Mothers’ Education, Family Language Policy, and Hebrew Plural Formation among Bilingual and Monolingual Children

Abstract

The present study examines the role of maternal years of education and family language policy (FLP) in monolingual and bilingual children’s acquisition of Hebrew plural morphology. The case of the Hebrew plural system is especially interesting when examining the influence of the above factors on morphological performance, given that it demands both a mastery of morphological rules (characterized by a high degree of transparency in Hebrew) and a lexicon-based mastery of exceptions. Participants were 146 children, 74 bilinguals (heritage language: Russian; societal language: Hebrew) and 72 Hebrew monolinguals, aged 5–8 (kindergarten, first grade, and second grade), from the same schools and neighborhoods. A Hebrew pluralization, sentence completion task that included 99 items from two categories: fully regular words whose plural forms are based on a morphological rule and non-regular words whose plural forms (also) require lexical and/or morpho-lexical knowledge. The parents of the bilingual children filled out a questionnaire with questions on background variables (e.g., maternal education) and language practice in both languages by different family members and language use at home. The findings indicated that maternal education contributes differently and distinctly to the linguistic performance of children from different linguistic backgrounds. For monolingual children, an increase in the number of years of maternal education is associated with an increase in the likelihood of success in the lexical and morpho-lexical aspects of Hebrew. By contrast, for bilingual children, no significant contribution of maternal education to children’s performance was found. For bilingual participants, their performance in the lexical and morpho-lexical aspects of the Hebrew plural system was consistently influenced by FLP across all school settings—increased use of Russian at home was associated with a lower likelihood of success in the societal language. FLP characteristics were not found to be related to maternal education. These findings have clinical implications for both assessment and intervention processes when working with bilingual children.

1. Introduction

The amount of input in each language plays a predictive role for both languages in child bilingual development (e.g., Pearson et al. (1997) and Thordardottir (2011) for vocabulary and Gathercole (2007) for morphology). Yet, what contributes to the amount and quality of input beyond age at the onset of bilingualism and length of exposure is less studied. Maternal education, for example, was found to affect the input that monolingual children receive at home (Calvo and Bialystok 2014; Dollaghan et al. 1999), and family language policy (FLP) involves practices that affect the input in bilingual homes (Rose et al. 2023). The present study aimed to address this gap by showing different patterns of the contribution of maternal education among monolinguals and bilinguals alongside a significant contribution from FLP to the performance of bilinguals in the societal language (SL), with no correlation between the two variables among the bilinguals. Looking into the role of maternal education in shaping the language of bilingual children is of great interest, as maternal education has also been used as a key factor in family’s socio-economic status (SES) in monolingual and bilingual populations, directly, as a sole measure (Dollaghan et al. 1999; Meir and Armon-Lotem 2017) or indirectly, within a scale for calculating SES (Bialystok and Shorbagi 2021; Fernald et al. 2013; Levie et al. 2017; Weiler et al. 2021). Parents’ questionnaires were used to capture maternal education (in years) and FLP measured according to the relative proportion of input received in each language in the homes of Russian–Hebrew bilingual children aged 5–8, acquiring Hebrew as the SL. The study explored the role of these measures in the acquisition of the plural morphology in SL-Hebrew, assessed using a controlled experimental design. Hebrew’s plural morphology was chosen as a test case since its mastery requires both the acquisition of a gender-based rule-governed system that is acquired early and is thus more likely to be resilient to input and multiple lexical exceptions that are likely to be affected by the amount of input. A deeper understanding of the relationship between maternal education, FLP, and the acquisition of this particular system will enable a better grasp of the precise mechanism through which maternal education and FLP influence children’s language.

1.1. Maternal Education in Language Acquisition

Higher maternal education is consistently associated with higher lexical abilities in children (Dollaghan et al. 1999; Hoff 2003; Hoff and Tian 2005; Huttenlocher et al. 2010; Magnuson et al. 2009; Rowe 2008; Zambrana et al. 2012). Dollaghan et al. (1999), for example, analyzed the spontaneous language samples of 240 monolingual three-year-old English-speaking children, divided into three maternal education groups (did not complete high school, high school graduates, and college graduates). They reported that children of more educated mothers had advantages both in expressive vocabulary (measured using the number of different words) and in receptive vocabulary. Hoff (2003) examined the productive vocabulary of monolingual 2-year-old children at two time points, 10 weeks apart. She found that higher maternal education (fathers’ and mothers’ education levels were similar) was associated with a greater increase in the child’s productive vocabulary. Dollaghan et al. (1999) also showed an advantage in the morphosyntactic domain for the group of children whose mothers were college graduates. This advantage was reflected in a higher number of morphemes per utterance. Studies consistently link higher education with more varied, diverse, and richer linguistic input (Dollaghan et al. 1999; Hoff 2003; Hoff and Tian 2005; Huttenlocher et al. 2010; Magnuson et al. 2009; Rowe 2008; Zambrana et al. 2012).

According to usage-based models (Diessel 2017; Tomasello 2001), the exposure to linguistic input that invites varied and frequent language use explains the advantage of children with more educated mothers in various language domains. This exposure occurs through both spoken and written channels, for example, through book reading and the application of diverse parental strategies during reading (Silvén et al. 2003).

Understanding the linguistic functioning of bilingual children requires consideration of factors relevant to language development in monolinguals, such as maternal education, as well as factors unique to bilinguals: bilingualism itself, FLP, age at the onset of bilingualism, and others (Bosch et al. 2019; Michaly and Prior 2024; Reznick and Armon-Lotem 2022; Shahar-Yames et al. 2018).

The influence of maternal education on linguistic performance has also been observed in bilingual children (De Cat 2021; Meir and Armon-Lotem 2017), although it has been less extensively researched in this population compared to monolinguals. When the focus was specifically on the distinct impacts of maternal education and bilingualism on the vocabulary of bilingual children, the two variables were found to have independent effects in their SL: a more educated background is associated with a larger vocabulary—for both monolinguals and bilinguals—and bilingualism is associated with a smaller vocabulary in both less and more educated backgrounds. This effect is valid for both receptive vocabulary (Calvo and Bialystok 2014) and expressive vocabulary (Meir and Armon-Lotem 2017) in studies that focused on bilingual participants with a wide variety of home languages (Calvo and Bialystok 2014) and in studies that focused on bilingual children with the same language pair (Meir and Armon-Lotem 2017). Likewise, Schwartz et al. (2009b), found a relationship between semantic-lexical abilities and the parents’ level of education, education/studies in Hebrew, and the length of stay in the country. According to those researchers, the contribution of education to higher achievements in the SL stems from a greater motivation to use SL-Hebrew, as reflected in the positive correlation between the number of years of education and the number of Hebrew books at home.

