How does maternal education influence the linguistic environment supporting bilingual language development in child second language learners of English?

Abstract

Aims:

In monolingual situations, mothers with higher levels of education are more likely to provide their children with enriched linguistic input. However, in bilingual situations, the relationship between maternal education and input is more complex because education may have occurred in one, but not both, of the languages. This study details this complexity by examining the relationship between maternal education and linguistic input in a group of children learning English as a second language (L2).

Method:

Participants were 89 immigrant/refugee children, living in Canada with diverse first languages (L1s). They completed a story-telling task in their L2 and parent reports provided a measure of L1 development. Parents also answered detailed questionnaires about family demographics and linguistic input within the family.

Results:

Regression modeling revealed that cumulative exposure to the L2 in school, maternal L2 fluency, relative quantity of L1/L2 use by the mother (input) and by the child (output) all influenced children’s L1 and L2 development. Relative quantity of input and maternal L2 fluency were, in turn, influenced by the amount of education mothers had completed and the language in which that education occurred. In instances where mothers were educated primarily in the L1, higher levels of education were associated with relatively more L1 input. For mothers who were educated predominantly in the L2, higher levels of education were associated with relatively more L2 input. Across both groups, mothers with higher levels of education had higher L2 fluency.

Conclusion:

This study revealed a complex relationship between the amount of education a mother has, the language in which that education occurred and the linguistic input she provides to her children. This complexity highlights that monolingual studies about input and environment may not offer a sufficiently nuanced perspective to account for the greater levels of variability found in bilingual situations.

Originality:

Previous research has offered valuable insights into the relationship between individual difference factors and bilingual acquisition. This study moves the field forward through the consideration of the interaction between factors in shaping bilingual acquisition.

Keywords

Bilingualism children input maternal education syntax lexical development

This study follows the lead of monolingual studies that have empirically evaluated the interdependencies between maternal education and linguistic input in children’s language development (e.g. Hart & Risley, 1995; Hoff, 2006). For example, in their seminal study, Hart and Risley (1995) reported that higher levels of education were associated with a greater quantity and quality of linguistic input and that these differences had profound effects on vocabulary development. For bilingual children, a similar relationship between maternal education and linguistic input is also predicted. The context of bilingual acquisition, however, is more complex because bilingual children can experience differences not only in how much language they hear but also differences in what language is used (Paradis & Grüter, 2014). In addition, mothers may have been educated in one but not both of the languages that their children are learning. As such, monolingual research may not offer a sufficiently nuanced perspective to account for the increased variability found in bilingual situations. To illustrate this, we detail the relationships between maternal education – in terms of both amount of education and the language of that education – linguistic input and bilingual acquisition among children learning English as a second language (L2).

Linguistic input and bilingual acquisition

For bilingual children who are already in school, linguistic input comes from a variety of sources (Paradis & Grüter, 2014). This study focuses on the input that children experience through conversations with their mothers. However, given that children’s input comes from multiple sources, an additional input variable is also considered in this study: cumulative exposure to the L2 at school. Through its inclusion, this study endeavors to provide an additional example as to the ways in which multiple sources of input are relevant to understanding bilingual development in children who are learning English as a L2.

Relative quantity of L1/L2 input from the mother

Research into the relationship between bilingual development and relative quantity of input has not reached a consensus, where relative input is a current (as opposed to cumulative) estimate of the proportion each language is used. In the case of first language (L1) development, the majority of researchers have reported that increases in the relative quantity of L1 input that school-aged children receive corresponds with increases in their L1 morphosyntax abilities and L1 vocabularies (e.g. Bohman et al., 2010; Mueller Gathercole et al., 2016; Place & Hoff, 2011; Prevoo et al., 2014). However, R. Jia and Paradis (2015) reported null results for the relationship between the relative language input children received and their Mandarin (L1) development. In the case of L2 development, several studies have demonstrated a positive relationship between relative L2 input and L2 abilities in school-aged children (e.g. Blom, 2010; Dijkstra, Kuiken, Joran, & Klinkenberg, 2016; G. Jia & Aaronson, 2003; G. Jia & Fuse, 2007; Mueller Gathercole et al., 2016; Prevoo et al., 2014; Unsworth, 2013). Yet, several other studies have reported null results for the relationship between relative L2 use at home and school-aged children’s emerging L2 skills (e.g. Chondrogianni & Marinis, 2011; Golberg, Paradis, & Crago, 2008; Páez et al., 2007; Paradis, 2011; Rojas et al., 2016). As a consequence, questions remain about the role of relative input in bilingual acquisition, especially in the case of L2 acquisition in a migrant context.

One reason for the above-noted discrepancies might be that previous studies have employed aggregate measures of relative input and each study may have aggregated across family members differently (e.g. Chondrogianni & Marinis, 2011; Collins et al., 2014; G. Jia & Fuse, 2007; Paradis, 2011). It is important to note that these aggregate input scores are robust and reliable measures of linguistic input (Paradis, 2017); however, from such aggregate measures, the impact of individual speakers, such as the mother, on children’s acquisition is difficult to ascertain. Presumably, maternal input is an important part of such aggregate scores and, consequently, as can be seen in the preceding paragraph, studies that used aggregate scores were taken as evidence as to the role of maternal input in bilingual development. In this study, maternal input was not aggregated with those of other family members.

