Introduction
Latino children of immigrants, the largest group of dual language children in the U.S., are widely understudied and underserved (Gutiérrez, Zepeda & Castro, Reference Gutiérrez, Zepeda and Castro2010; Tienda & Haskins, Reference Tienda and Haskins2011; Toppelberg & Collins, Reference Toppelberg and Collins2010) and at risk for poor academic outcomes (Santiago, Gudiño, Baweja & Nadeem, Reference Santiago, Gudiño, Baweja and Nadeem2014). Empirical research addressing the early development and education of this population is imperative to increase our understanding of the individual and contextual forces shaping their academic success. Latino dual language children typically enter school with a wide range of proficiencies in Spanish and English and make gains in one or both languages during their first school years. The extant literature supports the benefits of developing and supporting the home language of children of immigrants, without detriment to their ability to learn oracy or literacy in the larger community language (Wilson, Dickinson & Rowe, Reference Wilson, Dickinson and Rowe2013). It is therefore important to understand factors that contribute to the variability in proficiency in each language of the dual language child during the early school years in order to better promote the development of both languages.
Individual variation in children's language skills is often noted from an early age, partly due to differences in experiences and language exposure during the early developmental years before school. Children's home environment, family background, and early childhood settings play a fundamental role in their language development. We have known for some time that socioeconomic status (SES) impacts language skills in children (González, Reference González2001), however there is a need for more research to better understand the variability of candidate contributors that make up SES given the complexity of this construct. Moreover, the distinct aspects of socioeconomic factors should be jointly considered along with sociocultural factors in order to better understand these influences on dual language development within an ecological framework (Chen, Geva & Schwartz, Reference Chen, Geva and Schwartz2012). The ecology that surrounds the child and the influence of different contextual systems (families, siblings, peers and schools) and the interactions across and within these systems must be considered especially during critical development shifts such as the transition from home to school (Garcia-Coll & Szalacha, Reference Garcia-Coll and Szalacha2004; Rimm-Kaufman & Pianta, Reference Rimm-Kaufman and Pianta2000; Rogoff, Reference Rogoff2003). In the following sections, we review the current knowledge base on factors within these settings that have been shown to influence language development among monolingual children, and their role in the dual language development of Latino children of immigrants. We then discuss the present study, which aims to expand this knowledge base of dual language development within an ecological framework.
Gender
Studies of language development have documented gender differences, with girls having on average higher language abilities than boys (Hammer, Farkas & Maczuga, Reference Hammer, Farkas and Maczuga2010; Reilly, Wake, Ukoumunne, Bavin, Prior, Cini, Conway, Eadie & Bretherton, Reference Reilly, Wake, Ukoumunne, Bavin, Prior, Cini, Conway, Eadie and Bretherton2010). This has also been found to be the case among dual language children; although, to date, most gender differences favoring girls have been evidenced mostly in English and little is known on how gender may affect the development of both English and Spanish in dual language children. In addition, further research is needed in order to understand when gender effects may begin to manifest developmentally; whether they emerge in the early school years, earlier, or later during adolescence. In a study of dual language children, girls showed higher levels of engagement with parents from an early age and rated higher in measures of language development and school readiness (Winsler, Burchinal, Tien, Peisner-Feinberg, Espinosa, Castro, LaForett, Kim & De Feyter, Reference Winsler, Burchinal, Tien, Peisner-Feinberg, Espinosa, Castro, LaForett, Kim and De Feyter2014). While Latino dual language children often do not continue to develop their Spanish after beginning school, it has been documented that girls are more likely to be dual language proficient in adolescence (Portes & Rivas, Reference Portes and Rivas2011). In Latino immigrant families, girls and boys may have different types of exposure to Spanish and English. In a four year longitudinal study starting in preschool, Puerto Rican mothers were found to be five times more likely to speak in Spanish to their daughters than to their sons (Hammer, Lawrence, Rodriguez, Davison & Miccio, Reference Hammer, Lawrence, Rodriguez, Davison and Miccio2011).
