The acquisition of definiteness in Hebrew (L2) by bilingual preschool children (Russian-L1): A longitudinal multiple-case study

Abstract

In this longitudinal study, we examined bilingual children’s acquisition of definiteness in Hebrew (L2). Hebrew distinguishes between definite and indefinite nouns by marking definiteness by article. We recorded the speech of nine bilingual Russian–Hebrew-speaking children ranging in age from three to four years old in monthly sessions over a period of seven months. Their age of onset of acquisition of Hebrew (L2) ranged from six to 42 months. The L1 of all participants was Russian, which has no definite and indefinite articles.

We compared the bilingual children’s acquisition of definiteness in Hebrew to that of monolingual Hebrew-speaking children to outline the qualitative and/or quantitative differences in the patterns found. This comparison also enabled us to determine whether the bilinguals transfer their knowledge from L1 to L2 (transfer hypothesis) or show fluctuation in their article choice (fluctuation hypothesis).

Four tendencies were found with regard to the bilingual children: 1. All bilingual children made omission errors in definite contexts (evidence of transfer); 2. We observed misuse of articles, overuse and syntactic errors in all the children’s utterances (evidence of fluctuation); 3. We found both similarities and differences between bilinguals and monolinguals in acquisition of definiteness; and 4. Children with early age of L2 onset (before age three) demonstrated more accelerated acquisition of definiteness than children with later onset of L2 (after age three).

Keywords

Acquisition of definiteness Hebrew Russian transfer hypothesis fluctuation hypothesis bilingual children age range of second language acquisition

Introduction

The aim of the present study was to examine whether bilingual Russian–Hebrew-speaking children with a wide age range of onset of L2 acquisition (hereafter AOA) from six to 42 months, showed similar or different patterns of acquisition of definiteness in Hebrew (L2) compared to monolingual Hebrew-speaking children. This study aims to answer two questions: 1. What is the role of age of onset of L2 in L2 grammar acquisition? 2. How might the acquisition of definiteness by bilingual children be explained by the transfer hypothesis and the fluctuation hypothesis?

The role of AOA in second-language grammar acquisition

AOA and types of bilingual learners

Many studies over the last two decades have shown the complexity and challenges involved in acquiring grammatical knowledge in L2 (DeKeyser, 2005). The L2 learners need to acquire grammatical knowledge in both languages, including learning to use the grammatical forms productively, and aim to reach a native-like linguistic competence in the second language. In this context, recent research has highlighted the significant role of age of onset of L2 in L2 grammar acquisition (see Meisel, 2009; Unsworth & Hulk, 2009).

AOA is used as a one of the variables that distinguishes between various types of bilingual learners. Recent studies demonstrated the difference between children with simultaneous acquisition of two or more first languages (AOA from birth to age one) (2L1) (De Houwer, 1995; Döpke, 1992); children whose age of first exposure to L2 was between one and four years of age (early sequential bilingual/s – ESB/s) (Meisel, 2008; Rothweiler, 2008; Unsworth & Hulk, 2009); bilingual children with AOA in L2 between four and 10 years of age (cL2), and bilingual adults with AOA in L2 from around age 10 onward (aL2) (Meisel, 2009; White, 2003). It is notable that the distinction of the ESBs group has been proposed only recently in light of growing interest in bilingual development of children who “…are classed as L2 children, which could be interpreted as laying the boundary between … L1 and L2 acquisition at around age 3” (Unsworth & Hulk, 2009, p. 2). Note also that some researchers identify children as sequential bilinguals only if they are exposed to one language exclusively up to the age of four years (e.g. Montrul, 2008). However, other researchers do not differentiate between simultaneous and sequential bilinguals at this age, and do not distinguish the ESBs as a separate group. As was stressed by Unsworth and Blom (2010), making a decision as to which learners belong to which group depends to a large extent on the research question at hand. In addition, when focusing on the ESBs as a novel group of L2 learners, Unsworth and Hulk (2009) stressed a necessity to assume "as many factors as possible, such as methodology, L1/L2 pair, socioeconomic background, quantity and quality of input, are held constant" (p. 2). To what extent the ESBs are considered a unique group and need our particular research attention will be discussed later.

As the target population in the present study is early sequential bilinguals (ESBs), we only briefly address the existing data on the aL2 and cL2 learners and L2 grammar acquisition. Regarding aL2 learners, consistent findings indicate that even after considerable exposure to L2, the aL2 learners did not attain proficiency in some grammatical domains (Alfi-Shabtay, 2006; DeKeyser, 2005; Meisel, 2009), and that they showed significant quantitative and qualitative differences compared to monolinguals (Hulk & Cornips, 2006). For example, Alfi-Shabtay (2006) found that acquiring definiteness in Hebrew (L2) was very challenging for the Russian (L1) speaking immigrant who immigrated to Israel after age seven.

Concerning cL2 learners, the existing data point to qualitative similarities between cL2 and monolingual children in relatively transparent L2 grammar domains; that is, domains with the high extent of clarity of a stem or a suffix (Laaha & Gillis, 2007), e.g. verb inflections and adjectival inflection in Dutch (e.g. Bisschop & Punt, 2006; Blom, 2008; Blom, Polišenská, & Weerman, 2008; Weerman, Bisschop, & Punt, 2006) versus less transparent domains, e.g. acquisition of grammatical gender in Dutch which is difficult to determine based on morphological and semantic cues (Unsworth et al., 2014).

Referring to 2L1 learners, growing evidence indicates that these children acquire L1/L2 grammar principles just as well as their monolingual peers and achieve complete linguistic proficiency in L1/L2 (De Houwer, 1995; Meisel, 2008; Paradis & Genessee, 1996). In addition, their L1/L2 language tendencies are similar to those of monolingual children.

Early sequential bilinguals and second-language grammar acquisition

Unlike the aforementioned groups of L2 learners, the ESB group has been less investigated and it is still not clear how these bilingual children acquire grammar in L2 (Unsworth & Hulk, 2009). Two main characteristics differentiate the 2L1 and ESB groups: the sequence of L2 acquisition and grammatical knowledge acquisition in L1 before L2. This situation frequently occurs within immigrant families, where children are exposed to L2 (the dominant language of the host society) only after entering a preschool educational setting. In this case, the heritage or minority language is acquired first (Montrul, 2008).

With regard to achieving native-like competence in morphology and morpho-syntax in L2, Meisel (2009) suggested two sensitive periods: around ages four to six (cL2) and close to age four (ESB). More specifically, from age one to three years (approximately), acquisition of L2 grammar is relatively easy, followed by a gradual offset when the optimal periods begin to fade out from age four onward (Birdsong, 2005; Meisel, 2009). During this period, the children’s ability to acquire some grammatical sub-components in L2 diminishes, and the trajectory of sub-component acquisition is qualitatively different than that of 2L1 children and monolingual children and resembles that of adult L2 learners. This suggestion received empirical support in L2 morphology acquisition by L1-German-L2-French-speaking children (n = 35; AOA 2;8 and four) (Meisel, 2009). It was found that almost all children with AOA before age 3;6 showed similarities to 2L1 children in clitic forms, whereas the children with AOA 3;7 or older made errors in these forms that were unattested in monolingual or 2L1 acquisition of French. Additional support for Meisel’s claim emerged from Sopata’s (2008) recent longitudinal study, which focused on verbal inflection in L2 German by four Polish–German speaking children (AOA 2;6–4;7, with 10 months of data collection). The findings demonstrated a fading out effect as early as age three.

In another study that focused on gender determiner acquisition for marking of definiteness in Greek by bilingual English–Greek-speaking children, significant differences were found in accuracy rates on masculine and feminine gender between 2L1s and ESBs, whereas ESBs were similar to cL2 learners (Unsworth et al., 2010). Thus, in Greek, the age effect in gender determinant acquisition with masculine and feminine gender started rather early (before age four). Similarly, in the same international project, Unsworth et al. (2010) provided empirical evidence that ESBs showed similar performance to cL2 learners on producing the neuter gender determiner het in Dutch, with the significantly higher level of overgeneralization of the default form de of the common gender. These findings are consistent with Meisel’s (2009) suggestion.

At the same time, some recent studies found that the ESBs showed high accuracy and similarity between 2L1 and monolingual children. For example, Marinis (2007) found similarities between 30 monolingual English-speaking children (aged 6;9 to 8;9 years) and 28 bilingual Turkish–English-speaking children (aged 6;10 to 8;8 years, AOA from three to four years) in processing actives and passives in real time. Like monolinguals, the bilingual children used morphological cues to process actives and passives, and to (re)assign thematic roles. In this case, it is important to note that similar to English (L2), Turkish (L1) distinguishes between active and passive verbs.