Moving beyond the lexicon, Meir and Armon-Lotem (2017) showed the influence of maternal education and bilingualism to also be valid for the syntactic domain, assessed by a sentence repetition task. The study participants (120 children, both Russian–Hebrew bilinguals and Russian/Hebrew monolinguals, aged 5;7-6;7) were divided into two maternal education groups (12 years of education or less and above 12 years of education). The researchers showed that, similar to the expressive vocabulary of the participants, their ability to repeat various syntactic structures was independently influenced by maternal education and bilingualism.

To the best of our knowledge, no study has been conducted that focused on the relationship between maternal education and the morphological abilities of Hebrew-speaking children, monolingual or bilingual, which is central to the current research. Two studies focused on the relationship between SES and morphological abilities in Hebrew monolinguals, where differences in SES were associated with differences in their education levels. Schiff and Ravid (2012) examined the ability of 360 monolingual Hebrew-speaking children in grades 1–6, from low SES and medium-high SES, to form plural nouns. SES was defined by residential area and the Ministry of Education SES Index. Low SES was characterized by an average of 10 years of parental education and medium-high SES by 17 years. The researchers reported better performance among children from more educated backgrounds in both creating plural nouns and correctly marking the agreement structure. This advantage was particularly pronounced in structures that included nouns with a changing stem and/or irregular suffixes—structures that require lexical and morpho-lexical learning and more demanding morpho-phonological and morpho-syntactic processing. Likewise, Levie et al. (2017) found a clear effect of SES on the morpho-lexical abilities of 659 monolingual Hebrew-speaking children and adolescents aged 6–14 years, tested on deriving novel words out of words presented to them.

1.2. Maternal Education and FLP

The influence of maternal education on the linguistic abilities of children has been attributed to the characteristics of child-directed speech (CDS) that differ between different maternal education groups. Less educated mothers provided a more impoverished linguistic input, both quantitatively and qualitatively, presenting less lexical diversity and less syntactic complexity and resulting in less complexity in their children’s language (e.g., Hoff 2003; Hoff and Tian 2005; Huttenlocher et al. 2010; Magnuson et al. 2009). Rowe (2008), for example, shows that more educated parents possess greater knowledge about child development, which serves as a mediating factor between education background and the quantity and quality CDS. Moreover, Magnuson et al. (2009) demonstrate that advancement in maternal education led to enhancements in the quality of the home and family environment, with more stimulating and challenging interactions and settings, which in turn contribute to better lexical development in children.

In bilingual families, FLP has a direct influence on the quantity and quality of linguistic input in each language. FLP has three main components: ideology, management, and practice (Spolsky 2012). The ideological component includes attitudes and beliefs about bilingualism: for example, whether a parent believes that bilingualism has cognitive and/or social advantages. The management component refers to the linguistic approach and the rules surrounding language use at home, such as “One-Parent-One-Language” (OPOL; Barron-Hauwaert 2004), both parents using only the home language, what to do when a child speaks to a parent in the SL, etc. The practice component relates to how language use is actually implemented in the family context, that is, the practical application of the management component. The relationships between the three components are complex. Thus, ideology dictates the management pattern and linguistic practice, but at the same time, it can also be influenced by them (Altman et al. 2014). The question is whether there is a relationship between maternal education and language policy in the bilingual family.

Studies of populations with maternal education of over 12 years suggest the possibility that higher maternal education is associated with a desire to preserve the HL (Kennedy and Romo 2013; Rose et al. 2023; Schwartz et al. 2009b) alongside variations in language management and practice methods. Rose et al. (2023) described the FLP in the families of bilingual children who speak English as a HL and Hebrew as a SL, with maternal education of 13–24 years (mean: 17), showing that in most families, language management consisted of using English at home and Hebrew outside the home, but no reference was made to the impact of maternal education. For practice, Gonzalez-Barrero et al. (2021), for example, showed that bilingual families from a highly educated background (73% of the participating parents held a university degree or higher) provided higher literacy support in their dominant language—these families had more books in this language and engaged in reading more frequently, for longer durations, and started at an earlier age compared to activities in their non-dominant language. Sorenson Duncan and Paradis (2020) pointed out that, when mothers acquired their education in an HL, a higher level of education was associated with the higher use of the HL with the child (and less in the SL—English); meanwhile, for mothers who acquired their education in an SL, a higher level of education was associated with higher linguistic input in the SL. Finally, in the study of a population that was similar to the one in the present study, Schwartz et al. (2009b) focused on bilingual (Russian-HL and Hebrew-SL) participants among whom half of the parents held an academic degree. The researchers reported that parents believed in the importance of preserving Russian alongside the use of both Hebrew and Russian in communication with their children, but again, it did not refer to the different education groups. Some of the children reported the frequent use of Russian with their parents, while others reported using only Hebrew with their parents, about half with their siblings, and the majority (70%) with their Russian-speaking friends. In this context, it is important to mention the high correlation observed between language preference (SL, HL or both languages) among mothers and their children (Romero et al. 2004). However, even within the same bilingual family, there can be conflicting approaches to language policy and practice (Reichmuth 2024). The question arises regarding the impact of FLP on the language acquisition of a bilingual child.