Mother’s L2 fluency

Children from immigrant and refugee backgrounds often have mothers who are L2 learners of English and variation in mothers’ L2 proficiency has been suggested as a possible determinant of children’s L2 abilities (e.g. Golberg et al., 2008). For example, adult L2 learners of English often use less complex sentences when compared to native-speakers (e.g. Hinkel, 2003; O’Brien, Segalowitz, Collentine, & Freed, 2006). This reliance on simpler structures has predominantly been studied through written mediums, but similar effects are anticipated within oral language. As a consequence, parents with limited fluency in English may not be providing their children with sufficient exemplars of complex constructions to support the development of complex syntax, as has also been suggested for morphosyntax (e.g. Chondrogianni & Marinis, 2011; Hammer et al., 2012; Hoff, Welsh et al., 2014; Paradis, 2011).

Relative quantity of L1/L2 output from the child at home

Although the relative amount of output from children is an often overlooked factor in acquisition, more relative output has been associated with more advanced language skills (Bedore et al., 2012; Bohman et al., 2010; Paradis, 2011; Rojas et al., 2016). In fact, in the case of L2 development, the relative amount of output in the L2 at home appears to be more important as a determinant than the relative amount of input children receive from other family members (Bohman et al., 2010; Paradis, 2011). As such, studies that consider the role of maternal input in bilingual development must also take into account the extent to which the child’s own output is supporting their acquisition.

Cumulative exposure to the L2 at school

Interactions with people outside of the home, in particular at school, also provide important L2 input for child L2 learners (e.g. Altman, Armon-Lotem, Fichman, & Walters, 2016; Unsworth, 2013). As such, an additional input variable is also considered in this study: cumulative exposure to the L2 (English) at school. Not surprisingly, children who have had longer L2 exposure at school demonstrate greater L2 abilities (e.g. Altman et al., 2016; Hammer et al., 2012; G. Jia & Fuse, 2007; Paradis, 2011; Paradis, Rusk, Sorenson Duncan & Govindarajan, 2017; Unsworth, 2013). Notably, these past studies have tended to consider children who are slightly older and with greater amounts of exposure than is the case in the present study. Nevertheless, there is evidence that even small amounts of cumulative exposure can have profound effects on children’s emerging language abilities (Hoff, Rumiche et al., 2014). As such, to exemplify how multiple sources of input are relevant for child L2 learners, cumulative exposure at school was included in this study. Furthermore, the inclusion of cumulative exposure at school allows us to interpret the effect of maternal input on bilingual development above and beyond the variance that is predicted by school exposure (see Tables 3 and 4), at least at the onset of schooling.

Maternal education and the linguistic input children receive at home

Maternal education and mother’s relative L1/L2 use (input to the child)

Existing studies provide conflicting evidence about the relationship between mother’s education and her relative quantity of L1/L2 input to the child. For example, Prevoo et al. (2014) reported that higher levels of maternal education were associated with greater L2 (Dutch) and less L1 (Turkish) use among bilingual families in the Netherlands. This trend was also noted amongst Spanish–English bilingual families in the United States, where education was associated with higher proportions of English (L2) use (e.g. Bohman et al., 2010; Oller & Eilers, 2002). However, other studies have reported the opposite effect: more highly educated mothers were more likely to speak the L1 (e.g. Golberg et al., 2008; Hammer et al., 2012; R. Jia & Paradis, 2015; Mueller Gathercole et al., 2016). Thus, in the context of bilingual development, the interdependencies between maternal education and the input children receive are not straightforward, and likely influenced by additional factors.

One such factor is the language of the mother’s education. In fact, some research suggests that the language of education may have a greater influence on children’s emerging language abilities than the amount of education a mother has obtained (e.g. Hoff & Giguere, 2015). As such, one potential explanation is that increased L2 input to children is related to increased L2-medium education, as was the case for the mothers in Prevoo et al.’s (2014) study. In contrast, increased education levels are hypothesized to be associated with increased L1 input to children when mothers were educated in the L1, as was the case for the first-generation immigrants in Hammer et al.’s (2012) study. Such detailed information about parental background, however, is rarely included in the participant details of child acquisition studies and, consequently, it is not possible to fully evaluate whether such differences in the language of education might underlie the conflicting results as to the relationship between maternal education and relative input.

Maternal education and maternal L2 fluency

It would appear that the more education a parent has received in the L2, the better their L2 proficiency and hence the more effectively they can model the L2 for their children. However, this connection between education and fluency is largely taken for granted. Research explicitly testing this prediction is scarce. Existing research suggests that fluency level in Spanish–English mothers is linked to where that education was completed (e.g. Bohman et al., 2010; Hammer et al., 2012; Hoff & Giguere, 2015). Crucially, the location of education is synonymous in these cases for the language of that education. Bohman et al. (2010) found that parents who were more likely to have been educated in Spanish-speaking countries had children with higher Spanish language scores, presumably because these parents provided higher quality Spanish input to their children. In terms of L2 development, Hammer et al. (2012) reported that parents who were more likely to have been educated in the United States had children with higher English scores, presumably because the parents had more advanced English fluency and thus provided higher quality input. In summary, maternal fluency, like maternal relative input to the child, is predicted to be connected to both the amount of education and the language of that education.

Research questions

This study first considered the relationship between input and bilingual acquisition within a sample of child L2 learners.

How does the relative quantity of L1 input that children receive from their mother impact their L1 acquisition?

How does the relative quantity of L2 input that children receive from their mother impact their acquisition of L2 syntax and L2 vocabulary?

How does maternal L2 fluency influence children’s acquisition of L2 syntax and L2 vocabulary?

How does cumulative exposure to the L2 at school influence L2 syntax and L2 vocabulary development in the early stages of acquisition?

Having established that these input factors are determinants of acquisition for this sample of child L2 learners, the next goal of this study was to investigate the relationship between maternal education and the linguistic input children receive at home from their mothers.

5. How does maternal education influence the mother’s relative language use (input to the child)? Does the language in which the mother was educated impact this relationship?

6. How does education influence maternal L2 fluency? Does the language in which the mother was educated impact this relationship?