Birth order
In the general population, first born children tend to be more advanced in lexical and grammatical development than later born children (Hoff-Ginsberg, Reference Hoff-Ginsberg1998; Hoff, Reference Hoff2006; Havron, Ramus, Heude, Forhan, Cristia, Peyre & EDEN Mother-Child Cohort Study Group, Reference Havron, Ramus, Heude, Forhan, Cristia and Peyre2019). Among dual language children, first borns may have more exposure to the home language at a younger age and thus tend to develop and maintain their home language to a greater degree than their later born siblings who tend to use more English at home (Bridges & Hoff, Reference Bridges and Hoff2014). One reason for this is because, for a time, first born children grow up as an only child and, as a result, are likely to have more one-to-one parental interactions in their home language than their later born siblings (Hoff-Ginsberg, Reference Hoff-Ginsberg1998; Huttenlocher, Vasilyeva, Waterfall, Vevea & Hedges, Reference Huttenlocher, Vasilyeva, Waterfall, Vevea and Hedges2007; Havron et al., Reference Havron, Ramus, Heude, Forhan, Cristia and Peyre2019). In addition, older dual language siblings tend to use English more in talking to their younger siblings than other household members (Bridges & Hoff, Reference Bridges and Hoff2014), thereby increasing the amounts of English exposure and use at home for second-born and later children. Child directed language has been also shown to be different across families from different SES backgrounds, which must be considered when investigating the effects of birth order on dual language proficiencies in the early school years (Hoff-Ginsberg, Reference Hoff-Ginsberg1998).
Poverty and parental education
Studies have demonstrated SES-related differences in children's language experience, including the way parents talk to their children and the time spent on shared activities (Hoff-Ginsberg, 1991; Huttenlocher et al., Reference Huttenlocher, Vasilyeva, Waterfall, Vevea and Hedges2007). The home environment of families living in poverty may be less linguistically stimulating and responsive (Mistry, Benner, Biesanz, Clark & Howes, Reference Mistry, Benner, Biesanz, Clark and Howes2010) as a consequence of competing demands such as parents holding multiple jobs and other circumstances related to low SES including low parental education and single parenting (Hammer & Miccio, Reference Hammer and Miccio2006). As Latino children represent the largest segment of children living in poverty in the U.S. (Lichter, Sanders & Johnson, Reference Lichter, Sanders and Johnson2015) they are particularly exposed to poverty as a risk factor for language development. The variability in the language abilities evidenced in young dual language children may largely be attributed to SES (Guhn, Gadermann, Hertzman & Zumbo, Reference Guhn, Gadermann, Hertzman and Zumbo2010; Kieffer, Reference Kieffer2010; Mancilla-Martinez & Lesaux, Reference Mancilla-Martinez and Lesaux2011). However, the majority of studies showing SES effects on language development have been conducted with monolingual samples (Scheele, Leseman & Mayo, Reference Scheele, Leseman and Mayo2010). There is a specific need to understand the interplay between SES among other contextual factors across dual language children from different ethnic groups.
When considering the impact of poverty we must also take into account the effects of parent education on children's language development, which have been shown to contribute above and beyond the poverty threshold (Hoff, Reference Hoff2013). Therefore, it is particularly important to focus on the defining components of SES in immigrant families in addition to income, such as levels of education which may be significantly higher and disassociated from income levels due to immigration uprooting and low access to job opportunities, even among those living in poverty. Also, in studies involving Latino children, it is important to consider the higher involvement of fathers and higher marital and cohabiting rates among Latino families compared to other similar, low-SES minority families (Cabrera, Shannon, West & Brooks-Gunn, Reference Cabrera, Shannon, West and Brooks-Gunn2006). Paternal influences, such as paternal education and origin should be included in analyses, as “most Latino children live in households where fathers are accessible and share in their daily care” (Karberg & Cabrera, Reference Karberg and Cabrera2016). While there has been evidence that children from lower SES homes enter school with lower language abilities and, in the case of dual language children, lower English abilities, it is important to consider children's home language abilities (Farver, Xu, Eppe & Lonigan, Reference Farver, Xu, Eppe and Lonigan2006). There is limited research among dual language children on how different components of SES may be associated with both their home and school language proficiency.