In the same vein, Schwartz, Nir, Leikin, Levie, and Ravid (2014) found striking similarities between monolinguals (Hebrew, L1) and ESBs (Russian–Hebrew) with respect to the developmental trajectory of noun pluralization in Hebrew (L2). This similarity was attributed to the positive role of the ESBs’ relative maturity in Russian (L1). Like Hebrew, Russian is an inflectionally rich language in which nouns are marked for singular and plural. To summarize, in cases when L1 and L2 shared similar grammar domains (e.g. gender and plural inflectional marking as in the case of Russian and Hebrew), the ESBs showed acquisition patterns that were similar to those of their 2L1 or monolingual peers.

In conclusion, the empirical evidence is insufficient and inconclusive regarding the question of whether grammar acquisition in L2 is sensitive to AOA up to age four. The picture is complicated because the existing studies dealt with a variety of grammatical properties in L2 (e.g. verb and noun inflectional morphology and determiners as part of a freestanding grammatical morphology). In addition, they focused on second languages with a varying degree of transparency of grammatical properties (verb inflection in L2 as a rather transparent feature versus non-transparent gender agreement on definite determiners in Dutch) and diversity of L1 linguistic background (with and without the target property, e.g. German (L1) as a [–] clitic form language and French (L2) as a [+] clitic form language). Note, that the sign [–] indicates that the grammatical form does not exist in the language and the sign [+] indicates that the grammatical form does exist in the language.

With this theoretical background in mind, the present study focused on the acquisition of a complex definite system in Hebrew (L2) by the Russian–Hebrew-speaking children with a wide range of AOA (six to 42 months). We focused on this system for several reasons. First, definiteness is a fascinating phenomenon that brings together inflectional morphological knowledge in the case of Hebrew, and syntactic, semantic and pragmatic knowledge of all speakers (Abbott, 2008; Armon-Lotem & Avram, 2005; Wintner, 2000). Consequently, investigating definiteness in Hebrew provides the opportunity to observe several levels of linguistic representation. In this case, focus on definiteness takes us one step further in more precise pinpointing of grammatical development in L2 by ESBs: which grammatical features the ESBs manage to develop, which features are similar to their monolingual peers and which remain underdeveloped. Finally, focus on the definitive system acquisition in Hebrew as the L2 is particularly interesting in the context of the Russian–Hebrew dyad because Russian has neither articles nor a definite system.

Definiteness in Hebrew and Russian

Russian and Hebrew belong to two different linguistic families (Slavic and Semitic) and are based on two distinct morphologic and syntactic systems (Maslov, 1984, on Russian). Whereas Russian has a synthetic-inflectional grammar system, Modern Hebrew grammar is both analytic and synthetic (Glinert, 1989). One of the most important differences between these two morpho-syntactic systems is that Russian does not have determiners (and more specifically, the definite determiners). Furthermore, Russian conveys the concept of definiteness by a variety of means. They include: (a) lexical, by usage of demonstrative ( eta interesnaya kniga, “this interesting book;” Ona smotrela v tu storonu, otkuda šhli ludi, “She was looking in the direction from which the people were coming”) or possessive pronouns (moy drug, “my friend”); (b) syntactic, by means of word order, generally by inserting a clause after a subject to reflect its definiteness, e.g. prineci knigu kotoraya lezhít na stole, “bring the book which is lying on the table,” or placing the subject after the predicate to designate that the subject is definite, e.g. kniga lezhít na stole, “there is a book lying on the table;” (c) non-lexical, by means of agreement, stresses, and morphological means; (d) a combination of the above methods; and (e) based on the context in which the definite status of the noun phrase can only be identified based on the discourse context (Alfi-Shabtay, 2006; Avrutin & Brun, 2001; Chesterman, 1991; Lyons, 1999; Wexler, 1976).

Unlike in Russian, the definite article is used in Hebrew, but in a less uniform manner than in English or German (Armon-Lotem & Avram, 2005). Consequently, Hebrew poses a difficult and complex communicative problem for the child (Alfi-Shabtay, 2006; Zur, 1983). As in English, Hebrew distinguishes between definite and indefinite nouns. However, unlike English, only definite nouns are marked with the article. In other words, English has an article for definite nouns (the) and an article for indefinite nouns (a), whereas Hebrew has an article for definite nouns only (prefix –ha).

Example (a) below presents an example of a definite noun in Hebrew. In this sentence, the speaker refers to an object (“book”) that is already known to the hearer (there is a presupposition about the existence of that object). Therefore, the definite noun is marked by the prefix ha-. Example (b) gives an example of an indefinite noun in Hebrew. In this sentence, the speaker refers to an object that is unknown to the listener. The indefinite noun is not marked by the prefix ha-.

(a) קראתי את הַספר, qara‵ti ‵et ha - sefer, “I read the book”

(b) קראתי ספר, qara‵ti sefer, “I read a book”

Unlike English, Hebrew has some morpho-syntactic conditions that must be fulfilled by the speaker (Zur, 1983). First, the noun after the direct object marker must be definite (see example c).

In addition, if the noun is definite, its adjective must also be definite (d) and if the noun is indefinite, its adjective must also be indefinite (e):

(d) ראיתי את הסרט המעניין, ra‵iti ‵et ha - seret ha - me’anyen, “I saw the interesting movie”

(e) ראיתי סרט מעניין, ra‵iti seret me’anyen, “I saw an interesting movie”

Furthermore, if the noun is definite, its demonstrative pronouns must also be definite (f). Likewise, if the noun is indefinite, its demonstrative pronouns must also be indefinite (g):

(f) הַכלב הַזה, ha - kelev ha -ze, “this dog”

(g)כלב זה, kelev ze, “that dog”

Moreover, to specify noun compounds, the article (prefix ha-) must be added before the second nominal in the noun compound:

(h) חנות הַכפר, xanut ha - kfar, “the village store”

Finally, the preposition that comes before the definite noun will receive the article’s vowel (ha-) and the consonant of the article (ha-) will be replaced by the consonant of the preposition (ba-, “in” or la- “to”):

(i) בַּכפר, ba - kfar, “in the village”

(j) לַכפר, la- kfar, “to the village”

The complex syntactic conditions together with the unique semantic–pragmatic role of the definite system in Hebrew have an effect on the acquisition of definiteness by monolingual children (Zur, 1983), as described in the following section.

The acquisition of definiteness by monolingual and bilingual children

Most of the existing data on the acquisition of definiteness by monolinguals and bilinguals was based on English (Hawkins et al., 2006; Ionin, Ko, & Wexler, 2004; Lardiere, 2004; Zdorenko & Paradis, 2008). Most research on the acquisition of definiteness by bilinguals focused on aL2 (e.g. Ekiert, 2004; Zdorenko & Paradis, 2008, 2011) and cL2 populations (Ionin, Zubizarreta, & Philippov, 2009; Zdorenko & Paradis, 2008). These studies were based on two hypotheses with regard to definiteness acquisition in L2, the transfer hypothesis and the fluctuation hypothesis.

The transfer hypothesis

The transfer hypothesis emphasizes the role of L1 in L2 acquisition (Schwartz & Sprouse, 1994, 1996). According to this approach, the initial state of L2 acquisition is the final state of L1 acquisition. In other words, L1 has a major influence on L2 grammar acquisition, and thus, L2 learners are expected to transfer their knowledge of L1 grammar and rules while acquiring L2 linguistic principles. With regard to the cL2, Paradis and Genesse (1996) stressed that the phenomenon of transfer or cross-linguistic influence must be systemic: “By systemic, we mean influence at the level of representation or competence, sustained over a period of time” (p. 3). In addition, Yip and Matthews (2007) highlighted that the phenomenon of transfer must be recurrent “at least in three attestations in the same individual at the stage of development” (p. 38).

In the context of the present study, the role that the L1 may play in early L2 acquisition is particularly considerable in the case of a grammar feature that does not exist in L1 and cannot be transferred to L2 (Yip & Matthews, 2007). In other words, in the L1s which do not have an article system for the definiteness marking such as, for example, Russian and Japanese, this grammar feature cannot be transferred to L2. Up to now, most of the studies have found transfer effects in the acquisition of definiteness especially by aL2 learners (Morales, 2011; Salehuddin, Hua, & Maros, 2006; Sarko, 2008; Snape, 2005). At the same time few studies have focused on a cL2 population. Morales (2011) examined 30 Spanish [+article] speaking children (mean age = 9) learning English as L2 (mean length of exposure was 4.6 years), and living in Puerto Rico, and five native-English speakers (mean age = 9;2), living in Illinois. The elicited writing task was applied to examine the production of definiteness in English. The results suggested that the child L2 learners were affected by definiteness only in their article choice, or in other words, the Spanish-speaking children were able to transfer their L1 semantic knowledge to the L2, and thus resembled adult L2 learners (with an article system in their L1s).