1.3. Family Language Policy and Bilingual Language Acquisition

FLP is expected to have an impact on the linguistic functioning of bilingual children, as it directly influences their linguistic experiences in the different languages within the home environment (Altman et al. 2014; Armon-Lotem and Meir 2019; Dagenais and Day 1999; Hammer et al. 2009; Kennedy and Romo 2013; Rose et al. 2023). Altman et al. (2014), for example, found a relationship between FLP and the use of the two languages: a stronger pro-Russian ideology was associated with more extensive use of this language at home (practice). Although the effect of FLP is not identical across different language domains, all three components of FLP have a direct impact on the vocabulary of bilingual children—both in the HL and in the SL, as it is particularly sensitive to the quantitative and qualitative characteristics of linguistic input (Altman et al. 2014; Magnuson et al. 2009). Hwang et al. (2022) examined the relationship between parental beliefs about language acquisition in bilingualism and conceptual vocabulary among 190 Spanish–English bilingual children from three grade levels (kindergarten, second grade, and fourth grade). They found that the parental belief that children could successfully learn both languages simultaneously was more characteristic of families using Spanish at home. Additionally, for participants with limited English proficiency, it was found that a greater use of English at home was associated with higher scores in conceptual vocabulary. In other words, for these children, an indirect effect of parental beliefs on vocabulary was found via the choice of language at home. This was supported by Rose et al. (2023), who examined the impact of FLP on the vocabulary of 82 bilingual English–Hebrew-speaking children aged 5–14 and reported a direct effect of language policy on vocabulary in both languages. For Russian–Hebrew bilingual children, Altman et al. (2014) reported that a pro-bilingual approach, which is expressed through encouraging and promoting a bilingual environment at home, led to lower achievements in the Russian lexicon. At the same time, emphasizing the preservation of Russian while using Hebrew minimally or primarily for functional purposes at home (as done in pro-Russian families) did not lead to lower achievements in the Hebrew lexicon compared to the achievements of children from pro-bilingual homes. Hammer et al. (2009) presented similar findings for Spanish as an HL. For other linguistic domains, Altman et al. (2014) tested the impact of language policy on children’s performance in the morpho-phonology (via a non-word repetition task) and morpho-syntax (via a sentence repetition task) of both languages. The researchers reported a positive correlation between the extent of the use of the SL at home and the success rate in the task assessing morpho-phonological knowledge in that language. The differences in language policy did not lead to differences in the syntactic task, which is consistent with the existing literature, according to which different exposure measures predict abilities in the phonological and lexical domains but not in the syntactic domain (see Armon-Lotem and Meir 2019 for review and discussion). Sorenson Duncan and Paradis (2020) assessed the performance of 89 five-year-old children with English as their SL and a variety of HLs, also finding that greater input in the SL led to an advantage in the lexical domain but not in the syntactic domain. By contrast, Schwartz et al. (2009b), who focused on the semantic–lexical abilities of 70 7-year-old Russian–Hebrew bilingual children, found no significant relationship between the children’s linguistic abilities in the SL (Hebrew) and the parental linguistic input. The researchers explained this finding by noting that Russian still remained dominant in these families, and the use of Hebrew was limited to narrow and random functional contexts. This latter finding calls for further exploration of the impact of FLP on bilingual language acquisition among this study population.

1.4. Plural Morphology in Hebrew and Its Acquisition

The case of the Hebrew plural system is especially interesting when examining the influence of the above factors on morphological performance, given that it demands both a mastery of morphological rules (characterized by a high degree of transparency in Hebrew), and a lexicon-based mastery of the exceptions. The plural suffix in Hebrew includes information on grammatical gender and number. The suffix can be regular (-im for masculine and -ot for feminine) or irregular (the plural feminine suffix -ot for a masculine noun and the plural masculine suffix -im for a feminine noun). Feminine nouns with irregular plural forms are much less common compared to irregular masculine forms (Ornan 2003; Ravid et al. 2008; Ravid and Schiff 2012). When forming the plural, the stem of the word may remain unchanged or undergo a change (Ravid 1995).

Three main factors shape the process of acquiring plurals in Hebrew: stem type (lower performance in words with changing stems), suffix type (lower performance in words with irregular suffixes), and grammatical gender (lower performance in feminine nouns). The acquisition of the plural system in its regular aspects among monolingual children is rapid and early, alongside a longer and slower acquisition process for irregular forms (Berman 1981; Ravid 1995; Reznick and Armon-Lotem 2022; Ravid and Schiff 2009, 2012; Schiff et al. 2011; Schiff and Ravid 2012). Studies on the acquisition of the plural system at a young age in Russian–Hebrew bilinguals (Schwartz et al. 2009a, 2014; Reznick and Armon-Lotem 2022) found that the same central factors shape acquisition among both monolinguals and bilinguals, with differences mainly in the irregular aspects of the system. This pattern was attributed to reduced exposure to the SL (Hebrew) among bilingual children, well documented in research on morphology acquisition among bilingual children from various linguistic backgrounds (e.g., DeKeyser 2005; Goldschneider and DeKeyser 2001; Jensvoll 2004; Paradis et al. 2007; Rispens and De Bree 2015). These findings are consistent with the claims of usage-based models (Diessel 2017; Tomasello 2001), which emphasize the importance of exposure and language use, as well as the frequency effect, in the acquisition of inflectional morphology, both in its regular and irregular aspects (Ambridge et al. 2015; Tatsumi et al. 2018).

1.5. Rationale and Research Questions

Despite the extensive research on the impact of maternal education in monolinguals, little is known about the influence of this factor on morphological development in bilinguals. Bilingual language development was further found to be influenced by FLP (among other factors), as it impacts the use of different languages in different contexts, yet the relation between maternal education and FLP is less studied. Understanding the precise mechanism through which maternal education influences the language of bilingual children is needed. The current study aims to fill this gap, focusing on the number of years of maternal education, FLP practice, and the morphological performance of speakers in SL-Hebrew of Russian–Hebrew bilingual children.

The following research questions will be addressed:

• What is the contribution of maternal education and FLP, beyond factors of linguistic status (monolingual or bilingual), grade level, and the characteristics of Hebrew words, to the performance of Hebrew-speaking children (monolingual and bilingual) in the Hebrew plural system?
• Is there a relationship between maternal education and FLP?

In light of the abovementioned studies, we hypothesized that both maternal education and FLP would have an impact on the knowledge of plural morphology in SL-Hebrew and that maternal education would have an impact on FLP in the families of bilingual children. Maternal education might impact the acquisition of plural morphology in SL-Hebrew beyond the impact of FLP, as it was shown that maternal education impacts the amount and quality of child-directed speech, that is, the input children get. Yet, it could also be the case that no additional impact will be observed for maternal education if the Hebrew used at home is not used by the mothers or if the amount and quality of Hebrew used by the mothers is not enough to generate such an effect.