Method

Participants

Participants were 89 children who spoke their L1 at home and were learning English as a L2, primarily through preschool or kindergarten. These children were all registered in half-day school programs (2.5–3 hours/day) that were conducted in English and were either four or five days a week. All of the children had foreign-born parents. Thirty-two of the children were also foreign-born. Children were all typically-developing language learners and there was no indication of any hearing difficulties or cognitive impairments, according to parent reports. These children had diverse L1 backgrounds: Arabic, Cantonese, Farsi, Gujarati, Hindi, Mandarin, Punjabi, Somali, Spanish and Urdu. Demographic information is summarized in Table 1.

Table 1.

Summary of demographic information of participants.

Demographic variable	Mean	SD	Range
Age	60.43 months (5;0)	4.17 months (0;4)	50–66 months (4;2–5;6)
Cumulative L2 exposure at school (in months)	12.72 months	8.44 months	2–36 months
Mother’s level of education	11.95 years	5.00 years	0–20 years
Mothers’ relative L2 use to the child (input)	0.20	0.21	0–0.75
Mother’s (self-reported) L2 fluency	2.43 out of 4	1.13	0–4

L2: second language.

There were two distinct profiles of education amongst the mothers in this study. Sixty-three of the mothers were educated in the L1, pre-migration (Group A). Twenty-six of the mothers came to Canada from Somalia as refugees and these women were educated mostly post-migration in English (the L2 in this study) (Group B). In the case of the women from Somalia, civil war severely limited access to education prior to migration (Multicultural Family Connections Program, 2004). As a consequence, higher levels of education are likely associated with educational experiences post-migration, which often occurred in English when they were adults (specific examples from this sample of mothers include English education at a refugee camp in Kenya and adult education courses in Canada). In contrast, mothers in Group A had access to education throughout their childhood, adolescence and young adulthood. Consequently, they completed substantial amounts (if not all) of their education in L1-based programs, prior to migration.

It is worth noting that this sample of children formed the basis of the first author’s doctoral dissertation (Sorenson Duncan, 2017) and further details about the participants can be found there. Also, as detailed by Sorenson Duncan (2017), this sample contains an overlapping but not identical sample to several previous studies. Crucially, these previous studies did not include children from Somalia, who comprise Group B here. Furthermore, of these previous studies, only Paradis (2011) considered maternal education as a determinant of language acquisition but not its influence on maternal input and for different linguistic measures than those of the present study.

Procedures

Alberta Language and Environment Questionnaire

The Alberta Language and Environment Questionnaire (ALEQ; Paradis, 2011; Paradis, n.d.) was used to gather detailed demographic information, including maternal education and maternal input. It is administered as an oral interview between the parent and the researcher, often with the assistance of an interpreter/cultural broker. For the present study, the following information was included: the relative quantity of L1/L2 input that the mother provided to her child; the mother’s self-rated L2 fluency (i.e. English fluency); the child’s relative L1/L2 output and cumulative exposure to the L2 in school. Relative quantity of maternal input was scored on a five-point scale (0–1), ranging from no-English/all-L1 to all-English/no-L1. For the purposes of regression modeling, this scale was converted to a log odds ratio of English to L1. Maternal L2 fluency was also scored on a five-point scale (0–4), ranging from no English fluency to highly fluent. For the purposes of regression modeling, these scores were collapsed into a binary distinction of low fluency (scores of 2 or less) and high fluency (scores of 3 or 4). There were 45 mothers in the low-fluency group and 44 mothers in the high-fluency group.

Alberta Language Development Questionnaire

The Alberta Language Development Questionnaire (ALDeQ; Paradis, Emmerzael, & Sorenson Duncan, 2010; Paradis, n.d.) was used to generate L1 scores. Like the ALEQ, it is an oral interview conducted between the parent and researcher, often with the assistance of an interpreter/cultural broker. This particular questionnaire was specifically designed for use with children from diverse L1 backgrounds in multilingual contexts (Paradis et al., 2010). Section B: Current L1 Abilities, which asks parents a series of questions about their child’s current L1 abilities, was administered. On this section, parents rate their child’s productive language abilities compared to other children’s skills on a scale ranging from 0 to 3. Responses for each question are summed to obtain children’s L1 scores. The maximum possible score is 18/18 or 1.00.

Edmonton Narrative Norms Instrument

The Edmonton Narrative Norms Instrument (ENNI) story-generation activity was used to sample children’s productive L2 abilities (Schneider, Dubé, & Hayward, 2005). Stories were video-recorded and later transcribed and analyzed by a native-speaker of English. A second researcher transcribed a subset of the data to verify the reliability of these transcripts. The transcription reliability was 91% (word-by-word). Transcripts were then analyzed for L2 syntactic and lexical development. Following Paradis et al. (2017), syntactic development was estimated by comparing the number of simple and complex sentences that children produced. To estimate lexical development, a type:token ratio was calculated by dividing the number of distinct word types by the total word tokens (e.g. Roy, Frank, & Roy, 2009). These two language variables were chosen because prior research has noted the influence of maternal education and of input factors on these linguistic sub-domains (e.g. Hammer et al., 2012; Oller & Eilers, 2002; Wiechmann, Kerz, & Steinfeld et al., 2016).

Results

Descriptive statistics

L1 abilities

The mean ALDeQ B score was 0.65 (SD = 0.27, range = 0.00–1.00), suggesting that children, on average, had strong L1 abilities. In fact, 64% of children scored 0.67 or higher. Scores of 0.67 or higher indicate that parents ranked them as having appropriate to above-average language skills. Despite these high L1 scores, half of the parents (51%) reported attrition in their child’s L1 abilities and a preference for English compared to the L1.