Ethnic origin
Cultural background factors are predictive of development among all children; however, they bear particular significance on dual language children. While according to the U.S. Census (2010), Latinos (or “Hispanics”) share Spanish as a common or ancestry language, there are relevant immigration and cultural differences across origin groups that the “Latino” ethnic categorization fails to capture. A number of circumstances contribute to how Latino parents’ backgrounds are associated with the development of Spanish and English in their children. Transnational immigration patterns, diminished rooting and acceptance in this country may arguably contribute to stronger investment in maintaining the home culture and language (Dawson, Perez & Suárez-Orozco, Reference Dawson, Perez and Suárez-Orozco2012). Immigrants from the Dominican Republic demonstrate strong connections to their home culture and have been found to maintain Spanish language use into third-generation Dominicans (Alba, Logan, Lutz & Stults, Reference Alba, Logan, Lutz and Stults2002). In comparison, mainland Puerto Ricans have reported speaking significantly less Spanish at home compared to other Latino immigrant groups (Robert Wood Johnson Foundation, 2014; Alba et al., Reference Alba, Logan, Lutz and Stults2002), suggesting home language practices may be less protected among mainland Puerto Ricans, while ethnic culture and Spanish language allegiance is likely higher among foreign-born and non-U.S. citizen Latinos (Hammer, Nimmo, Cohen, Draheim & Johnson, Reference Hammer, Nimmo, Cohen, Draheim and Johnson2005; Hammer, Rodriguez, Lawrence & Miccio, Reference Hammer, Rodriguez, Lawrence and Miccio2007).
Language use at home
The language environment at home is in many ways dependent upon the family members’ own language backgrounds and use at home. In homes where both parents use the minority language, children have a higher chance of developing and maintaining this language (Portes & Rivas, Reference Portes and Rivas2011; De Houwer, Reference De Houwer2007; Collins, O'Connor, Suárez-Orozco, Nieto-Castañon & Toppelberg, Reference Collins, O'Connor, Suárez-Orozco, Nieto-Castañon and Toppelberg2014). For example, a child living in a family where both parents regularly use Spanish will hear substantially more conversations and interactions in Spanish than in families where one parent speaks Spanish but the other speaks English. Portes and Hao (Reference Portes and Hao1998) found that parental language input is highly associated with home language retention in adolescents from diverse immigrant backgrounds in the United States. Additionally, Cha and Goldenberg (Reference Cha and Goldenberg2015) documented that children from Spanish dominant homes were more likely to have higher outcomes in both Spanish and English, whereas English dominant homes led to better oral proficiency in only English at the expense of loss of Spanish proficiency. As such, the role of parent language use must be considered concurrently with other sociocultural and socioeconomic factors associated with children's Spanish and English proficiency in the early school years.
Home literacy
Literacy practices at home play a critical role in children's dual language development and proficiencies as they transition into kindergarten and early school years. Parental literacy habits toward reading influence the child's exposure to print and consequently their language development (Weigel, Martin & Bennett, Reference Weigel, Martin and Bennett2006). For dual language children, literacy experiences such as joint book reading, storytelling, and the number of books at home affect their development in both languages (Caspe, Reference Caspe2009). Goldenberg, Reese and Rezaei (Reference Goldenberg, Reese and Rezaei2011) found that home language use and literacy activities in English tended to predict higher English language and literacy scores and lower Spanish scores while home language and literacy in Spanish predicted higher Spanish and lower English scores. However, a recent study showed home language and literacy experiences of Latino dual language children had a large effect on children's Spanish abilities with minimal effect on their English (Lewis, Sandilos, Hammer, Sawyer & Méndez, Reference Lewis, Sandilos, Hammer, Sawyer and Méndez2016).