2. Method

2.1. Participants

One hundred and forty-six children, 74 bilinguals (Russian–Hebrew) and 72 monolinguals (Hebrew), aged 5–8 (kindergarten, first grade, and second grade), from the same schools and neighborhoods, participated in the study, within a more comprehensive study of regular and irregular morphology in Russian–Hebrew bilingualism (Reznick 2019). All participants but five (3.4%) were born in Israel. Parents reported no language-related challenges (e.g., hearing difficulties, repeated ear infections in the prior year, or parental concerns over their child’s linguistic development).

Table 1. Background information on the participants.

			ML (N = 72)			BL (N = 74)
			Kindergarten	1st Grade	2nd Grade	Kindergarten	1st Grade	2nd Grade
			(N = 26)	(N = 25)	(N = 21)	(N = 30)	(N = 23)	(N = 21)
Age (months)	Mean (SD)		68.88 (5.19)	83.08 (5.28)	94.57 (4.19)	70.60 (5.04)	84.57 (3.31)	94.86 (3.84)
Boys	N (%)		11 (42.3)	11 (44)	10 (47.6)	14 (46.7)	6 (26.1)	8 (38.1)
Mother Ed. (years)	Mean (SD)		14.88 (2.55)	13.80 (2.18)	15.24 (3.19)	14.07 (2.13)	14.35 (2.79)	14.71 (2.69)
Birth Order	N (%)	1st born:	9 (34.6)	13 (52)	8 (38.1)	13 (43.3)	9 (39.1)	8 (38.1)
		2nd born:	9 (34.6)	6 (24)	9 (42.9)	11 (36.7)	11 (47.8)	7 (33.3)
		3rd born:	6 (23.1)	5 (20)	4 (19)	6 (20)	3 (13)	4 (19)
		4th born:	2 (7.7)	1 (4)	0 (0)	0 (0)	0 (0)	2 (9.5)
Age of Onset of BL (months)	Mean (SD)					26.37 (17.88)	34.87 (20.58)	29.95 (23.59)
Born in Israel	N (%)					29 (96.7)	21 (91.3)	19 (90.5)

presents participant characteristics by linguistic status (monolingual/bilingual) and grade level (kindergarten/first grade/second grade). A significant age difference was observed for the different grade levels (p < 0.001), with no significant differences between monolingual and bilingual children attending the same grade level (p ≥ 0.86). The groups did not differ in their years of maternal education (F(5, 140) = 1.04, p = 0.40), with a significant positive correlation between the educational levels of mothers and fathers (monolingual: r = 0.63, p < 0.001; bilingual: r = 0.50, p < 0.001). Most groups exhibited no gender differences (χ² ≤ 2.35, p ≥ 0.13), and there were no significant group differences in the number of children with older siblings (χ² ≤ 2.28, p ≥ 0.13). The bilingual groups did not significantly differ in their age at the onset of bilingualism (i.e., initial exposure to Hebrew) (F(2, 71) = 1.13, p = 0.33).

2.2. Tools

2.2.1. The Parents’ Questionnaire

For each participant in the study, a parental questionnaire was filled out, which included diverse demographic information, including information on the parents’ education (see the questionnaire in the Supplementary Materials). In addition, the parents of the bilingual children answered FLP practice questions on the family members’ use of the HL and the SL (Russian, both Russian and Hebrew, or Hebrew) with the child. This focused on six interactive conditions, including the language spoken at home and the language spoken with each family member: the mother, the father, siblings, grandparents, and friends. For each of these six categories (language used at home and five interlocutors), parents were asked to indicate one of three options: the use of Russian, the use of both languages, or the use of Hebrew. Consequently, for each bilingual participant, the score for language choice (both at home and with the five interlocutors) in each of the FLP practices (Russian, both languages, and Hebrew) ranged from zero to six. The questionnaire was presented in Hebrew to the parents of the monolingual children and in Russian to the parents of the Russian–Hebrew-speaking children.

2.2.2. The Hebrew Pluralization Task

The pluralization task evaluated participants’ ability to produce the Hebrew plural form of 99 given singular nouns in a sentence completion task. The participants were told, “Here there is a {noun}, and here there are many…”. The pluralization task also included verb inflections to serve as distractors. Since both in Hebrew and in Russian, the word suffix in the singular form typically indicates the grammatical gender—from which the plural suffix is derived—each of the 99 words had a regular singular form in both languages. This was done to neutralize the potential impact of irregular singular forms on the children’s performance. The nouns varied in stem type (non-changing or changing), suffix type (regular or irregular), and grammatical gender in Hebrew (masculine or feminine), and the variable of congruence/incongruence in the grammatical gender of the word in both languages was controlled for. Thus, out of the 99 test words, 41 had the same grammatical gender in both languages, and 58 words had different grammatical genders. For 57 masculine words in Hebrew, 24 were also masculine in Russian, 21 were feminine in Russian, and 12 were neutral in Russian. For 42 feminine words in Hebrew, 17 were also feminine in Russian, 15 were masculine, and 10 were neutral.

Table 2. Distribution of the 99 task nouns by stem and suffix type in the plural.

		Suffix
		Regular	Irregular
Stem	Non-Changing	34	20
	Changing	31	14

presents the distribution of the 99 test items by stem and suffix types.

For the purposes of the current study, these four word types were grouped into two main categories: fully regular words, whose plural forms are based on a morphological rule (words with a non-changing stem and a regular suffix), and irregular words, whose plural forms require (also) lexical and/or morpho-lexical knowledge (words with a changing stem and/or an irregular plural suffix). For information on the role of stem and suffix characteristics in the performance of study participants, see Reznick and Armon-Lotem (2022).

2.3. Administration of the Task

Each test item was presented with visual stimuli in a randomized direction (either right to left or left to right), followed by an oral stimulus presented through headphones two seconds later. The visual stimuli included a pair of pictures: one depicting the item in singular and the other in plural. The oral stimuli included a sentence for completion. Before the test began, four trial sentences were administered. The participants’ responses were recorded both by a computer and by the experimenter in real time. The study commenced in a quiet setting either in the child’s educational environment or at home. Parental consent and the children’s assent were secured, and IRB approval was obtained from both the University Humanities Ethics Committee and the Ministry of Education (No. 8883.646, issued 27 December 2015).