This pattern of results is consistent with the data reported by Zdorenko and Paradis (2008, 2011), who also found evidence supporting the transfer hypothesis in bilinguals from L1s with no definite articles. The children whose L1s lacked articles made more omission errors than children with [+article] L1s. However, by the end of this longitudinal study, the bilingual children with [–article] L1s were more accurate in article use. The researchers concluded that L1 transfer is limited and occurs most frequently at the early stages of L2 acquisition. To summarize, the findings presented above indicated evidence of transfer effect in the case of L1s with a definite system.

The fluctuation hypothesis

This hypothesis assumes that L2 learners of [–article] L1 languages sometimes divide articles based on definiteness and sometimes based on specificity setting with regard specifically to English L2. This division was defined by Ionin, Ko, and Wexler (2004) as the article choice parameter. In this context, it should be noted that in some languages, article choice is constrained by the definiteness setting (e.g. in English, the definite article, the, is used for definite nouns and the indefinite article, a, is used for indefinite nouns). In other languages, articles are organized according to the specificity setting (e.g. in the Samoan language, the article, le, is used for specific nouns and the article, se, is used for non-specific nouns).

Access to the semantic universals (Morales, 2011), any aspect of meaning or property of the semantic subsystem that is represented in all languages (Talmy, 1985), should mean that L2 learners fluctuate between these two settings until they are exposed to sufficient input to lead them to set the parameter correctly (e.g. definiteness in English or specificity in Samoan). Definiteness refers to “making reference to the common ground between speaker and hearer” (Ko, Ionin, & Wexler, 2009, p 4), whereas specificity refers to knowledge known only to the speaker.

In contrast to English L2, the Hebrew language doesn’t distinguish between the two markers: the for [+definite] and a for [–definite]. In Hebrew, there is only one definite marker (-ha). When a noun phrase is [–definite], the L2 learner should not use the definite marker (-ha), while if the context is [+definite], then the definite article, (-ha) should be used. In this context, in Hebrew L2 the fluctuation hypothesis can be presented in the sustained variations of errors of definiteness. These variations demonstrate that the cL2 learner fluctuates between two parameters of choice of settings in definiteness: [–article] and [+article]. Thus, one can expect that the sustained and systemic overuse in indefinite contexts (as shown in multiple cases) presents a fluctuation effect in Hebrew L2.

Evidence for fluctuation hypothesis in English L2 learners

Over the past decade, studies found support for the fluctuation hypothesis with regard to cL2 and aL2 (English as L2) without a definite system in their L1s (for cL2, see Zdorenko & Paradis, 2008 and for aL2, see Ionin, Zubizarreta, & Maldonado, 2008; Lardiere, 2004). In another study, Zdorenko and Paradis (2008) examined a longitudinal corpus of narratives (two years) of 17 bilingual children with and without an article system in their L1s, living in Canada: six Mandarin-L1 [–article] children, two Cantonese-L1 [–article] children, three Spanish-L1 [+article] children, one Romanian-L1 [+article] child, one Japanese-L1 [–article] child, one Korean-L1 [–article] child and three Arabic-L1 [+article] children. The children’s mean age was 5;4 years and mean exposure to English was nine months at the onset of the study. The participants were asked to tell a story while looking at pictures (“elicited storytelling game”). The researchers found that all children overused the definite article the (instead of the indefinite article a) in specific indefinite contexts. In addition, the children were more accurate in their use of the definite article the in the definite contexts than with the indefinite article, a, in the indefinite contexts, regardless of L1 backgrounds. The misuse errors decreased with increased exposure to English (L2 input). In sum, these findings provided evidence for the fluctuation hypothesis and demonstrated that L1s (whether with or without articles) have limited influence on children’s L2 acquisition of definiteness, and this apparently occurs only at the early stages of acquisition.

To recap, the findings on cL2 and aL2 in acquiring definiteness in L2 emphasize the importance of the L1 background. More and more evidence is demonstrating support for the fluctuation hypothesis with regard to bilingual children and adults without an article system in their L1s, whereas bilingual children and adults with a definite system in their L1s will apparently transfer their knowledge to L2. Considering very young bilingual children (i.e. 2L1 and ESB), research examining these two hypotheses (the fluctuation hypothesis and the transfer hypothesis) is lacking. Therefore, the focus of the present study was to examine the transfer hypothesis and the fluctuation hypothesis as two possible explanations for acquisition of definiteness by young bilingual children with sequence of AOA ranging from six to 42 months (i.e. 2L1s and ESBs) and with Russian-L1 [–article], and to demonstrate the trajectory of acquisition of definiteness in Hebrew among them. Our question is whether very young bilingual learners, coming from L1-Russian [–article], will show no sensitivity to the determiner category in L2-Hebrew (i.e. will show more omissions in definite contexts than other article errors, supporting the transfer hypothesis) or will fluctuate between different parameters until they receive sufficient input (i.e. will make both types of article errors, omissions and overuses, in different contexts).

The acquisition of the definite article in Hebrew by monolingual children

The acquisition of the definite system in Hebrew has not been widely examined to date (Armon-Lotem & Avram, 2005). In Zur’s (1983) cross-sectional study on children (ages two to 12), a U-shaped curve was found as a reflection of children’s acquisition of definiteness. Zur (1983) analyzed children’s spontaneous utterances using different elicitation tests and described their errors. The findings demonstrated that at an early age (up to age four to five), children use articles in their utterances and make many errors, such as misuses and omissions. However, these errors decreased over time (especially after age six). These findings indicate that age five is critical for the acquisition of the definite system in Hebrew. After this age, children have sufficient knowledge of both syntactic and semantic–pragmatic rules that help them understand the meaning and the purpose of definiteness. In addition, Zur (1983) demonstrated specific errors typical to each age group. These errors included misuse of definiteness, omissions and difficulties in establishing the syntactic rules of definiteness in Hebrew (see sub-section “The acquisition of definiteness by monolingual and bilingual children”).

In a cross-sectional study, Armon-Lotem and Avram (2005) examined acquisition of definiteness by 32 Hebrew-speaking children aged two to five. The data were collected using an elicitation task (i.e. the child listened to stories and had to complete them by identifying the objects and using definite articles). Similar to the data obtained about English (L1) acquisition, the findings indicated that two-year-old children overgeneralized the definite article to indefinite referential contexts. Around age three, the children started to understand the pragmatic concept (i.e. distinguishing between the hearer’s and listener’s knowledge) and, therefore, to use the definite article only in definite referential contexts. However, as opposed to acquisition of definiteness in English, where the definiteness system stabilizes both pragmatically and syntactically by the age of three, Hebrew-speaking children continued to omit the definite article until age four. The researchers argued that the complexity of the definite system in Hebrew accounts for the difference between L1-English and L1-Hebrew speakers in the process of acquisition of definiteness. That is, in Hebrew, a different pragmatic principle (than in English) needs to be applied, such as the concept of shared knowledge (speaker’s and hearer’s mutual knowledge and beliefs), to master the definite system. Consequently, Hebrew-speaking children master the definite system only after they acquire the concept of shared knowledge, which happens later (at school age).

As mentioned above, the acquisition of definiteness in Hebrew is a complex and challenging task for native speakers. An interesting question is how L2 bilingual children and adults (with or without a definite system in L1) acquire definiteness in Hebrew. It is surprising that only few studies have examined this question among bilingual children and adults, as presented earlier in this section.

The acquisition of definiteness in Hebrew by bilingual children

There is only one study, that of Armon-Lotem (2005), which examined the acquisition of definiteness in Hebrew by children (aged 10 to 12) with Russian-L1 [–article]. The children were divided into three groups according to different AOAs (three, four-and-a-half and six) and were compared to monolingual age-matched controls using two tasks (a yes/no judgment task and a picture elicitation task). The findings indicated that AOA had a more significant influence on the successful acquisition of L2 definiteness than the length of residence in the host country, difference in learning style or difference in task. Children exposed to Hebrew at the age of three scored significantly higher than those exposed later (age four-and-a-half and six), and only marginal differences were found between them and their monolingual peers. In contrast, children with AOA 4;6 and after showed considerable quantitative differences compared to their monolingual peers (negative correlations were found between success and age). These patterns of results cannot support Meisel’s (2009) claim that the first sensitive age for acquisition of grammar by L2 learners is actually between age three and four because the range between three and four of AOA was not included in the scope of this study. In addition, Armon-Lotem (2005) analyzed only the frequency of errors and did not address qualitative differences between the bilingual and monolingual peers. The lack of the qualitative analysis of the errors could be partly attributed to the methodological limitation of this study, namely, its cross-sectional as opposed to longitudinal design. The cross-sectional design does not allow more than one observation of the same child regarding the target linguistic phenomenon, i.e. error pattern. Consequently, we cannot judge the relative frequency and stability of this phenomenon (Yip & Matthews, 2007), and hence, we cannot reach conclusions about patterns of acquisition of a specific domain of grammar. In the present study we applied a longitudinal design, which gave us the opportunity of observing the relative frequency of various phenomena, as well as the opportunity to determine their “stability” (Montrul, 2008). In addition, the study by Armon-Lotem (2005) focused only on Russian–Hebrew-speaking children with AOAs after three (three, four-and-a-half and six), whereas in the present study, we examined the definite system with regard to the simultaneous and early sequential acquisition of L2 with a wide range of AOAs from age six to 42 months. As far as we know, the current study was the first time that the patterns of acquisition of definiteness within multiple cases and longitudinal design in Russian and Hebrew were explored.