L2 abilities

The child L2 learners, on average, produced 55.33 simple sentences (SD = 21.42, range = 9–125) and 10.27 complex sentences (SD = 7.62, range = 0–34). On average, these children produced 395.90 total words (SD = 157.31, range = 78–935) when completing the story-telling task and 103.57 different word types (SD = 29.72, range = 46–199), leading to an average type:token ratio of 0.27 (SD = 0.07, range = 0.16–0.60).

Regression modeling

Regression modeling was used to evaluate sources of individual differences. For each outcome variable, backwards elimination was used to arrive at the optimal model from a full-model that overfit the data. Predictors that were not significant were removed in a step-wise fashion to identify the optimal model, which was verified by comparing nested models using maximum likelihood ratio tests. Logistic regression was used in all instances, except for the models with L1 abilities as the outcome variable. In this case, linear regression was used.

Determinants of children’s L1 abilities

Children’s L1 abilities were estimated using Section B (Current Language Abilities) of the ALDeQ (Paradis et al., 2010). To avoid a proportion score, the numerator of the ALDeQ score was used as the outcome variable. The denominator for all children was 18; as such, the numerator conveys the same information as the proportioned score. Linear regression modeling revealed that the relative quantity of English (L2) input that the mother provided to the child and the relative amount of English that the child spoke at home significantly and negatively influenced children’s L1 abilities (maternal input: estimate =−1.594, p = 0.005; child’s output: estimate =−1.622, p = 1.66e–05). That is, the more English (L2) that was used by the mother (input) and by the child (output), the lower children’s L1 scores. The cumulative exposure to English (L2) in school, mother’s L2 fluency, L2 syntactic and L2 lexical abilities did not significantly predict L1 scores (see Table 2).

Table 2.

Determinants of children’s first language scores.

	Estimate	t	p
Intercept	15.806	21.549	<2e-16
Mother’s relative L2 input	−1.594	−2.889	0.005
Child’s relative L2 output	−1.622	−4.566	1.66e-05

L2: second language.

Determinants of children’s L2 syntactic abilities

Children’s L2 (English) abilities were measured using a story-generation task (ENNI; Schneider et al., 2005). L2 development of complex syntax was estimated by tabulating children’s use of simple and complex sentences. For the purposes of regression modeling, the log odds ratio of complex to simple clauses was the outcome variable. The relative amount of English (L2) spoken by the child to the mother (output), the cumulative exposure to English in school and mother’s L2 fluency were significant and positive predictors of children’s complex syntax use (child’s output: estimate = 0.098, p = 0.004; school exposure: estimate = 0.020, p = 1.05e-05; maternal fluency: estimate =−0.348, p = 1.68e-05). That is, greater relative English use by the child (output), more time in school and having a mother with higher English fluency were all associated with greater use of complex sentences in the narrative task. The relative amount of English input from the mother was also a significant, but negative, predictor of complex syntax (estimate = −0.113, p = 0.034). That is, children whose mothers spoke more English (L2) to them (input) used fewer complex sentences in their L2 stories (see Table 3). L1 abilities were not a significant predictor of use of L2 complex syntax.

Table 3.

Determinants of children’s second language (L2) complex syntax use.

	Estimate	z	p
Intercept	−1.853	−16.230	<2e-16
Cumulative L2 exposure at school	0.020	4.407	1.05e-05
Mother’s relative L2 input	−0.113	−2.076	0.034
Child’s relative L2 output	0.098	2.869	0.004
Mother’s L2 fluency: low^a	−0.348	−4.03	1.68e-05

The high-fluency group is the reference level for this variable.

Determinants of children’s L2 lexical abilities

L2 lexical abilities were also estimated from the story-telling generation task (ENNI; Schneider et al., 2005). In this case, the total number of words (tokens) and the total number of different words (types) were tallied. For the purposes of regression modeling, the log odds ratio of types to tokens was the outcome variable. Higher type frequency was positively and significantly predicted by cumulative exposure to English (L2) at school and the relative amount of English (L2) spoken by the mother to the child (input) (maternal input: estimate = 0.059, p = 8.82e-05; school exposure: estimate = 0.003, p = 0.028). That is, children who had been in school longer and heard more relative English from their mothers produced a greater number of word types on an English story-telling task. Mother’s L2 fluency, the relative amount of L2 spoken by the child to the mother (output) and L1 abilities were not significant predictors of this L2 vocabulary measure (see Table 4).

Table 4.

Determinants of children’s second language (L2) productive vocabulary.

	Estimate	z	p
Intercept	−1.129	−41.133	<2e-16
Mother’s relative L2 input	0.059	3.921	8.82e-05
Cumulative exposure to L2 at school	0.003	2.200	0.028

Determinants of relative quantity of input from the mother

The relative amount of maternal input was estimated through a parental interview. For the purposes of regression modeling the score was converted to a log odds ratio of English to L1 for the outcome variable. Mother’s relative quantity of English use with her child (input) was significantly and positively predicted by the relative amount of English the child spoke to the mother (output) (estimate = 0.418, p = 0.0002). The amount of maternal education also significantly predicted the relative amount of English spoken by the mother to the child. This relationship, however, was mediated through an interaction with language of education (for the interaction: estimate = 0.192, p = 0.014). Within Group A (i.e. the group of mothers who were educated predominantly in the L1), higher levels of education were associated with more relative L1 input to the child (and less L2 input). In contrast, within Group B (i.e. the group of mothers who were educated in the L2), higher levels of education were associated with more relative L2 (English) input to the child (see Table 5 and Figure 1).

Table 5.

Determinants of the amount of second language (L2) spoken by the mother to the child (input).