2.4. Data Coding

Family language policy: One of the ways to define the practice component in FLP is through the languages actually spoken at home in different contexts (e.g., Rose et al. 2023; Schwartz et al. 2009b). This measure was used in the current study. For each participant, the number of interactive conditions in which each language (Russian and Hebrew) or both languages were used was summed up. The number of interactive conditions for each language ranged from zero to six (see the Section 2.2, parent questionnaire).

Noun Pluralization: A valid response was defined as the first utterance containing a complete word, if produced.

Table 3. The categories for error analysis.

Category		Examples
Wrong plural form
Stem error	No change of the singular form	kelev “dog”—kelev-im (instead of klav-im)
	Incorrect or unrequired change	kelev “dog”—kalev-im (instead of klav-im)
Suffix error	Masculine instead of feminine	sulam “ladder”—sulam-im (instead of sulam-ot)
	Feminine instead of masculine	nemala “ant”—nemal-ot (instead of nemal-im)
Lexical substitution
	An existing noun altering a target noun’s meaning	kise “chair”—kis-im “pockets” (instead of kisa-ot)
	A neologism created via the incorrect inflection of a lexical substitution or a minor distortion in the lexical substitutions	sefel “mug”—kos-im (instead of sfal-im)
Productions of the singular		kelev “dog”—kelev (instead of klav-im)
Other: No response, “I don’t know” responses, responses in Russian, and irrelevant speech

shows categories for error analysis (for a more detailed presentation of the categories for error analysis, see Reznick and Armon-Lotem 2022).

2.5. Analysis

To examine the contributions of maternal education and FLP, beyond factors of linguistic status, grade level, and the characteristics of Hebrew words, to the performance of the research participants, the data were analyzed using generalized linear mixed models (GLMMs). These models assessed the contributions of various independent variables to a binary dependent variable—success/failure in the Hebrew task items that capture the errors made by the children. To investigate the correlation between the bilingual child’s maternal education and FLP, partial Pearson correlations controlling for chronological age and length of exposure to Hebrew were conducted. The analyses were performed using IBM SPSS Statistics 29.

3. Results

Descriptive statistics presenting the distribution of success according to the main variables: linguistic status, grade level, and word characteristics in Hebrew (fully regular and irregular) appear in Table S1 (Supplementary Materials). In the first stage, GLMM analysis was conducted for all participants and all task items to evaluate the contributions of key research variables to participant performance. The GLMMs were constructed with a binary dependent variable, “success”, for each task item (success/failure). The following continuous predictors were included in the model: the number of years of the mother’s education, the number of interactive conditions in which Russian was used (0–6), and the grade level (kindergarten, first grade, and second grade, with approximately one year apart between them). Additionally, the model included categorical predictors of linguistic background (two levels: monolingual/bilingual) and the regularity of Hebrew words (two levels: fully regular item/irregular item).

The analysis results are presented in

Table 4. GLMM analysis: effects of independent variables on the likelihood of success in the Hebrew task.

Effect	B	Std Error	Lower	Upper	Wald χ2	df	p
(Intercept)	−0.861	0.1514	−1.158	−0.564	32.316	1	<0.001
Mother Education	0.023	0.0090	0.005	0.041	6.514	1	0.011
FLP—Russian Use	−0.152	0.0188	−0.189	−0.115	64.900	1	<0.001
Grade Level	0.486	0.0286	0.430	0.542	289.819	1	<0.001
Language Background	0.478	0.0556	0.369	0.587	73.876	1	<0.001
Item regularity	2.217	0.0735	2.073	2.361	910.160	1	<0.001

. For continuous variables, a positive slope (B) indicates that an increase in the variable’s value increases the likelihood of success in the plural task items. A negative slope indicates that an increase in the variable’s value decreases the likelihood of success. For categorical variables, the following were defined as the reference levels in each category: linguistic background = bilingual, regularity of Hebrew words = irregular. A positive slope in these cases means that the first level of each variable compared to the second level results in a positive likelihood of success. Thus, the positive slope indicates a higher likelihood of success for this reference level compared to other levels in the category.

The analyses revealed several significant effects (Table 4). First, a significant effect of FLP was found—an increase in the number of interactive conditions where Russian was used correlated with a lower likelihood of success in the Hebrew plural system (B = −0.152, SE = 0.0188, χ² = 64.900, p < 0.001). Second, a significant effect of grade level was observed—an increase in grade level correlated with an increase in the likelihood of success (B = 0.486, SE = 0.0288, χ² = 289.819, p < 0.001). Third, a significant effect of linguistic background was identified—bilinguals had a lower likelihood of success compared to monolinguals (B = 0.478, SE = 0.0556, χ² = 73.876, p < 0.001). Finally, a significant effect was found for word type, with a higher likelihood of success for fully regular items compared to items with a changing stem and/or an irregular suffix (B = 2.217, SE = 0.0735, χ² = 910.160, p < 0.001). The number of years of the mother’s education, despite being significant, showed a slope close to zero, indicating its weak effect (B = 0.023, SE = 0.0090, χ² = 6.514, p = 0.011).

In the second stage, the contributions of maternal education and FLP were examined using four models: one for each linguistic background (monolingual/bilingual) and word type (fully regular/irregular). In each model, the grade level variable effect and its interactions with the primary variables of interest—maternal years of education and language policy (for bilinguals only) were included. To avoid collinearity, maternal education and language policy were included not as main effects but only as interaction variables. As a result, for the significant interactions, we proceeded with simple slope-effects analyses, while for the non-significant interactions, we examined the main effects of maternal education and language policy.

Significant interactions were found within each linguistic background for words with a changing stem and/or irregular suffix: for monolinguals, a significant interaction was observed between the grade and maternal education (B = 0.021, SE = 0.0050, χ² = 17.618, p < 0.001). For bilinguals, a significant interaction was found between the grade and language policy (B = −0.054, SE = 0.0068, χ² = 63.431, p < 0.001). Additionally, for bilinguals, a marginally significant interaction was identified between the grade and language policy for fully regular words (B = −0.039, SE = 0.0201, χ² = 3.773, p = 0.052) (Table S2, Supplementary Materials).

For significant and marginal interactions, simple slope effects were conducted for maternal education and language policy (Table S3, Supplementary Materials). For the interactions that were not found to be significant, we examined the main effects of maternal education and language policy (Table S4, Supplementary Materials).