Research aims and questions

The study tried to achieve a number of aims. The first was to examine the acquisition of definiteness in Hebrew (L2) among bilingual preschool children with Russian as L1, in light of the two hypotheses: the transfer hypothesis and the fluctuation hypothesis. The second aim was to examine the effect of age of onset of L2 on the L2 acquisition of definiteness system in light of growing evidence emphasizing the role of age in L2 grammatical knowledge acquisition (Lardiere, 2004; Meisel, 2009; White, 2003; Zdorenko & Paradis, 2011). This aim was examined in light of Meisel’s (2009) hypothesis on gradual fading out effects in AOA as early as between age three and four years. As only a few studies have focused on age one to four (the ESB group versus 2L1) grammar acquisition in L2, we examined children with a wide range of AOAs (six months to 42 months).

Research questions:

Do bilingual children with Russian L1 [–article] show quantitative and/or qualitative differences from monolingual Hebrew (L1) speaking children in the acquisition of definiteness in Hebrew L2 [+article]?

Do early bilingual children with Russian L1 [–article] show L1- transfer (only omission errors) or fluctuation (overuse in indefinite contexts) or both transfer and fluctuation effects in their L2 acquisition of definiteness? If both phenomena are evident are there any differences in frequencies of appearance?

Do children with different AOAs show quantitative and/or qualitative differences in the production of definiteness?

Method

Participants

The acquisition of the definite system in Hebrew was studied during one academic year (seven months) among nine preschool bilingual children. The children were between three and four years old at the onset of the study (M age in months = 43.11, SD = 4.1), and their AOAs ranged from six to 42 months. The children’s mean length of utterances in Hebrew (L2) ranged between 1.3 and 4.8. All participants were Israeli-born children from families of Russian-speaking immigrants (aged 30 to 42 years) from the former Soviet Union. Russian was the first language of all children and their parents, as well as the dominant language in parent–child communication. The parents reported that all children acquired Russian (L1) with no observable delay (onset of L1 ranged from eight to 21 months, see Table 1).

Table 1.

Background data of the bilingual children (n = 9).

Child’s name Variables	FIM	UMK	MIC	JON	MAS	DAN	UMA	AVI	YAI
Gender	Male	Female	Female	Male	Female	Male	Female	Female	Male
AOA in the preschool period (in months)	6	9	13	21	36	39	39	40	42
Age at the beginning of the study (in months)	42	40	41	40	42	44	42	52	48
Age at onset of L1 (in months)	18	17	18	19	11	21	13	8	11
Mother’s education (in years)	12	21	21	18	19	16	15	18	13

AQA: Age of second language acquisition.

In all families, both parents were bilinguals, and reported relatively high language competence in Russian and Hebrew. The length of residence in Israel was fairly long (M = 13.3 years, SD = 4.8). Mothers’ self-reported education level was relatively high: all mothers had at least a high school diploma (10 years of education in the former Soviet Union) and seven mothers had a university degree (a minimum of five years of education; M = 18.28 years, SD = 2.28). All families reported medium–high socioeconomic status.

All parents also reported “first language first” practice at home. All parents and children cooperated willingly with the researchers. Six children (three girls – UMA, MAS, AVI; and three boys – DAN, YAI, JON) were enrolled in a bilingual Russian–Hebrew-speaking preschool located in northern Israel. The explicit aim of this institution was to promote Russian-L1 maintenance and Hebrew-L2 acquisition. It is important to point out that this particular kindergarten adheres to a “first language first” approach until the age of three years, so that all teacher–child communications and instructions are conducted in Russian (L1) and by a Russian-speaking teacher. From the age of three to five years, intensive immersion in Hebrew is conducted by a native Hebrew-speaking teacher. During this period, Hebrew is used for most of the classroom time (70%). The day is divided into periods of instruction, and only one language is used in each period. Although this is a private institution, it functions under the supervision of the Ministry of Education and teachers apply the same curriculum as in the monolingual kindergartens.

Three children (two girls – UMK, MIC; and one boy – FIM) attended a Hebrew-speaking kindergarten (day care) located in northern Israel, where all teacher–child communication and instruction is conducted in Hebrew. This kindergarten includes monolingual Hebrew-speaking children and bilingual Russian–Hebrew-speaking children.

Tools and procedure

Bilingual corpus and testing

The data on the bilingual children were obtained from a long-term study (Schwartz et al., 2011). The bilingual children’s speech was recorded in monthly sessions over a seven-month period, about 20 minutes per month per child (overall, approximately 140 minutes per child). The recordings took place in a separate, quiet room in the kindergarten. The data were collected during the morning hours, following the communal session with the Hebrew-speaking kindergarten teacher, to ensure that the target language was primed. At the beginning of each session, the children were asked to talk about their previous weekend’s activities (spontaneous speech). Later, a game was organized in such a way that the children were encouraged to produce utterances with definite nouns (partly-elicited task). All sessions consisted of free play with games, toys and books that were available at the kindergarten (and thus familiar to the children) and that were used at the discretion of the investigator based on her familiarity with each individual child.

Each child was recorded separately by a research assistant (a native Hebrew-speaking graduate student), yielding a spoken corpus of child speech consisting of 13,913 words across 4,967 utterances.

Data analysis

The obtained data were analyzed for each bilingual child with regard to the following categories: (a) frequencies of production of definite articles, (b) contexts of production (morpho-syntactic, semantic–pragmatic, and imitations), (c) types of incorrect responses (overuse and omission) and (d) tendencies after seven months.

Frequencies of production of definite articles

The child -investigator interactions were transcribed, coded and analyzed by CHILDES conventions and programs (MacWhinney, 2000). Frequencies of production of definite articles were collected and calculated (in percentages) for each child individually over a 7 month period.

Contexts of production

We analyzed each bilingual child’s definite utterances in accordance with the methodology elaborated by Fruchtman (1979), specifically for the analysis of the definite system in Hebrew. Definite nouns can be used in some specific linguistic contexts in Hebrew. Fruchtman (1979) collected and presented all specific contexts of the definiteness production in Hebrew, using a definite marker (-הַ, ha-, “the”) or other phrases (such as pronouns, e.g.זה, ze, “this”). Based on Fruchtman’s (1979) classification, all definite utterances of each bilingual child in the present study were collected and classified according to three main definite contexts: morpho-syntactic contexts, semantic–pragmatic contexts and imitation contexts. The morpho-syntactic contexts included all production of definite articles by each child that related to the grammatical rules and terms (see Table 2).

Table 2.

Analyses of definiteness: productions of definiteness in morpho-syntactic contexts.

Morpho-syntactic contexts	Examples
Using definite nouns before direct objects with the preposition et and the prefix –ha (‵et ha-)	ראיתי את הסרט, ra‵iti ‵et ha - seret, “I saw the movie”
Using definite adjectives after definite nouns	הסרט המעניין, ha - seret ha - me’anyen, the movie the nteresting, “the interesting movie”
Using definite pronouns after definite nouns	הסרט הזה, ha - seret ha - ze, the movie the this, “this movie”
Using definite construct states	חוף הים, xof ha - yam, shore the sea, “the sea-shore”

The semantic–pragmatic contexts included all production of definite articles by each child, involving semantic–pragmatic knowledge (see Table 3).

Table 3.

Analyses of definiteness: productions of definiteness in semantic–pragmatic contexts.

Semantic–pragmatic contexts	Examples
Using definite nouns that were mentioned earlier in the conversation (mutual knowledge)	אתמול ראיתי סרט, הסרט היה מעניין, ‵etmol ra‵iti seret, ha - seret haya me’anyen, “yesterday, I saw a movie, the movie was interesting”
Using definite nouns that were mentioned during a game or seen in a picture – known both to investigator and child (mutual knowledge)	זה הפיל הגדול, ze ha - pil ha - gadol, this is the big the elephant, “this is the big elephant”
Using definite locative nouns known both to investigator and child	הגן, ha - gan, “the kindergarten”
Using definite temporal nouns known both to investigator and child	היום בערב, hayom ba - ‘erev, “today in the evening”
Using definite compounds	דמבו הפיל, Dumbo ha - pil, “Dumbo the elephant”

The imitation contexts included all children’s repeated definite utterances that were produced by the investigator:

*INV: aval hi lo’ po ba-gan naxon?