Model A	Estimate	z	p
Intercept	−1.282	−1.691	0.091
Amount of mother’s level of education (in years)	−0.065	−1.256	0.209
Group: Group B^a	−1.953	−2.131	0.033
Child’s relative L2 use to mother (output)	0.418	3.751	0.0002
Amount of education * Group:B^a	0.192	2.448	0.014
Model B	Estimate	z	p
Intercept	−1.277	−6.771	1.28e-11
Amount of mother’s level of education (in years) (scaled)	−0.234	−0.893	0.372
Group: Group B^a	0.046	0.116	0.907
Child’s relative L2 use to mother (output) (scaled)	0.618	3.895	9.84e-05
Amount of education * Group:B^a	0.802	2.078	0.038

Group A (first language-based education) is the reference level.

Figure 1.

More education leads mothers who were predominantly educated in the first language (L1) to speak relatively more of the L1 (and less of the L2 [English]) to their children, whereas the opposite effect is found for mothers who were predominantly educated in the L2.

It is important to note that the intercept in Model A is approaching significance, but did not reach significance. Non-significant intercepts are not, in principle, a problem (e.g. Peng et al., 2002). They suggest that if all predictors had a value of zero then the outcome variable would be zero (Frost, 2013). That is, in this case, the mother would only speak in the L1 to her child. Although this mathematically makes sense, we must be certain that the math represents a plausible reality (Frost, 2013). In this scenario, a complete lack of English (L2) is unlikely because English is the majority language and lingua franca in this community (Golberg et al., 2008). In this case, the varied scales of the predictor variables likely resulted in the non-significant intercept. Maternal education (in years) is a much larger scale (ranging from 0–20) than the relative amount of English (ranging from 0–1). Consequently, variables were scaled before modeling. Scaling creates variables on similar scales to one another, which, in this case, led to the expected significant intercept without affecting the interpretation of the results (see Table 5 – Model B).

Determinants of maternal L2 fluency

Maternal fluency was estimated based on the self-reported fluency during the parental interview. As noted in the Procedures section, mothers were classified as having either low or high L2 (English) fluency. This binary fluency measure was the outcome variable for the logistic regression modeling. Mothers were more likely to be in the high English fluency group if they had higher levels of education, although this effect was more pronounced in Group A (L1-based education) compared to Group B (L2-based education), as indicated by the difference in the slopes in Figure 2 (see Table 6) (for the interaction: estimate =−0.368, p = 0.036). The relative amount of L2 (English) spoken by the child to the mother (output) was not a significant predictor of the mother’s L2 fluency.

Table 6.

Determinants of the mother’s English (second language) fluency.

	Estimate	z	p
Intercept	−8.199	−3.957	7.58e-05
Mother’s amount of education (in years)	0.580	4.040	5.34e-05
Group: Group B^a	6.560	2.928	0.003
Amount of education * Group:B^a	−0.368	−2.102	0.036

Group A (first language-based education) is the reference level.

Figure 2.

Higher levels of education are associated with higher second language (English) fluency, especially within the group of mothers who were educated predominantly in the first language (Group A).

Discussion

This study considered children’s language in relation to the linguistic input they received and, in turn, this input was considered in relation to mothers’ amount of education and the language of that education. The relative quantity of L1/L2 input that these children received from their mothers, the relative amount they spoke each language (output), their mother’s L2 fluency and children’s cumulative L2 exposure in school were all found to be significant predictors of their bilingual development. Maternal education was found to predict both maternal L2 fluency and maternal relative L1/L2 use (input). However, this relationship was not straightforward because this effect was influenced by the language in which the mothers’ education had occurred. As discussed below, mothers with higher levels of L1-based education were less likely to use English (the L2) with their children, whereas mothers who had higher levels of education in L2-based programs were more likely to use English when speaking to their children.

Linguistic input and bilingual acquisition

Relative quantity of L1/L2 input from the mother

In this study, children who received more relative L1 input from their mothers had higher L1 scores. Notably, increases in relative L1 input corresponded with decreases in relative L2 input. This finding aligns with the dominant trend in previous research, which also suggests a positive relationship between the relative language input that children hear in the minority language and their proficiency in that language (e.g. Bohman et al., 2010; Mueller Gathercole et al., 2016; Place & Hoff, 2011; Prevoo et al., 2014). The effect of relative input from the mother on children’s L2 development was more nuanced. For syntactic development, children who received more relative L2 input used less complex sentences. However, for lexical development, children who received more relative L2 input from their mothers used more different words. In interpreting these seemingly conflicting findings, it is important to consider the nature of the input. On one hand, adult L2 learners are known to use simpler sentences (e.g. Hinkel, 2003; O’Brien, et al., 2006); as such, input from these mothers may have lacked sufficient exemplars of complex sentences to support these children’s development in this domain, as has also been reported for older school-aged children (e.g. Chondrogianni & Marinis, 2011; Paradis, 2011). Vocabulary, on the other hand, is a skill with which adult L2 learners make steady gains (e.g., Coady, 1997), perhaps allowing mothers to more easily support their young children’s L2 vocabulary development, as was the case in this study. It is worth noting that the children in this study were enrolled in half-day educational programs. It is possible that as they begin to attend full-day programs, maternal input may have a more limited effect on their vocabulary scores (e.g. Golberg et al., 2008; Paradis, 2011).

Mothers’ L2 fluency

The above interpretation is supported by the results for maternal L2 fluency. This quality of input factor supported these children’s L2 syntactic development but had no effect on their L2 lexical acquisition. That is, children with mothers in the high-fluency group used more complex syntax. This finding aligns with previous research about morphosyntactic development that suggests that mothers with higher levels of L2 fluency are less likely to omit grammatical morphemes and thus more likely to provide the necessary exemplars to support morphosyntactic development (e.g. Chondrogianni & Marinis, 2011; Hammer et al., 2012; Hoff, Welsh, et al., 2014; Paradis, 2011). Future research that explicitly examines the distribution of complex sentences in L2 mothers’ speech is still required.