Access to children's books during their early years has been shown to positively influence literacy outcomes, yet in many lower SES homes there are limited amount of books or even no books at all (Bracken & Fischel, Reference Bracken and Fischel2008; Neuman, Reference Neuman1999). Latino immigrant families have fewer books at home, and limited access to books in both Spanish and English through public libraries or commercial outlets (Ramírez, Yuen & Ramey, Reference Ramírez, Yuen and Ramey1991). Spanish children's books, in particular, are limited and constitute a very small percentage of the total books available in the U.S. (Reese, Thompson & Goldenberg, Reference Reese, Thompson and Goldenberg2008). Therefore, homes that have books in Spanish tend to reflect a proactive effort from caregivers.
In addition to the quantity of books, the type of literacy experiences is also related to children's language outcomes (Cline & Edwards, Reference Cline and Edwards2017). What is defined as book reading may differ between families as cultural, economic and gender differences often play a role in how a parent reads to their child and how the child perceives the act of reading (Caspe, Reference Caspe2009; Cline & Edwards, Reference Cline and Edwards2017; Downing, Ollila & Oliver, Reference Downing, Ollila and Oliver1975; Wollscheid, Reference Wollscheid2014). For example, Cline and Edwards (Reference Cline and Edwards2013) found that high extra-textual talk was positively related to children's expressive language skills in English speaking homes whereas low extra-textual talk was more positively associated in Spanish speaking homes. Given the additional factors related to literacy within multilingual homes it is essential to investigate language differences in relation to accessibility of books and reading habits, and their effect on dual language development.
Language use in early childhood settings
The early childhood education environment is more than an extension of the home and is critical in supporting children's language development (Dearing, Zachrisson, Mykletun & Toppelberg, Reference Dearing, Zachrisson, Mykletun and Toppelberg2017). For Latino dual language children, early education provided in English only settings may lead to home language loss; the younger the children are, the greater the effect. Dual language children who attend early childhood education settings where English is primarily used are subsequently less likely to speak their home language than children who attend early childhood education settings where the home language is used (Buysse, Peisner-Feinberg, Páez, Hammer & Knowles, Reference Buysse, Peisner-Feinberg, Páez, Hammer and Knowles2014). Bilingual early childhood education settings where the home language is used as part of instruction do not limit progress in English and have the advantage of supporting the growth and development of the home language. Contrary to common misunderstandings, teaching children to read in their home language can support their literacy development in English (National Academies of Sciences, Engineering, and Medicine, 2017). In a longitudinal study of Spanish-speaking low-income preschool children who attended high quality bilingual schools, significant gains in both languages were made each year, greater than those for the control group who attended preschools with English only programs (Barnett, Yarosz, Thomas, Jung & Blanco, Reference Barnett, Yarosz, Thomas, Jung and Blanco2007). Furthermore, early childhood environments that use the children's home language allow for more opportunities for children to participate and express themselves. In a study of pre-kindergartner classrooms, teachers who spoke some Spanish to dual language children led to better teacher-student relationships which in turn led to higher levels of children's Spanish abilities (Chang, Crawford, Early, Bryant, Howes, Burchinal, Barbarin, Clifford & Pianta, Reference Chang, Crawford, Early, Bryant, Howes, Burchinal, Barbarin, Clifford and Pianta2007).
As language skills develop over time, it is especially important to consider that the contexts and experiences of dual language children may differ in each language. Research has recognized certain types of language support, practices, and exposures that have been shown to have an on-going positive effect on developing proficiencies. Although, many studies conducted to date among this population have focused primarily on children's English language proficiencies. It is critical to consider both languages of dual language children, as the contextual factors may have divergent effects for each language when they begin school, as well as over time.