These analyses revealed the differential behavior of the maternal education factor within each linguistic background. Among monolingual participants in second grade, an increase in the number of years of the mother’s education was associated with an increase in the likelihood of success in word types requiring lexical and morpho-lexical learning (words with a changing stem and/or an irregular suffix, B = 0.142, SE = 0.0297, χ² = 22.942, p < 0.001). This factor did not significantly impact children’s performance in these words in lower grades (kindergarten and first grade, |B| ≤ 0.026, SE ≥ 0.0222, χ² ≤ 1.376, p ≥ 0.241). Additionally, maternal education did not contribute to the performance of monolinguals with fully regular words (B = 0.081, SE = 0.0525, χ² = 2.379, p = 0.123). Among bilinguals, no contribution of maternal education to participant performance was found in any word type (|B| ≤ 0.024, SE ≥ 0.0127, χ² ≤ 0.444, p ≥ 0.505).

In addition to the lack of contribution from maternal education to the performance of bilingual children, it was found that an increase in the number of interactive conditions where Russian was used was associated with a decrease in the likelihood of success in Hebrew. For words with a changing stem and/or irregular suffix, this effect of language policy was observed across all three grade levels (kindergarten, first grade, and second grade), with a stronger slope observed in the first grade (B ≤ −0.134, SE ≤ 0.0344, χ² ≥ 15.514, p < 0.001). For fully regular words, this contribution of language policy was found only in first grade (B = −0.209, SE = 0.1020, χ² = 4.199, p = 0.040).

Table 5. Bilinguals: effect of FLP on the likelihood of success in the Hebrew task.

	Regular	Irregular
Kindergarten	x	√
1st Grade	√	√
2nd Grade	x	√

√ = Significant decrease in the likelihood of success with an increase in the use of Russian at home. x = The absence of an effect of FLP on the success rate.

presents a summary of the contribution of language policy among bilingual participants.

In order to examine the possibility that FLP mediates the relationship between maternal education and participant performance, which might explain why maternal education was not significant for bilinguals in the above models, additional GLMMs were constructed to test the total relationship between maternal education and the performance of bilingual children in each word type. It was found that, even in these models, the variable of maternal education did not significantly contribute to the performance of bilingual participants (|B| ≤ 0.038, SE ≥ 0.0122, χ² ≤ 1.290, p ≥ 0.256). In contrast, a similar examination of the total relationship between language policy and children’s performance revealed a significant contribution of language policy to performance on words with a changing stem and/or irregular suffix (B = −0.147, SE = 0.0198, χ² = 55.151, p < 0.001), alongside a lack of contribution for fully regular words (B = −0.048, SE = 0.0540, χ² = 0.797, p = 0.372). Therefore, the hypothesis that maternal education is mediated by language policy was rejected.

To examine the correlation between the number of years of maternal education and the number of interactive conditions in which Russian was used at home, a partial correlation was conducted for bilingual participants, controlling for chronological age and length of exposure to Hebrew. The correlation was not significant (r = 0.05, p = 0.70).

As noted, the grammatical gender factor was excluded from the above analyses due to the small number of items in each cell. Despite this limitation, an examination of the three characteristics in Hebrew (stem, suffix, and grammatical gender), alongside other variables (grade level, linguistic status, maternal education, and language policy), suggests a potential connection between maternal education and the likelihood of success in a very specific category of words among second-grade bilingual children. These words have the least frequent plural forms in Hebrew and require lexical learning (feminine words with a non-changing stem and an irregular plural suffix, seven items in the category). With an increase in the number of years of the mother’s education, the likelihood of success in these words decreased (B = −0.189, SE = 0.0696, χ² = 7.40, p = 0.007). This contribution of maternal education to the performance of bilingual children using these words was also found when examining the total relationship (without language policy) between maternal education and children’s performance (B = −0.190, SE = 0.0685, χ² = 7.703, p = 0.006). This direction warrants further research under more controlled conditions, with a sufficient number of items in each cell.

4. Discussion

The primary objective of this study was to characterize the roles of maternal education and FLP in the acquisition of the SL by bilingual children. The research goal was achieved through a systematic examination of the performance of Russian–Hebrew-speaking children aged 5–8 in the Hebrew plural system. Maternal education was used as a continuous predictor, measured by the number of years of education as reported in the parent questionnaires. The language policy was assessed based on the practice component, measured through the number of family interactions where HL-Russian, SL-Hebrew, or both languages were used.

The current study had several key findings. First, it was found that maternal education contributes differently and distinctly to the linguistic performance of monolingual and bilingual children. For monolingual children, an increase in the number of years of maternal education was associated with an increase in the likelihood of success in the lexical and morpho-lexical aspects of Hebrew. By contrast, for bilingual children, no significant contribution of maternal education to children’s performance was found. Second, for bilingual participants, their performance in the lexical and morpho-lexical aspects of the Hebrew plural system was consistently influenced by FLP across all grade levels—increased use of Russian at home was associated with a lower likelihood of success in the societal language. Finally, in our sample, FLP characteristics were not found to be related to maternal education.

4.1. The Role of Maternal Education among Bilinguals and Monolinguals

The lexical and morpho-lexical performance of monolingual participants in Hebrew was aligned with their maternal education—higher maternal education was found to be associated with better performance by the children. This finding, observed for the first time among Hebrew-speaking elementary school students, aligns with consistent reports on the advantage of monolingual children from more educated backgrounds in the lexical domain (Dollaghan et al. 1999; Hoff 2003; Hoff and Tian 2005; Huttenlocher et al. 2010; Magnuson et al. 2009; Rowe 2008; Zambrana et al. 2012; but see Meir and Armon-Lotem 2017 for findings on the relationship between maternal education and the lexical performance of monolingual Hebrew-speaking kindergarten children).