Relative quantity of L1/L2 output from the child at home

The children’s relative language output was related to both their L1 and L2 development. Children with higher levels of relative English language use (output) had lower L1 scores and higher L2 syntactic scores. This finding aligns with previous studies, which have also highlighted the importance of considering children’s language output as a determinant of their acquisition (Bedore et al., 2012; Bohman et al., 2010; Paradis, 2011; Rojas et al., 2016).

Cumulative exposure to the L2 in school

As predicted, even in the early stages of L2 acquisition, cumulative exposure to English is a positive and significant predictor of children’s L2 development. That is, in this study, children who had been in school longer used more complex sentences and a larger variety of words when completing the L2 story-telling task. This finding is expected, given the wide range of studies that have previously reported a relationship between the amount of cumulative L2 exposure and L2 development (e.g. Altman et al., 2016; Hoff, Rumiche et al., 2014; G. Jia & Fuse, 2007; Paradis, 2011; Paradis et al., 2017; Unsworth, 2013). The noteworthy point is that acquisition was simultaneously influenced by this school-based input variable and the previously discussed home input variables. That is, the inclusion of this variable exemplifies previous claims that multiple sources of input are relevant to preschool and school-aged children’s language acquisition (e.g. Paradis & Grüter, 2014).

Maternal education and the linguistic input children receive at home

In this study, the input children received was considered in relation to maternal education. Based on monolingual research, the prediction was that children who have mothers with higher levels of education will hear more language and also better quality language than children whose mothers have lower levels of education. This study investigated this prediction in a group of child L2 learners and found that the relationship between education and input is more nuanced for this group.

Maternal education and mother’s relative language input to the child

Maternal education influenced the relative proportion of mothers’ L1/L2 use with their children. However, this relationship was not straightforward because it was influenced through an interaction with the language of that education. That is, in Group A, higher levels of education were associated with less relative English (L2) input from the mother to the child. Notably, the mothers in this group were educated predominantly in the L1. This finding aligns with those of some previous studies (e.g. Golberg et al., 2008; Hammer et al., 2012; R. Jia & Paradis, 2015; Mueller Gathercole et al., 2016). In contrast, in Group B, higher levels of education were associated with more relative L2 input from the mother to the child. Notably, because of civil war, the mothers in this group were educated predominantly in the L2 (see participant details in the Method section). Interestingly, this finding aligns with those of other studies (e.g. Bohman et al., 2010; Oller & Eilers, 2002; Prevoo et al., 2014).

In summary, both of the seemingly conflicting findings from the previous literature are represented within this single study. This apparent conflict arises when only the highest level of education achieved by mothers is considered. The interaction reported in this paper highlights that the relative quantity of input that mothers provide to their children is best understood from a more nuanced perspective that considers both the amount and the language of education.

Maternal education and maternal L2 fluency

Maternal fluency was related to maternal education overall (regardless of the language of education); that is, mothers with higher levels of education reported having higher levels of L2 proficiency. This finding is somewhat surprising, given the interaction noted just above. That is, differences were expected based on the language of education of the two groups. Furthermore, previous research has suggested that when mothers are educated primarily in their home country before migration, they have lower proficiency than mothers who attended L2-medium education, for example, education in English after moving to the United States (e.g. Hammer et al., 2012). However, this was not the case in this study; fluency levels were comparable across the two groups (the mean for Group A [L1-based education] was 2.46/4 and for Group B [L2-based education], 2.35/4), suggesting that mothers who experienced L2-based instruction were not more proficient than those who were primarily educated prior to migration.

In seeking to understand the comparable proficiency between these different groups, it seems that different types of education may have contributed to comparable proficiency levels. In the case of Group A, increased access to foreign-language courses (presumably English classes) is hypothesized to underlie the finding that education and English (L2) fluency are positively connected for these women. In Group B, in contrast, increased English-medium schooling post-migration is hypothesized to contribute to their proficiency levels. Consequently, more education, regardless of where it took place, likely involved more English language training. In interpreting this finding, it is important to keep in mind that these two groups were not equivalent in the absolute level of education that was attained; Group B generally had lower levels of education, as is expected given that civil war caused an extreme interruption to their education. As such, it is expected that if the Somali women (Group B) had completed higher levels of education, their L2 proficiency would surpass those who completed post-secondary education prior to migration, but who studied English as a foreign language. Future research that considers specific details about maternal education across more educational scenarios and in a more nuanced way is, thus, necessary in order to understand the relationship between mothers’ level of education, their L2 proficiency and children’s bilingual development.

Conclusion

Children learn language from the linguistic input they receive. However, input varies on an individual basis, and this variation leads to differences in children’s emerging language skills (e.g. Lieven, 2010). One explanation for this variation is that the environments in which children live shape the specific linguistic input that they receive (e.g. Hart & Risley, 1995; Hoff, 2006). To test the extent that this explanation can be extended to bilingual populations, we examined the relationship between maternal education and linguistic input. As was expected, we found that a more nuanced perspective on these variables was necessary. Understanding bilingual development, thus, requires that researchers not only consider individual variables but also rich details about how input and environment interact to shape acquisition.