Present study
The present study investigates Spanish and English proficiencies of Latino dual language children across the critical early school years in relation to variations in socioeconomic and sociocultural factors. The focus of this study is motivated by research showing a wide variability in dual language proficiencies at school entry (Collins et al., Reference Collins, O'Connor, Suárez-Orozco, Nieto-Castañon and Toppelberg2014). The potential benefits of dual language proficiency among this at-risk population underscores the need for additional research in this area. Our study investigates oral language proficiencies in Spanish and English critical to function in academic settings (August, Reference August2003; Gottlieb & Corwin, Reference Gottlieb2006) and is guided by the following research questions:
1. What are the individual effects of child characteristics, poverty, parent education and origin, home language use and literacy practices, and early childcare language use on children's English and Spanish proficiencies at kindergarten and second grade?
2. When simultaneously considered, what are the unique and shared effects of these variables on children's English and Spanish proficiencies at kindergarten and second grade?
Methods
The present study is based on a longitudinal sample of 228 Spanish speaking Latino children attending urban public schools in a large city in the Northeast United States. The variables analyzed were collected as part of the Childhood Bilingualism and Developmental Psychopathology study funded by the National Institutes of Health addressing multiple aims encompassing dual language development. All children were screened to meet the following inclusion criteria. All children were born in the continental U.S. with at least one parent who was born in Puerto Rico, Dominican Republic or other Latin American country and currently residing in the mainland US. Puerto Rican families were included as they share many characteristics of other Latino immigrant families and account for a substantial proportion of this population in the Northeast U.S.. Children must have been living in a predominantly Spanish-speaking home with minimal or no exposure to English before age three. Children with developmental disorders or other severe sensory-motor sequelae of neurological trauma or disorder were excluded. After receiving IRB, school district and principal approval, student information from school enrollment lists was used to determine potential eligibility. Potential participants and their caregivers were invited to participate with a recruitment letter explaining the study. Bilingual and bicultural research assistants made phone calls to caregivers to confirm eligibility criteria and willingness to participate in the study. Upon obtaining written parental consent and child assent, children were assessed in three sessions, 45 minutes each, conducted at the school on three separate days, in most cases within a week. Children were individually assessed in kindergarten (mean age 6;1) and second grade (mean age 8;1) for language abilities, in offices outside of the classroom on separate days, counterbalanced so that half of the children were tested in English first and the other half were tested in Spanish first. Protocols from the published assessment manuals were followed by trained research assistants who were native speakers of Spanish and English. Trained bilingual research assistants conducted parent interviews in the children's homes and collected information on family background and history. These parent interviews took place mostly with children's mothers who chose in most cases to conduct the interview in Spanish.
Our sample included equal numbers of boys and girls, with the majority being first-born children (52%). Most participants were living in poverty (86%) and had parents with low to moderate levels of education. The majority of mothers were born in the Dominican Republic (52%) and Puerto Rico (21%) and used only Spanish (55%) or mostly Spanish (24%) when speaking to the child. Almost all children (94%) had at least some books at home and had some sort of early childhood care outside the home before kindergarten (75%). Children attended early childhood settings including Head Start (45%), preschools (18%), home-based daycares (13%), or did not have care outside the home prior to kindergarten (25%). Additional demographic and descriptive data of sample subgroups are presented in Table 1.
Descriptive Statistics for Study Variables and Analysis of Covariance (ANCOVAs) of Variable Subgroups
Note. The table consists of imputed data with a total N of 228. Superscripts A, B, C, D, E and F differentiate groups for ANCOVA Bonferroni post-hoc analyses.