(“But she isn’t here in the kindergarten, right?”)

*CHI: hi lo’ po ba-gan.

(“She isn’t here in the kindergarten”) (MIC, MAY)

Types of errors (overuse, omission, and syntactic errors)

To examine the two hypotheses concerning acquisition of definiteness within a bilingual context (the transfer hypothesis and the fluctuation hypothesis), we also analyzed each child’s definite errors and classified them into three different categories: overuses, omissions and syntactic errors (see Table 4). These categories were used based on previous studies by Zur (1983) and Armon-Lotem and Avram (2005).

Table 4.

Definite error analysis.

Definite uses and errors	Examples
Overuses of the definite article	CHI: ואיפה עוד הסוס? ve efo od ha-sus?* [And where else is there the horse?] (“And where else is there a horse?”) (UMK, APRIL)
Omissions of the definite article	CHI: איפה הכלב קטן? efo ha-kelev katan [: ha- katan] []? [Where is the little the dog?] (“Where is the little dog?”) (JON, MARCH)
Definite syntactic errors (definite construct states’ or prepositions’ misuses)	CHI: להבית le- ha- bayit [: la- bayit] []. (“To the house”) (AVI, APRIL)

Tendencies after seven months

To examine the improvement relating to definiteness of each child over time we also created a new measure, which we called the discrepancy score. For each child individually, we calculated the discrepancy between the correct uses of definiteness at the last data collection point (the seventh month of the study) and the correct uses of definiteness at the first data collection point (the first month of the study). Thus, a positive discrepancy score meant that the child’s usage of definiteness increased with time, whereas a negative discrepancy score meant that the child’s correct usage of definiteness decreased with time.

Results

The results will be presented in two sub-sections: (1) The acquisition of definiteness of the bilingual Russian-Hebrew speaking children and the monolingual Hebrew-speaking children; and (2) A within-group comparison: The link between age of L2 onset and acquisition of definiteness by bilingual children. First, we will address our first and second questions by analyzing the production of definite articles by the nine bilingual children with regard to existing data on monolingual Hebrew-speaking children. In analyzing bilingual and monolingual data, we will also discuss differences between our bilinguals with earlier and later AOA. Second, we will examine our third question (Do children with different AOAs show quantitative and/or qualitative differences in the production of definiteness?) by focusing only on differences between bilingual children with a wide range of AOAs (from age six to 42 months).

The acquisition of definiteness of the bilingual Russian–Hebrew-speaking children and the monolingual Hebrew-speaking children

To analyze the differences and similarities between L2 acquisition of definiteness of the monolingual and bilingual children, we present the development of the definiteness by monolinguals first (see Table 5), based on data obtained by Zur (1983) from speech elicitation tasks.

Table 5.

Monolingual acquisition of definiteness: descriptive data (according to Zur, 1983).

Age\definiteness use	Frequencies and types of definiteness	Overuses	Omissions	Unique pattern
1;10–3;6	Most of the children used the definite article between the ages of 2;6–3;0. However, they made lots of errors since their definite competence and knowledge were not full and complete. In addition, they used more articles in morpho-syntactic than semantic–pragmatic contexts, since their pragmatic competence was not complete or fully developed.	33% of the definite articles among children’s spontaneous speech and definite tests were overuses.	40% of the definite articles among children’s spontaneous speech were omitted. Lots of the omission errors were in definite noun contexts.	Children omitted the definite article after the direct object marker (את, ‵et-). Children used more definite prepositions in a synthetic way (בַּ-, ba-, “in the”) rather than in an analytic one (בתוך ה-, betox ha-, “in the”).
3;6–4;6	After the age of 3;6, 95% of the production of definite articles in children’s spontaneous speech and definite tests were corrected. In addition, they used more articles in morpho-syntactic than semantic–pragmatic contexts, since their pragmatic competence was not complete or fully developed.	11% of the definite articles in children’s spontaneous speech and tests of definiteness were overuses.	–	Children used more definite noun compounds after age 3;6. In addition, children produced more direct object markers (את, ‵et-) as well as specifying the nouns that followed as definite.
5;0–6;0	The age of 5;0 was considered a critical age to comprehend and organize the definiteness system. After age 5;0, children understood the syntactic and pragmatic functions of the definite article and their definite uses were more correct.	–	7% of the definite articles in children’s spontaneous speech and in definiteness tests were omitted.	Children aged 5;0–8;0 tended to use a shortened form of the direct object (- תַּ, ta-).
7;0–8;0	At the ages of 7;0–8;0, children seemed to understand the pragmatic contexts of the definiteness and external circumstances (previous knowledge).	0.8% of the definite articles in children’s spontaneous speech and definiteness tests were overuses.	–

As can be seen in Table 5, most of the younger children used the definite article (between the ages of 2;6–3;0). However, they made a lot of errors since their definite competence and knowledge were neither full nor complete. After the age of 3;6, 95% of the production of definite articles in children’s spontaneous speech and definite tests were correct. As for definite types, children used more definite noun compounds (after age 3;6) and produced more direct object markers (את, ‵et-) as well as specifying the nouns that followed as definite. The age of 5;0 was considered a critical age to comprehend and organize the definiteness system. After age 5;0, children understood the syntactic and pragmatic functions of the definite article and their definite uses were more correct. Their definite mistakes were significantly reduced (below 8%).

In sum, the monolingual children tend to use more definite articles as they get older. With an increase in age, the monolingual children made fewer errors and their definite knowledge and competence were more target-like and complete. In addition, the monolinguals made more omission errors than overuses at all ages, and with the increase in age became more proficient in using definite nouns in semantic–pragmatic contexts.

Table 6 summarizes the L2 acquisition of definiteness of each bilingual child arranged by AOA. More specifically, in this table, the production of definite articles of each bilingual child is analyzed in different contexts of production (morpho-syntactic, semantic–pragmatic, imitation), and with regard to different types of errors (overuse, omission) and tendencies.

Table 6.

Frequencies of definiteness production and types of errors among bilingual children by child and AOA.

	FIM	UMK	MIC	JON	MAS	DAN	UMA	AVI	YAI
AOA	6	9	13	21	36	39	39	40	42
Frequencies of definiteness (%)	14.61	11.42	11.72	9.2	3.51	29.75	11.95	11.99	10.76
Overuses of definiteness (%)	3.64	5.45	5.76	4.71	0	5.35	7.09	0.76	11.86
Omissions of definiteness (%)	3.64	12.12	13.46	13.2	3.44	8.27	9.21	13.07	10.16
Semantic–pragmatic uses of definiteness (%)	42.33	35.75	35.89	38.67	65.51	41.11	46.8	40.76	33.89
Morpho-syntactic uses of definiteness (%)	48.9	42.42	42.94	38.67	24.13	41.6	21.98	40	39.83
Imitations (%)	1.45	3.03	0.64	1.88	6.89	2.18	14.89	3.07	4.23
Syntactic definite errors (%)	0	1.21	0.64	2.83	0	1.45	0	1.53	1.69
Discrepancy score^a (Mon7–Mon1)	18.89	11.02	–2.48	26.68	0	15.39	5.78	–20	–30.9

Discrepancy score = subtraction between the correct use of definiteness at month seven and the correct use of definiteness at month one in percentages (%).

The percentages in Table 6 were calculated per child per month (seven months) – relating to utterances where an article was required and then summarized. The percentages of the frequencies of definiteness were calculated per month per child, relating to all child utterances, and then summarized.

AOA: Age of second language acquisition.

The similarities between bilinguals’ and monolinguals’ acquisition of definiteness

As can be seen in Tables 5 and 6, there were a number of similarities between monolinguals’ and bilinguals’ acquisition of definiteness patterns with regard to error patterns and contexts of production. These similarities were observed among all nine bilinguals regardless of their AOA. Thus, similar to the monolingual peers, the bilingual children made more errors of omission than errors of overuse. In addition, the bilingual children used more definite articles in morpho-syntactic contexts than in semantic–pragmatic contexts, as monolinguals did before age five. Moreover, the bilingual children preferred the synthetic form of the definite preposition over the analytic one (e.g. ba - bayit, “in the house;” instead of betox ha - bayit, “inside the house”). This linguistic behavior resembled that of the monolingual children, as described by Zur (1983). Finally, as in the case of the monolinguals’ acquisition of definiteness under age 3;6, the bilinguals hardly used definite noun compounds in their speech.

It should be noted, however, that some interesting tendencies were found within the bilingual group regarding the range of AOA in comparison to the monolinguals’ tendencies (see Tables 5 and 6).

First, the bilinguals with earlier AOAs (up to age three) used more definite articles in morpho-syntactic contexts than bilingual children with later AOAs. In addition, the monolingual children used more definite articles in morpho-syntactic contexts than in semantic–pragmatic contexts.

