVO, B. (1979). Peabody picture vocabulary test. [Hebrew edition of Dunns Peabody Picture Vocabulary Test]. Haifa: University of Haifa. NICOLADIS, E. (2002). Whats the difference between toilet paper and paper toilet? FrenchEnglish bilingual childrens crosslinguistic transfer in compound nouns. Journal of Child Language, 29, 843863. NICOLADIS, E., PALMER, A., & MARENTETTE, P. (2007). The role of type and token frequency in using past tense morphemes correctly. Developmental Science, 10(2), 237254. NICOLADIS, E., & PARADIS, J. (October, 2006). The role of frequency in cross-linguistic differences in bilingual childrens past tense acquisition. Paper presented at the 5th Annual Conference on the Mental Lexicon, McGill University, Montreal, QC. NIR-SAGIV, B., & RAVID, D. (May, 2006). Noun plurals in Hebrew child-directed speech: Frequency of stem and suffix categories. Paper presented at Workshop on Acquisition and impairments of inflectional morphology, Morphology conference, Budapest, Hungary. PARADIS, J., NICOLADIS, E., & CRAGO, M. (2007). French-English bilingual childrens acquisition of the past tense. Paper presented at the 32nd Annual Boston University Conference on Language Development, Boston University. PEARSON, B. Z., FERNNDEZ, S. C., LEWEDAG, V., & OLLER, D. K. (1997). The relation of input factors to lexical learning by bilingual infants (ages 10 to 30 months). Applied Psycholinguistics, 18, 4158. PEARSON, B. Z., FERNNDEZ, S. C., & OLLER, D. K. (1993). Lexical development in bilingual infants and toddlers: Comparison to monolingual norms. Language Learning, 43(1), 93120. PINKER, S. (1991). Rules of language. Science, 253, 530535. PINKER, S., & PRINCE, A. (1988). On language and connectionism: Analysis of a parallel distributed processing model of language acquisition. Cognition, 28, 59108. PINKER, S., & ULLMAN, M. T. (2002). The past and future of the past tense. Trends in Cognitive Sciences, 6(11), 45663. RAVEN, J., RAVEN, J. C., & COURT, J. H. (1976). Ravens progressive matrices. Oxford: Oxford Psychologists Press. RAVID, D. (1995a). Language change in child and adult Hebrew: A psycholinguistic perspective. New York: Oxford University Press. RAVID, D. (1995b). The acquisition of morphological junctions in Modern Hebrew. In H. Pishwa & K. Maroldt (Eds.), The development of morphological systematicity: A crosslinguistic perspective (pp. 5577). Tubingen: Gunter Narr. RAVID, D. (October, 2007). Morpho-phonological categories of noun plurals in Hebrew. Paper presented at the workshop on Language, University of Haifa, Israel.
The International Journal of Bilingualism
Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

522

INTERNATIONAL JOURNAL OF BILINGUALISM 13 (4)

RAVID, D., & SCHIFF, R. (2009). Morpho-phonological categories of noun plurals in Hebrew: A developmental study. Linguistics, 47(1), 4563. RAVID, D., & YAGEV, I. (July, 2003). Language knowledge in preschool children with cleft palate and their peers. Paper presented at Child Language Seminar, University of Newcastle upon Tyne, UK. ROTHWEILER, M. (JulyAugust, 2008). Verbal inflection, case morphology and the acquisition of sentence structure in early successive bilinguals. Paper presented at the 11th Congress of the International Association for the Study of Child Language, Edinburgh, UK. SCHELLETTER, C. (MayJune, 2007). Childrens acquisition of regular and irregular morphology: Differences between monolinguals and bilinguals. Paper presented at the 6th International Symposium on Bilingualism, Hamburg, Germany. SCHWARTZ, M. (2006). The impact of literacy acquisition in L1 Russian on literacy acquisition in L2 Hebrew and in L3 English among Russian-speaking (L1) children: Bi-literate bilingualism versus mono-literate bilingualism. Unpublished PhD thesis [In Hebrew], Department of Education, University of Haifa, Israel. SCHWARZWALD, O. (1991). Grammatical vs. lexical plural formation in Hebrew. Folia Linguistica, 25(34), 577608. SHARE, D. L. & LEVIN, I. (1999). Learning to read and write in Hebrew. In M. Harris and G. Hatano (Eds.), Learning to read and write: A cross-linguistic perspective. (pp. 89111). Cambridge: Cambridge University Press. SUPPES, P., SMITH, R. & LVEILL, M. (1973). The French syntax of a childs noun phrases. Achives de psychologie, 42, 207269

The International Journal of Bilingualism


Downloaded from ijb.sagepub.com by Miren Morales on October 20, 2011

Você também pode gostar