The novelty of this study lies in its clear and systematic demonstration of the selective nature of the contribution of maternal education to monolingual children’s performance. There was no impact on the mastery of the morphological rule itself (words with non-changing stems and regular suffixes), but rather, an impact on the acquisition of the lexical and morpho-lexical aspects of the system (words with changing stems and/or irregular morphology) was observed, and this impact was limited to the second grade. The nature of the differences in the impact of maternal education—affecting the lexical and morpho-lexical aspects of the system but not the morphological rule itself—aligns with what is known about the acquisition of plurals in Hebrew among monolinguals (Berman 1981; Ravid 1995; Ravid and Schiff 2009, 2012; Schiff et al. 2011; Schiff and Ravid 2012). The Hebrew plural system, in its regular aspects, is transparent, productive, highly regular (Berman 1981), and most frequent (Ravid et al. 2008). The transparency and productivity of the morphological rule (Berman 1981; Ravid 1995) explains the better mastery of these forms (Ravid and Schiff 2009; Reznick and Armon-Lotem 2022; Schiff et al. 2011) and the minimal, if any, relevance of maternal education to the acquisition of the morphological rule in our sample. Even at young ages, performance is at ceiling levels, regardless of maternal education (Schiff and Ravid 2012; Reznick and Armon-Lotem 2022). This is due to the characteristics of the Hebrew plural system, as mentioned earlier.

By contrast, the lexical and morpho-lexical aspects of the system, that is, the acquisition of the changing stems and/or irregular morphology, appear later and require exposure to these less frequent forms. That is, for the more lexical aspects of the system, there appears to be a gradual consolidation process of knowledge, which becomes evident at older ages. Our findings suggest that, as some children start to master these aspects in the second grade, maternal education impacts the rate of success. These findings align with the assumptions of usage-based models (Tomasello 2001) in language acquisition. As the process of forming and acquiring language rules is directly influenced by factors of frequency and complexity (e.g., Ambridge et al. 2015; Ravid et al. 2008; Ravid and Schiff 2009, 2012; Tatsumi et al. 2018), the superior performance of monolingual children with more educated mothers could be attributed to their exposure to richer and more varied linguistic input and more frequent, consistent, and diverse language usage opportunities (e.g., Diessel 2017; Dollaghan et al. 1999; Hoff 2003; Huttenlocher et al. 2010; Rowe 2008; Zambrana et al. 2012).

Unlike the findings for monolinguals, the current study revealed that, generally, maternal education did not contribute to bilingual children’s performance—not only in the highly regular aspects of the system but also in the irregular (lexical) aspects. As noted in the review of the literature, relatively few studies have focused on examining the relationship between maternal education and the language performance of bilingual children, though many have assumed that it is a relevant measure for SES, as in monolingual populations. The findings of the current study do not align with those of previous studies that have investigated this relationship and reported a positive correlation between maternal education and the lexical abilities of bilingual children in their societal language (i.e., better language performance with higher maternal education). For example, studies have found a positive correlation with receptive vocabulary (Calvo and Bialystok 2014), productive vocabulary and syntax (Meir and Armon-Lotem 2017), and a composite measure of SL lexical knowledge that includes both production and comprehension tasks (Schwartz et al. 2009b).

The discrepancy between the findings of different studies can be explained by two main arguments: significant methodological differences between the (few) studies and the presence of additional variables (potential mediators) that can influence the performance of bilingual children.

Firstly, the current study tested morpho-lexical knowledge, which pertains to the interface between morphology and the lexicon, and not the comprehension and production of lexical items. More specifically, it used a pluralization task with 99 items that assessed both inflectional morphology and its lexical aspects, while other studies used different tasks such as the Peabody Picture Vocabulary Test (Calvo and Bialystok 2014), a productive vocabulary task (Meir and Armon-Lotem 2017), and a composite measure from six lexical tasks that include both productive and comprehension tasks (Schwartz et al. 2009b). Knowledge of irregular morphological forms that are lexically governed at the interface between a lexicon and morphology seems to be more sensitive to the availability of more complex input, and possibly to crosslinguistic influence as well (Sorace 2004). Further research is definitely needed to understand the differential role of maternal education in different linguistic modules. Another explanation might be related to the way maternal education was used in the analysis, as different methods were employed in the different studies. In the current study, as in that of Schwartz et al. (2009b), maternal education was treated as a continuous variable (number of years of education). In contrast, Calvo and Bialystok (2014) and Meir and Armon-Lotem (2017) used a categorical variable with two maternal education groups: 12 years or less and more than 12 years of education (with a wide range of years of education in the latter group). Categorizing our data yielded groups too small to apply such an analysis.

As maternal education did not exhibit the expected impact on children’s morpho-lexical knowledge, the relevance of additional variables that could be potential mediators to bilingual children’s performance arises. Among these, FLP is a central factor contributing to bilingual children’s language performance (Altman et al. 2014; Armon-Lotem and Meir 2019; Dagenais and Day 1999; Hammer et al. 2009; Kennedy and Romo 2013; Rose et al. 2023). A consideration of other additional factors is discussed in Section 4.3.

4.2. Language Policy and Bilingual Children’s Performance in an SL

One of the central factors influencing the acquisition of the lexicon in an SL among bilinguals is the quantity and quality of linguistic input. Often, this input is a result of the FLP (Altman et al. 2014; Armon-Lotem and Meir 2019; Dagenais and Day 1999; Hammer et al. 2009; Kennedy and Romo 2013; Rose et al. 2023). Therefore, we examined the characteristics of the language policy among bilingual families. A key component of language policy is the practice component (Spolsky 2012), which is measured, among other things, by the choice of language/languages used at home in various interactive conditions.

In this study, it was found that the amount of linguistic input in Russian at home affected the bilingual child’s mastery of the Hebrew plural system: as the amount of input in HL at home increased, performance in the SL plural system decreased. The decline in performance in the SL was significant across all grade levels, and it was particularly pronounced in the lexical and morpho-lexical aspects of the plural system—words with a changing stem and/or irregular suffix. The negative relation between the use of an HL at home and morpho-lexical knowledge in an SL is indicative of a positive relation between the use of at least some SL at home (as when the use of both an SL and an HL is reported), or an SL at home and better performance in the SL. This finding aligns with numerous studies that have directly linked the characteristics of linguistic input to the acquisition of the lexicon and morpho-phonological/morpho-lexical knowledge in general (Ambridge et al. 2015; Diessel 2017; Tatsumi et al. 2018; Tomasello 2001) and in the SL particularly (Altman et al. 2014; Hoff et al. 2018; Kennedy and Romo 2013; Rose et al. 2023; Sorenson Duncan and Paradis 2020): the greater and more diverse the input in the SL, the better the performance in the SL.