In addition, the bilinguals with earlier AOAs (up to age three) used more direct object markers and definite nouns than bilingual children with later AOAs. In this way, they were more similar to the monolingual children over age 3;6. Finally, similar to the findings in monolinguals over age five, the bilinguals with earlier AOAs (up to age three) used shorter forms of the direct object marker (ת-/ta-).

The differences between bilinguals’ and monolinguals’ acquisition of definiteness

In the present study, we also found clear-cut differences between the bilinguals’ and the monolinguals’ patterns of definiteness production.

First, regardless of children’s AOAs, the bilinguals tended toward correct usage of definite nouns after direct object markers (e.g. ‵et ha - kadur, “the ball”). Thus, we can assume that the bilingual children acquired the syntactic rule in Hebrew (i.e. the definite article must appear after the direct object marker), and used it in their utterances. In contrast, the monolingual children up to age 3;6 omitted more definite articles after the direct object marker (e.g. ‵ani ‵ohevet ‵et xamor, “I love donkey”), and younger monolinguals even tended to omit the direct object marker completely before the definite noun (e.g. ‵ani ‵oxel tapuax, “I’m eating an apple”). In this case, the bilingual children seemed to acquire the object word +definite noun even before the monolingual children, who acquired it correctly only after age 3;6.

In addition, as was reported earlier (Table 5), between the ages of 1;10 and 3;6, the monolingual children made omission errors and especially omitted definite nouns (e.g. sefer ha - gadol, “book the big” – only using the definite article before the adjective, whereas in Hebrew, the definite article should also have been used before the noun). In the same vein, regardless of the bilinguals’ AOAs, they also omitted definite nouns. However, unlike the monolinguals, this also happened with adjectives (e.g. ha -kelev qatan, “the dog small”). These omission errors can indicate a transfer effect because Russian has no determiner category. Consequently, the bilinguals have no linguistic representation of definiteness (neither definite nouns nor definite adjectives) in their L1 to transfer to L2, and thus they show difficulties in its acquisition at early stages. (This will be discussed in depth in the Discussion section.) In addition, this pattern indicates that the bilinguals did not acquire the syntactic rule of noun–adjective agreement on the definite/indefinite article.

A within-group comparison: the link between age of L2 onset and acquisition of definiteness by bilingual children

The differences in production of definite articles between the bilingual children with different AOAs are demonstrated in Figures 1 and 2 and Tables 6 and 7.

Figure 1.

Frequencies of definiteness (FD) in percentages as observed in the child’s speech by age of acquisition.

Figure 2.

The discrepancy score of production of definite articles for each bilingual child arranged by age of acquisition.

Table 7.

The qualitative and quantitative differences within-group: under and over AOA 3.

Groups\differences	Description of differences
AOAs before age 3 (FIM, UMK, MIC, and JON)	• The definite discrepancy score tended to be more positive (from 18.89% to 0%), meaning that the bilingual children used definite articles more correctly over time (relating to earlier and later observations – comparing month 7 to month 1).
	• Overall, the bilingual children hardly used definite noun compounds in their utterances. However, under AOA 3;0, we found a higher production of definite noun compounds than over AOA 3;0.
	• An interesting finding was the shortened use of the definite direct object marker (-’ת/ta-) that was seen only among bilingual children with AOA before 3;0 and also in DAN’s utterances.^a
AOAs age 3 and after age 3 (MAS, DAN, UMA, AVI and YAI)	• The definite discrepancy score (after AOA 3;0) tended to decrease and became negative over time – relating to earlier and later observations, comparing month 7 to month 1 (from 0% at AOA 3;0 to −30.9% at AOA 3;6).

The production of compounds and the shortened use of definite direct object words (-‵ת/ta-) were observed in DAN’s utterances even though his AOA was over age 3. These findings may be attributed to DAN’s L2 language preference and substantial amount of L2 input at home, which are presented above.

AOA: Age of second language acquisition.

First, we found that the children with earlier AOAs showed a more positive discrepancy score; that is, more correct production of definite articles in the child’s speech over time, as opposed to children with later AOAs. For example FIM, with the AOA of six months, had a positive discrepancy score (18.89%), whereas YAI with the latest AOA (42 months) had a negative one (–30.9%).

At the same time, the bilinguals with both early and late AOAs showed similarities in definite frequency and in types of errors (omissions and overuse).

Surprisingly, the children with later AOA (after age three) showed a higher frequency of production of definite articles in the semantic–pragmatic context compared to the children with earlier AOA. The average gap between the semantic–pragmatic uses of definiteness and the morpho-syntactic uses of definiteness increased as AOA increased. We found that the average gap between the semantic–pragmatic contexts and the morpho-syntactic contexts among the bilinguals with AOA before age three is 5.07 as opposed to 12.1 among the bilinguals with AOA after age three.

Discussion

The first aim of this study was to examine how long young bilingual children with L1-Russian, which has no definite articles, take to acquire the L2-Hebrew definite system compared to monolingual Hebrew-speaking children. The second aim of the study was to investigate the effect of age of L2 onset on acquisition of the definiteness in L2 by children with a wide range of AOAs (age six to 42 months). This study adds to the existing literature in three ways. First, the longitudinal multiple-case study supported both the full transfer and the fluctuation hypotheses on L2 acquisition of definiteness by young bilinguals with Russian L1 [–article]. Second, we found both similarities and differences between bilinguals and monolinguals in acquisition of definiteness. Finally, concerning Meisel’s (2009) hypothesis relating to gradual fading out effects in AOA, which are already apparent between the ages of three and four years, our findings only partly support the hypothesis.

The fluctuation and transfer hypotheses with regard to acquisition of definiteness

Our data support both hypotheses on acquisition of definiteness: the transfer hypothesis and the fluctuation hypothesis. It appears, therefore, that these two hypotheses complement rather than contradict each other in reflecting the developmental processes as a possible explanation of acquisition of definiteness by the participants.

On the one hand, our findings demonstrate transfer effect, supporting the transfer hypothesis. Thus, the L1-Russian [–article] has a significant effect on L2-Hebrew definiteness acquisition. All bilingual children made more omission errors than any other type. In the case of Russian as a language without articles in its system, according to the transfer hypothesis, absence of definiteness or omission errors in bilinguals’ speech was expected and observed. These data are in line with the results reported by Zdorenko and Paradis (2008, 2011), who also found evidence supporting the transfer hypothesis in bilinguals from L1s with no definite articles. Similar to the conclusions of Zdorenko and Paradis (2008, 2011), our data suggest that L1 transfer occurs most frequently at the early stages of L2 acquisition.

An additional similarity to Zdorenko and Paradis’s (2008, 2011) data was our finding of a fluctuation effect among our bilinguals’ production of definiteness. All bilingual children used definite articles in their utterances and fulfilled some of the L2 grammatical rules (e.g. definite nouns after direct object markers). In addition, the bilingual children had linguistic sensitivity to L2 definiteness in their speech and they made article misuse errors besides omissions (overuses in indefinite contexts and other syntactic errors). However, it is important to stress that in this age range the fluctuation evidence in cL2 learners’ production of definiteness is considerably less frequent than transfer evidence. It may be considered that growing exposure to Hebrew L2 will lead to fluctuation effects in the production of definiteness and the reduction of omission errors. This is due to the assumption that for the young bilingual children, access to Universal Grammar is more efficient than it is for the older L2 learners, who rely more on the grammatical rules of their L1 (Zdorenko & Paradis, 2007). However, further investigations should be made in examining whether these children will continue to fluctuate until they establish a new language-specific grammar while receiving sufficient L2 input. In addition, further longitudinal research is needed to better understand the nature of the L2-definite developmental process: whether young bilinguals – ESB [–article] L1s − first transfer their knowledge from L1s until they receive some exposure to L2 and then demonstrate more fluctuation patterns than transfer in their speech.

The similarities and differences between bilingual and monolingual children in acquiring definiteness in Hebrew

In our study, we found some qualitative similarities between the bilingual children and natives in acquiring definiteness. Similar to Hebrew-speaking monolingual peers up to age five (Zur, 1983), all bilinguals made more omission errors than other types (such as overuse or syntactic errors). In addition, they hardly used definite compound nouns in their utterances.

At the same time, there was also evidence of clear-cut differences between the bilinguals’ and the monolinguals’ patterns of definiteness acquisition. First, we found a clear tendency among all bilinguals for omissions of definiteness in morpho-syntactic contexts (e.g. omission of the definite article in noun–adjective definite agreement: הכלב קטן, ha -kelev qatan, “small the dog” instead of הכלב הקטן, ha -kelev ha -qatan, “ the small dog”) in contrast to what was evidenced among monolinguals (Zur, 1983). This tendency can be explained by the complexity of this context because a child needs to decide each time whether noun + adjective are definite or indefinite in the specific context, as opposed to nouns that are always definite after the object word את, ‵et. This complexity is even more challenging for our bilinguals because of two possible complementary explanations: less L2 input and L1 background (i.e. lack of definite system in Russian).