Footnotes

Acknowledgements

We would like to thank the parents and children who participated in this research, and the Edmonton Public School Board who assisted in recruiting families. Special thanks go to the Multicultural Health Brokers Cooperative for their assistance in recruitment and in providing interpretation services. Finally, we would like to express our appreciation to the agencies who provided financial support for this research.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding was provided to Tamara Sorenson Duncan from the Vanier Canada Graduate Scholarship (770-2010-0038) and the Killam Foundation (Isaac Walton Killam Doctoral Scholarship at the University of Alberta, Edmonton, AB and IWK Research Fellowship at the IWK Health Centre, Halifax, NS: #22266). Funding was provided to Johanne Paradis from the Alberta Heritage Foundation for Medical Research (Population Health Investigator Award #200800174), the Alberta Centre for Child, Family and Community Research (#090415INV-Paradis), the Canadian Language and Literacy Research Network (Investigator Initiated Research Award #27061500) and Early Child Development Support Services, Edmonton (Innovative Research Grant – 2008).

Author biographies

Tamara Sorenson Duncan holds a joint post-doctoral position between the Autism Research Centre at the IWK Health Centre and the Language and Literacy Lab at Dalhousie University, Halifax, Nova Scotia, Canada. Her main research interests are (a) the input and environment factors that lead to language learning success for children from migrant backgrounds and (b) the relationship between language development and reading comprehension in children from a variety of backgrounds, including monolingual and bilingual children and children with autism spectrum disorder.

Johanne Paradis is a professor of Linguistics at the University of Alberta, Edmonton, AB, Canada. Her main research interests involve children learning English as a second language who are from immigrant and refugee families, including children with typical development, with developmental language disorders and with autism spectrum disorder. Specifically, her work details how these children approach native-speaker competence, what unique language development profiles they display and the factors explaining why some individual children learn English at different rates than others.

References

Altman

Armon-lotem

Fichman

Walters

(2016). Macrostructure, microstructure, and mental state terms in the narratives of English – Hebrew bilingual preschool children with and without specific language impairment. Applied Psycholinguistics, 37, 165–193.

Bedore

L. M.

Peña

E. D.

Summers

C. L.

Boerger

K. M.

Resendiz

M. D.

Greene

Gillam

R. B.

(2012). The measure matters: Language dominance profiles across measures in Spanish–English bilingual children. Bilingualism: Language and Cognition, 15(03), 616–629.

Blom

(2010). Effects of input on the early grammatical development of bilingual children. International Journal of Bilingualism, 14, 422–446.

Bohman

T. M.

Bedore

L. M.

Peña

E. D.

Mendez-Perez

Gillam

R. B.

(2010). What you hear and what you say: Language performance in Spanish–English bilinguals. International Journal of Bilingual Education and Bilingualism, 13, 325–344.

Chondrogianni

Marinis

(2011). Differential effects of internal and external factors on the development of vocabulary, tense morphology and morpho-syntax in successive bilingual children. Linguistic Approaches to Bilingualism, 3, 318–345.

Coady

(1997). L2 vocabulary acquisition: A synthesis of research. In Coady

Huckin

(Eds.), Second language vocabulary acquisition: A rational for pedagogy (pp. 273–290). Cambridge: Cambridge University Press.

Collins

O’Connor

Suárez-Orozco

Nieto-Castañon

Toppelberg

(2014). Dual language profiles of Latino children of immigrants: Stability and change over the early school years. Applied PsychoLinguistics, 35, 581–620.

Dijkstra

Kuiken

Joran

R. J.

Klinkenberg

E. L.

(2016). The role of majority and minority language input in the early development of a bilingual vocabulary. Bilingualism: Language and Cognition, 19(1), 191–205.

Frost

(2013). Regression analysis: How to interpret the constant (Y intercept). Retrieved from http://blog.minitab.com/blog/adventures-in-statistics/regression-analysis-how-to-interpret-the-constant-y-intercept (accessed 29 August 2016).

10.

Gathercole

S. E.

(2006). Nonword repetition and word learning: The nature of the relationship. Applied Psycholinguistics, 27, 513–543.

11.

Golberg

Paradis

Crago

(2008). Lexical acquisition over time in minority first language children learning English as a second language. Applied Psycholinguistics, 29(01), 41–65.

12.

Hammer

C. S.

Komaroff

Rodriguez

B. L.

Lopez

L. M.

Scarpino

S. E.

Goldstein

(2012). Predicting Spanish – English bilingual children’s language abilities. Journal of Speech, Language, and Hearing Research, 55, 1251–1264.

13.

Hart

Risley

T. R.

(1995). Meaningful differences in the everyday experiences of young American children. Baltimore, MD: Paul H. Brookes Publishing.

14.

Hinkel

(2003). Simplicity without elegance: Features of sentences in L1 and L2 academic texts. TESOL Quaterly, 37, 275–301.

15.

Hoff

(2006). How social contexts support and shape language development. Developmental Review, 26(1), 55–88.

16.

Hoff

Giguere

(2015). Bilingual mothers’ language of education, not level of education, predicts children’s bilingual development. Paper presented as part of the thematic session, Characterizing maternal contributions to the language environment of emerging, Spanish-English bilingual children. International Symposium in Bilingualism, New Brunswick, NJ.

17.

Hoff

Rumiche

Burridge

Ribot

K. M.

Welsh

S. N.

(2014). Expressive vocabulary development in children from bilingual and monolingual homes: A longitudinal study from two to four years. Early Childhood Research Quarterly, 29(4), 433–444.

18.

Hoff

Welsh

Place

Ribot

(2014). Properties of dual language input that shape bilingual development and properties of environments that shape dual language input. In Grüter

Paradis

(Eds.), Trends in language acquisition research: Input and experience in bilingual development (pp. 119–140). Amersterdam: John Benjamins Publishing Company.

19.

Jia

Aaronson

(2003). A longitudinal study of Chinese children and adolescents learning English in the United States. Applied Psycholinguistics, 24, 131–161.

20.