The impact of FLP was again observed mainly for the lexical aspects of the plural system—namely, regardless of FLP in bilingual settings, children are still rapidly acquiring regular Hebrew plurals because of their transparency and abundance of occurrences, but they lack the acquisition of lexical/morphological aspects of the plural system due to limited input (Ravid 1995; Reznick and Armon-Lotem 2022; Schwartz et al. 2009a, 2014). This aligns with the importance of linguistic system characteristics such as transparency, frequency, regularity, and productivity in their acquisition (Ambridge et al. 2015; Diessel 2017; Tatsumi et al. 2018; Tomasello 2001). An additional catalyst in the case of Russian–Hebrew speakers is a certain similarity between the languages—both have a rich grammatical gender system and even a similar morpho-phonological marking of masculine and feminine in the singular (Corbett 1991). This similarity between the systems could facilitate the acquisition of morphological rules in the SL (DeKeyser 2005; Goldschneider and DeKeyser 2001; Reznick and Armon-Lotem 2022).

4.3. Open Questions, Limitations of the Study, and Future Research

The findings of previous studies suggested a relationship between maternal education and language policy (Kennedy and Romo 2013; Rose et al. 2023; Schwartz et al. 2009b). However, this hypothesis was refuted in the current study, which consistently showed the direct contribution of FLP to bilingual children’s performance. Notably, it was also found that maternal education does not positively impact children’s performance in the societal language, unlike the findings for monolinguals, and unlike the reports from the few studies that examined the contribution of maternal education to bilingual children’s performance.

We consider these findings to primarily invite further research for several reasons. Firstly, one key variable to consider when discussing the relationship between the number of years of maternal education and FLP is the language in which the mother’s education was acquired. If the education was obtained in the SL, higher education levels are likely to result in a tendency to use the SL more at home. Conversely, if the education was acquired in the HL, there would be a greater use of this language at home (Sorenson Duncan and Paradis 2020). The current study did not examine the language of education, which might explain the lack of correlation found between the number of years of maternal education and FLP. Moreover, we did not collect more detailed information regarding maternal education characteristics beyond the number of years of education. For example, the same number of years of education can reflect different types of education, which may impact the research findings. Likewise, the length of the mother’s stay in the SL country can also affect the choice of home language, and this information was not available for the current study. Therefore, characterizing mothers only by the number of years of education may be insufficient in the case of bilingual children. This group is highly heterogeneous, and even within the same family, there can be opposing approaches to language policy (Reichmuth 2024). It seems that there are certainly cases where a highly educated mother provides limited and poor linguistic input in the HL, for various reasons, and vice versa. Thus, the relationship between the number of years of maternal education and the FLP is influenced by a variety of additional factors, and it is advisable to continue investigating this relationship in more detailed and extensive future studies.

Furthermore, the measurement of language policy was limited to the number of interactive conditions in which each language or both languages together were used. Although this metric provides significant insights into the FLP practice, it still does not fully and effectively reflect the quantitative and qualitative characteristics of the linguistic input to which the child is exposed. To accurately assess the characteristics of linguistic input, a further detailed examination is necessary—considering various contexts (Abutbul-Oz and Armon-Lotem 2022), literacy-related aspects such as the number of books in each language available at home (Schwartz et al. 2009b), and even language samples from family interactions (Blum-Kulka 1994). Furthermore, it is essential to account for the composition of the familial and social environment (whether there are two parents or one and the number of siblings, grandparents, and friends), as well as the number of friends with whom frequent interactions occur—an aspect that was not addressed in this study beyond controlling for the variable of older siblings. Additionally, parents’ beliefs also correlate with the language policy practiced at home (Hwang et al. 2022), but ideology and management were not checked in our questionnaire. Furthermore, it appears that the prestige level of the HL also influences the language policy at home. For example, in the study by Rose et al. (2023), the HL was English, whereas in the current study, it was Russian. In both cases, the societal language was Hebrew, and the differences in the status of these two languages might have contributed to the variations in the tendency to use the HL at home among the participants’ families. Likewise, the use of three categories instead of a continuous scale between the languages (the use of Russian, the use of both languages, or the use of Hebrew) does not allow for distinguishing between using only Russian or mainly Russian. In the future, it would be better to use a continuous scale between the two languages.

Finally, the GLMMs in this study did not introduce noun genders, alongside other word characteristics, as a factor due to the small number of items per cell reflecting the frequency of highly irregular words in the language. The potential of such an analysis is suggested at the end of the Section 3. Thus, the current study did not allow for a definitive and well-supported answer concerning this effect due to the limited number of relevant items in Hebrew suitable for young children (e.g., only seven feminine words with a non-changing stem and an irregular suffix in this research; see Ornan 2003; Ravid et al. 2008).

4.4. Conclusions and Implications

The performance of bilingual children in the SL is related to the characteristics of FLP. This correlation is evident in the lexical aspects of the societal language but not in the highly regular and productive morphological system. These findings have clinical implications for both assessment and intervention processes when working with bilingual children. The question of the contribution of the mother’s education to the bilingual child’s performance in the SL, as well as understanding the connection between the mother’s education and the characteristics of the FLP, requires further research.

The findings of the study have direct implications for the linguistic assessment and intervention processes for bilingual children. The significant effect of FLP on the performance of bilingual children in the SL emphasizes the need to gather information about FLP during the assessment of a bilingual child’s language abilities and when planning their intervention program. Characteristics of the FLP can explain the language performance deficits of bilingual children, which should be considered when making diagnoses and/or recommendations for interventions.

The study also indicates that, during assessment and intervention, it is essential to clearly distinguish between the mastery of the morphological rule itself and the mastery of its exceptions, with the latter being primarily influenced by the FLP. Professionals should remember the importance of preserving and reinforcing the HL while simultaneously aiming for adequate and high-quality exposure to the societal language to ensure balanced bilingualism.

Additionally, the presence of a different pattern of the relationship between the two variables (number of years of maternal education and the success rate in Hebrew) in each linguistic status (bilinguals vs. monolinguals)—positive for monolinguals and either non-existent or negative for bilinguals—highlights the importance of considering maternal education in a differentiated and tailored manner for monolinguals and bilinguals.