Second, all bilingual children showed a tendency for early acquisition of the rule in Hebrew (i.e. definite noun after direct object marker את, ‵et), and produced it without any errors, as opposed to the monolinguals under age 3;6. The accelerated mastery of this role might be attributed to the bilinguals’ experience with the Russian L1, which is characterized by a rich and complex case marking system, and specifically with the case which marks the direct object, the accusative case. Similarly to the direct object marker את, ‵et in Hebrew, the accusative case in Russian might be used after the prepositions в v (to, into) and на na (on, to) when they indicate motion towards something. In general, the accusative is formed from the nominative singular nouns with the help of the following endings: -у, u, -ю, ju, -а, a, -я, ja (singular); -ей, ej, -ой, oj, -ов, ov, -ев, ev (plural). The basic transparent forms of this system are adopted by monolingual Russian-speaking children around age two (Ceytlin, 2000). To summarize, the grammatical experience in the young bilinguals’ L1 cannot be ignored and may play a significant role in their L2 acquisition (Paradis, 2008).

Third, the analysis of bilingual productions showed that the bilinguals tended to use more imitation in their speech as a strategy to acquire L2 grammar. This is consistent with other studies that have demonstrated repetition or imitation as a developmental process to acquire language (L2) by young bilingual children (e.g. de Guerrero, 2005; Tabors, 1997).

Age of Hebrew (L2) onset in acquisition of definiteness

Our bilingual children came from similar home language input backgrounds in terms of Russian (L1) clear dominance in parent–child communication from birth, parents’ socio-cultural background and length of residence in the host country. However, the children differed in AOA. To understand the role of age of Hebrew (L2) onset in acquisition of definiteness in Hebrew, we examined early bilingual children with different AOAs (from six to 42 months) and investigated whether these children showed qualitative and quantitative differences in acquisition of definiteness. In addition, based on Meisel’s (2009) and Armon-Lotem’s (2005) studies, we also examined whether AOA from age three (36 months) and over (in our study, up to 42 months) is a sensitive period in relation to acquisition of definiteness by ESBs.

In general, our data brought some evidence for the role of age of L2 onset, as was proposed by Meisel’s (2009) hypothesis of an early sensitive period in L2 grammar acquisition. We found that the children with earlier AOAs showed a more positive discrepancy score, that is, more correct production of definite articles in the child’s speech with time, as opposed to the children with later AOAs. Thus, it seems that the children with earlier ages of L2 onset acquire definiteness as a complex and multidimensional grammar domain at a more accelerated rate than the children with later ages of L2 onset. This delay among bilinguals with later AOAs could be attributed to differences in the amount of input in L2 during their life span with a clear advantage of input to earlier AOA bilingual groups. In this context, it is important to note that even monolingual Hebrew-speaking children take around six years to reach full competence in production of definite articles. Thus, it could be that the gap between children with earlier and later AOA might eventually narrow with gaining a critical mass of definite input during the life span (Gathercole, 2006).

In addition, our findings about qualitative and quantitative similarities between the two groups of bilinguals in types of errors (omissions and overuse) and in frequency of production of definite articles limit the role of age of onset of L2 only to the rate of acquisition of definiteness. Consequently, our data cannot be interpreted as unambiguous evidence for a sensitive period in L2 grammar acquisition with regard to definiteness in Hebrew (L2).

It was interesting and unexpected that the children with later AOAs (after age three) even showed some kind of advantage in acquisition of definiteness by showing higher frequency of production of definite articles in semantic–pragmatic contexts compared to the children with earlier ages of L2 onset. The average gap between the semantic–pragmatic uses of definiteness and the morpho-syntactic uses of definiteness increased as AOA increased. This pattern of data could be explained by two complementary explanations. First, these findings point out higher pragmatic competence among children with AOA after age three. This might be attributed to their relative maturity in interpersonal communication contexts due to older chronological age, ranging from 42 to 52 months at the beginning of the study versus 40 to 42 months among the children with AOA under age three. Second, the specific rule of the morpho-syntactic context (i.e. using the definite noun after the direct object marker) is more dependent on the amount of exposure and input to this specific unique compound form (‵et ha-) in comparison to the semantic–pragmatic context.

Conclusions

In the present study, we found new evidence supporting both fluctuation and transfer hypotheses with regard to bilingual children without definite articles in their L1. In addition, we found similarities and differences between bilinguals and monolinguals in acquisition of definiteness. Finally, we presented some effects of age of onset of L2 on acquisition of definiteness in L2 that were limited only to rate of acquisition. Future follow-up research is necessary to examine whether bilingual children with later ages of L2 onset could eventually close the gap in acquisition of definiteness over time by reaching a critical mass of linguistic information provided by L2 input.

It is noteworthy that our conclusions might be limited by the fact that we did not examine age-matched monolinguals as a control group or adult L2 users that also come from an [–article] L1 (in this case, Russian). At the same time, the data on monolinguals’ and adult bilinguals’ acquisition came from wide-scale research projects conducted in Israel with research populations from similar socioeconomic backgrounds.

Finally, because the role of quantitative and qualitative aspects of input (Unsworth & Blom, 2010) was outside the scope of the present study, future research needs to investigate the complex relations between such factors as age of L2 onset, L2 input characteristics and success in acquisition of definiteness.

Footnotes

Funding

The author(s) received no financial support for the research and/or authorship of this article.

Author biographies

Mila Schwartz, PhD, is Senior Lecturer in the Oranim Academic College of Education. Her research focuses on language acquisition by early sequential bilinguals speaking different language as L1 and L2 and on effect of early bilingual education on children’s cognitive, linguistic and social development.

Hadas Rovner, MA, studied and specialized at Learning Disabilities Department (Diagnosis and Therapy- MA), Haifa University. Her BA was in Special Education and Linguistic Departments (Haifa University). This is her first research from her thesis work focusing on early bilinguals and language acquisition.

References

Abbott

(2008). Issues in the semantics and pragmatics of definite descriptions in English. In Gundel

J. K.

Hedberg

(Eds.), Reference: Interdisciplinary perspectives (pp. 61–72). Oxford: Oxford University Press.

Alfi-Shabtay

(2006). Hebrew knowledge of veteran Russian-speaking immigrants in Israel. Unpublished doctoral dissertation. Tel-Aviv University.

Armon-Lotem

(2005). How critical is the critical period: The acquisition of definiteness in L2 Hebrew by children with L1 Russian. ISB5, Barcelona.

Armon-Lotem

Avram

(2005). The autonomous contribution of syntax and pragmatic: The acquisition of the Hebrew definite article. In Di Sciullo

Anne-Marie

Delmonte

(Eds.), Universal Grammar and the External systems (pp. 169–182). Amsterdam: John Benjamins.

Avrutin

Brun

(2001). The expression of specificity in a language without determiners: Evidence from child Russian. In Do

Domínguez

Johansen

(Eds.), Proceedings of the 25th annual Boston University conference on language development (pp. 70–81). Somerville, MA: Cascadilla Press.

Birdsong

(2005). Interpreting age effects in second language acquisition. In Kroll

J. F.

de Groot

A. M. B.

(Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 109–127). New York: Oxford University Press.

Blom

Polišenská

Weerman

(2008). Articles, adjectives and age of onset: The acquisition of Dutch grammatical gender. Second Language Research, 24, 297–331.

Ceytlin

S. N.

(2000). Jazyk i rebenok. Lingvistika detskoj rechi. Moskow: Vlados.

Chesterman

(1991). On definition. A study with special reference to English and Finnish. Cambridge: Cambridge University Press.

10.

de Guerrero

M. C. M.

(2005). Inner speech − L2: Thinking words in a second language. New York: Springer.

11.

De Houwer

(1995). Bilingual language acquisition. In Fletcher

McWhinney

(Eds.), The handbook of child language (pp. 219–250). Oxford: Blackwell.

12.

DeKeyser

(2005). What makes learning second language grammar difficult? A review of issues. Language Learning, 55, 1–25.

13.

Döpke

(1992). One parent one language: An interactional approach. Amsterdam: John Benjamins.

14.

Ekiert

(2004). Acquisition of the English article system by speakers of Polish in ESL and EFL settings. Columbia University Working Papers in TESOL and Applied Linguistics, 4(1), 1–23.

15.

Fruchtman

(1979). Definite categories in Hebrew. Unpublished doctoral dissertation. Tel-Aviv University. [In Hebrew]

16.

Gathercole

V. M.

(2006). Miami and North Wales, so far and yet so near: Morpho-syntactic development in bilingual children. University of Wales Bangor.

17.

Glinert

(1989). The grammar of modern Hebrew. Cambridge University Press.

18.