Jia

Fuse

(2007). Acquisition of english grammatical morphology by native mandarin-speaking children and adolescents: Age-related differences. Journal of Speech, Language, and Hearing Research : JSLHR, 50(5), 1280–1299.

21.

Jia

Paradis

(2015). The use of referring expressions in narratives by Mandarin heritage language children and the role of language environment factors in predicting individual differences. Bilingualism: Language and Cognition, 18(04), 737–752.

22.

Lieven

(2010). Input and first language acquisition: Evaluating the role of frequency. Lingua, 120(11), 2546–2556.

23.

Mueller Gathercole

V. C.

Hoff

. (2006). Input and the acquisition of language: Three questions. In Hoff

Shatz

(Eds.), Blackwell handbook of language development (pp. 107–127). Oxford, UK: Blackwell Publishing.

24.

Mueller Gathercole

V. C.

Kennedy

Thomas

E. M

. (2016). Socioeconomic level and bilinguals ’ performance on language and cognitive measures. Bilingualism: Language and Cognition Language and Cognition, First View, 19(5), 1057–1078.

25.

Multicultural Family Connections Program. (2004). Mapping the life experiences of refugee and immigrant families with preschool children. (Unpublished raw data).

26.

O’Brien

Segalowitz

Collentine

Freed

(2006). Phonological memory and lexical, narrative, and grammatical skills in second language oral production by adult learners. Applied Psycholinguisitcs, 27, 377–402.

27.

Oller

D. K.

Eilers

R. E.

(2002). Language and literacy in bilingual children. Clevedon, UK: Multilingual Matters.

28.

Páez

M. M.

Tabors

P. O.

López

L. M.

(2007). Dual language and literacy development of Spanish-speaking preschool children. Journal of Applied Developmental Psychology, 28, 85–102.

29.

Paradis

(n.d.). CHESL Centre. Retrieved from http://www.linguistics.ualberta.ca/CHESL_Centre.aspx (accessed 11 May 2016).

30.

Paradis

(2011). Individual differences in child English second language acquisition: Comparing child-internal and child-external factors. Linguistic Approaches to Bilingualism, 1(3), 213–237.

31.

Paradis

(2017). Commentary: Parent report data on input and experience reliably predict bilingual development and this is not trivial. Bilingualism: Language and Cognition, 20(1), 128–129.

32.

Paradis

Emmerzael

Sorenson Duncan

(2010). Assessment of English language learners: Using parent report on first language development. Journal of Communication Disorders, 43(6), 474–497.

33.

Paradis

Grüter

(2014). Introduction to “Input and experience in bilingual development.” In Paradis

Grüter

(Eds.), Trends in language acquisition research: Input and experience in bilingual development (pp. 1–14). Amsterdam, Nl: John Benjamins Publishing Company.

34.

Paradis

Rusk

Sorenson Duncan

Govindarajan

(2017). English second language children’s acquisition of complex sentences: The role of age, input and cognitive factors. Annual Review of Linguistics, 20, 1–20.

35.

Peng

C. J.

Lee

K. L.

Ingersoll

G. M.

(2002). An introduction to logistic regression analysis and reporting. Journal of Educational Research, 96(1), 3–14.

36.

Place

Hoff

(2011). Properties of dual language exposure that influence 2-year-olds’ bilingual proficiency. Child Development, 82(6), 1834–1849.

37.

Prevoo

M. J. L.

Malda

Mesman

Emmen

R. A. G.

Yeniad

Ijzendoorn

M. H. V.A. N.

Linting

(2014). Predicting ethnic minority children’s vocabulary from socioeconmoic status, maternal language and home reading input: different pathways for the host and ethnic language. Journal of Child Language, 41, 963–984.

38.

Rojas

Iglesias

Bunta

Goldstein

Goldenberg

Reese

(2016). Interlocutor differential effects on the expressive language skills of Spanish-speaking English learners. International Journal of Speech-Language Pathology, 18, 166–177.

39.

Roy

B. C.

Frank

M. C.

Roy

(2009). Exploring word learning in a high-density longitudinal corpus. In Proceedings of the Thirty-First Annual Conference in Cognitive Science Society (pp. 2106–2111). Austin, TX: Cognitive Science Society, Inc.

40.

Santrock

J. W.

(2001). Child development (9th Edition). New York, NY: McGraw-Hill Higher Education.

41.

Scheele

A. F.

Leseman

P. M.

Mayo

A. Y.

(2010). The home language environment of monolingual and bilingual children and their language proficiency. Applied Psycholinguistics, 31(01), 117–140.

42.

Schneider

Dubé

R. V.

Hayward

(2005). The Edmonton Narrative Norms Instrument. Retrieved from University of Alberta Faculty of Rehabilitation Medicine website: www.rehabresearch.ualberta.ca/enni.

43.

Schneider

Hayward

(2010). Who does what to whom: introduction of referents in children’s storytelling from pictures. Language, Speech and Hearing Services in Schools, 41, 459–473.

44.

Sorenson Duncan

. (2017). How Does Maternal Education Influence Language Acquisition? Interdependencies between Environment and Input in the Bilingual Development of Immigrant and Refugee Children. Unpublished Doctoral Dissertation, University of Alberta, Edmonton, AB, Canada.

45.

Wiechmann

Kerz

Steinfeld

(2016). The Effect of Dual Language Exposure on Single Language Development: Comparing Acquisition Rates in Bilingual and Monolingual Children. In Bannink

Honselaar

(Eds.), From variation to iconicity: Festschrift for Olga Fischer on the occasion of her 65th birthday (pp. 425–438). msterdam: Uitgeverij Pegasus.

46.

Unsworth

(2013). Assessing the role of current and cumulative exposure in simultaneous bilingual acquisition: The case of Dutch gender. Bilingualism: Language and Cognition, 16(1), 1–25.