Hawkins

Al-Eid

Almahboob

Athanasopoulos

Chaengchenkit

…Velasco-Zarate

(2006). Accounting for English article interpretation by L2 speakers. In Foster-Cohen

S. H.

(Ed.), EUROSLA Yearbook, 6, (pp. 7–25). Amsterdam: John Benjamins.

19.

Hulk

A. C. J.

Cornips.

(2006). Between 2L1 and child L2 acquisition: An experimental study of bilingual Dutch. In Lleo

(Ed.), Interfaces in multilingualism: Acquisition and representation (pp. 115–137). Amsterdam: John Benjamins.

20.

Ionin

Wexler

(2004). Article semantics in L2-acquisition: The role of specificity. Language Acquisition, 12, 3–70.

21.

Ionin

Zubizarreta

L. M.

Maldonado

B. S.

(2008). Sources of linguistic knowledge in the second language acquisition of English articles. Lingua, 118, 554–576.

22.

Ionin

Zubizarreta

L. M.

Philippov

(2009). Semantics by child and adult L2-English learners. Bilingualism: Language and Cognition, 12, 337–361.

23.

Ionin

Wexler

(2009). L2-acquisition of English articles by Korean speakers. In Chungmin

Simpson

G. B.

Youngjin

(Eds.), The handbook of East Asian psycholinguistics: Korean (pp. 286–304). Cambridge: Cambridge University Press.

24.

Laaha

Gillis

(Eds.) (2007). Typological perspectives on the acquisition of noun and verb morphology. Antwerp Papers in Linguistics 112. Antwerp: University of Antwerp.

25.

Lardiere

(2004). Knowledge of definiteness despite variable article omission. In Brugos

Micciulla

Smith

C. E.

(Eds.), BUCLD 28 Proceedings (pp. 328–339). Somerville, MA: Cascadilla Press.

26.

Lyons

(1999). Definiteness. Cambridge University Press, Cambridge.

27.

MacWhinney

(2000). The CHILDES project: Tools for analyzing talk. Hillsdale: Erlbaum.

28.

Marinis

(2007). On-line processing of passives in L1 and L2 children. In Belikova

Meroni

Umeda

(Eds.), Proceedings of the 2nd conference on Generative Approaches to Language Acquisition North America (GALANA) (pp. 265–276). Somerville, MA: Cascadilla Proceedings Project.

29.

Maslov

(1984). Ocherki po aspektologii. Leningrad (In Russian).

30.

Meisel

J. M.

(2008). Child second language acquisition or successive first language acquisition? In Haznedar

Gavruseva

(Eds.), Current Trends in Child Second Language Acquisition (pp. 55–80). Amsterdam: Benjamins.

31.

Meisel

J. M.

(2009). Second language acquisition in early childhood. Zeitschrift für Sprachwissenschaft, 28, 5–34.

32.

Montrul

(2008). Incomplete acquisition in bilingualism. Amsterdam, Netherlands: John Benjamins.

33.

Morales

(2011). The role of the L1 in the acquisition of English articles by Spanish-speaking children. In Herschensohn

Tanner

(Eds.), Proceedings of the 11th Generative Approaches to Second Language Acquisition conference (GASLA 2011) (pp. 83–89). Somerville, MA: Cascadilla Proceedings Project.

34.

Paradis

(2008). Are simultaneous and early sequential bilingual acquisition fundamentally different? Paper presented at Models of Interaction in Bilinguals: International Conference of Centre for Research on Bilingualism in Theory and Practice, Oct. Bangor University, Bangor, UK.

35.

Paradis

Genesee

(1996). Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition, 18, 1–25.

36.

Rothweiler

(2008). Verbal inflection, case morphology and the acquisition of sentence structure in early successive bilinguals. Paper presented at the 11^th congress of the international association for the study of child language, July–August, Edinburgh, UK.

37.

Salehuddin

Hua

T. K.

Maros

(2006). Definiteness and indefiniteness: A contrastive analysis of the use of determiners between the Malay language and English. GEMA: Online Journal of Language Studies, 6(1), 21–30.

38.

Sarko

(2008). Morphophonological or syntactic transfer in the acquisition of English articles by L1 speakers of Syrian Arabic? In Slabakova

Rothman

Kempchinsky

Gavruseva

(Eds.), Proceedings of the 9th Generative Approaches to Second Language Acquisition conference (GASLA 2007) (pp. 206–217). Somerville, MA: Cascadilla Proceedings Project.

39.

Sopata

(2008). Finiteness in child second language acquisition. Paper presented at the Colloquium on Language Acquisition and Change: Across the Lifespan and across Generations. University of Hamburg.

40.

Schwartz

B. D.

Sprouse

(1994). Word order and nominative case in nonnative language acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In Hoekstra

Schwartz

B. D.

(Eds.), Language acquisition studies in generative grammar (pp.317–368). Amsterdam: John Benjamins.

41.

Schwartz

B. D.

Sprouse

(1996). L2 cognitive states and the full transfer/full access model. Second Language Research, 12, 40–72.

42.

Schwartz

Nir

Leikin

Levie

Ravid

(2014). Acquisition of Noun Plurals among Early Sequential Russian-Hebrew Speaking Bilinguals: A Longitudinal Multiple Case Study. Heritage Language Journal, 11: 144–178.

43.

Snape

(2005). The use of articles in L2 English by Japanese and Spanish learners. Essex Graduate Student Papers in Language and Linguistics, 7, 1–23.

44.

Snape

(2006). The acquisition of the English determiner phrase by Japanese and Spanish learners of English. Unpublished doctoral dissertation, University of Essex.

45.

Tabors

P. O.

(1997). One child, two languages: A guide for preschool educators of children learning English as a second language. Baltimore: Brookes.

46.

Talmy

(1985). Lexicalization patterns: Semantic structure in lexical forms. In Shopen

(Ed). Language typology and syntactic description, vol. 3: Grammatical categories and the lexicon (pp. 56–149). Cambridge: Cambridge University Press.

47.

Unsworth

(2005). Child L2, adult L2, child L1: Differences and similarities. A study on the acquisition of direct object scrambling in Dutch. Doctoral dissertation, Utrecht University.

48.

Unsworth

Argyri

Cornips

Aafke

Sorace

Tsimpli

(2014). On the role of age of onset and input in early child bilingualism in Greek and Dutch. Applied Psycholinguistics, 35, 765–805.

49.

Unsworth

Argyri

Cornips

Hulk

Sorace

Tsimpli

(2010). Disentangling age and input effects in a crosslinguistic study of the acquisition of grammatical gender in child bilinguals. Paper presented at the 32th Annual conference of Deutsche Gesellschaft für Sprachwissenschaft, Berlin, Germany.

50.

Unsworth

Argyri

Cornips

Hulk

Sorace

Tsimpli

(2011). On the role of age of onset and input in early child bilingualism in Greek and Dutch. In Pirvulescu

. (Eds.), Selected Proceedings of the 4th Conference on Generative Approaches to Language Acquisition North America (GALANA 2010) (pp. 249–265). Somerville, MS: Cascadilla Proceedings Project.

51.

Unsworth

Blom

(2010). Comparing L1 children, L2 children and L2 adults. In Unsworth

Blom

(Eds.), Experimental methods in language acquisition research (pp. 201–222). Amsterdam: John Benjamins.

52.

Unsworth

Hulk

A. C. J.

(2009). Early successive bilingualism: Disentangling the relevant factors. Zeitschrift für Sprachwissenschaft, 28, 69–77.

53.

Weerman

Bisschop

Punt

(2006). L1 and L2 acquisition of Dutch adjectival inflection. ACLC Working Papers, 1. (Accessible though website of Amsterdam Center for Language and Communication.)

54.

Wexler

(1976). On the non-lexical expression of determinedness. (With special reference to Russian and Finnish). Studia Linguistica, 30(1), 34–67.

55.

White

(2003). Second language acquisition and universal grammar. Cambridge, England: Cambridge University Press.

56.

Wintner

(2000). Definiteness in the Hebrew noun phrase. Journal of Linguistics, 36, 319–363.

57.

Yip

Matthews

(2007). The bilingual child: Early development and language contact. Cambridge University Press.

58.

Zdorenko

Paradis

(2007). The role of the first language in child second language acquisition of articles. In Belikova

(Ed.), Proceedings of the 2nd conference on Generative Approaches to Language Acquisition North America (GALANA) (pp. 483–490). Somerville, MA: Cascadilla Proceedings Project.

59.

Zdorenko

Paradis

(2008). The acquisition of articles in child second language English: Fluctuation, transfer or both? Second Language Research, 24(2), 227–250.

60.

Zdorenko

Paradis

(2011). Articles in child L2 English: When L1 and L2 acquisition meet at the interface. First Language, 31(4), 1–25.

61.

Zur

(1983). The acquisition of definiteness among Hebrew speakers. Thesis Work. Tel-Aviv University. [In Hebrew]