Measures
Dual Language Proficiency. Spanish and English oral language abilities were directly assessed using the Woodcock Language Proficiency Battery-Revised (WLPB-R) at kindergarten and again two years later (Woodcock, Reference Woodcock1991). The WLPB-R measures academic language proficiency and it has been widely used by empirical studies in educational and clinical settings. It has published validity, reliability, and norms for ages 2;0 to 90+ in English and Spanish. Cronbach's alpha reliability (internal consistency) coefficients for both languages range from .84 to .92 for each of the subtests and test clusters across all age ranges. Specific linguistic domains in English and Spanish were tested during separate sessions by native speakers of the corresponding language. Four oral WLPB test scores (Memory for Sentences; Picture Vocabulary; Listening Comprehension, and Verbal Analogies) in each language were used to yield an oral language cluster score, a global measure of general language proficiency normed by WLPB-R procedures, for each language. Memory for Sentences is a mixed expressive-receptive measure of syntactic and semantic competence, where the child is asked to repeat words, phrases and then whole sentences of increasing length. Picture Vocabulary is a measure of expressive semantics (vocabulary) that requires naming pictures of items. Listening Comprehension is a receptive measure of syntactic and semantic competence, where the child listens to increasingly complex sentences or passages and is asked to provide the word that is missing at the end of the sentence or the passage. Verbal Analogies is a receptive measure of semantics, where the child is asked to provide a brief verbal answer to a missing item in a sentence that carries logical relationship.
Child Characteristics, Poverty, Parent Education and Origin, Home Language Use and Literacy Practices, and Early Childcare Language Use data was collected when the children were in kindergarten, through multidimensional surveys administered during the parent interviews at home. Questions adapted from the Methods for the Epidemiology of Child and Adolescent Mental Disorders (MECA) study (Goodman, Hoven, Narrow, Cohen, Fielding, Alegria, Leaf, Kandel, Horwitz, Bravo, Moore & Dulcan, Reference Goodman, Hoven, Narrow, Cohen, Fielding, Alegria, Leaf, Kandel, Horwitz, Bravo, Moore and Dulcan1998) were used to determine the primary caregiver, parents’ place of birth (origin), socioeconomic status (poverty, and maternal/paternal education) child's birth order and child's gender. Children's birth order was determined based on parental reports of all siblings of participants. A child was considered to be living in poverty when his or her household was a recipient of at least one government program which ties eligibility to income threshold levels (U. S. Bureau of the Census, 2000); these programs included: WIC, Food Stamps, Head Start, Section 8 housing assistance, and Welfare. The highest level of schooling the parents completed (elementary only, partial high school, complete high school, etc.) was recorded using ordinal scores ranging from 0 (no formal education) to 12 (MA/PhD) reflecting both education level attained and years of education; in addition, the total number of years of schooling completed by the mothers and fathers was recorded. Language use at home for each caregiver and whether each spoke mostly or only Spanish, Spanish and English, or mostly English to the child were measured at baseline with the Language and Literacy Use (LLU; Tabors & Paez, Reference Tabors and Paez2001) parent survey. The majority of the home interviews (99%) were conducted with children's mothers who were determined to be the primary caregivers. Maternal language proficiency was directly assessed in both Spanish and English at their homes by a research assistant using WLPB-R listening comprehension subtests previously described. Information was collected about home literacy practices and included questions on the number of books and languages of the books available at home, frequency of parental reading, as well as the frequency and languages of an adult or older sibling reading to the child. Parents were also asked about whether children attended preschool, day care, head start or other childcare setting outside the home, and about the language(s) used in the early childhood education settings.
Non-verbal IQ is used as a control variable in all of the analyses for the present study. Children's non-verbal cognitive ability and general intelligence was measured with the Universal Nonverbal Intelligence Test (UNIT; Bracken & McCallum, Reference Bracken and McCallum1998) which is administered and completed without the use of language. The UNIT has been widely used in the cognitive assessment of children whose test performance may be affected by language impairment or second-language barriers. Extensive published studies have determined the UNIT's reliability and validity to be consistently high. The full-scale intelligence quotient standard score is used in the present analyses.
Results
Preliminary analyses of all study variables were conducted with measures of central tendency and variability and were determined to be suitable for further analyses. Regarding missingness, analyses of each variable were conducted to check whether missing status was associated with other study variables. For all of the independent variables collected in kindergarten, coverage was higher than 99% (with the exception of paternal education, which was 82%). Children's language proficiency variables collected in second grade coverage was 72%, mostly due to a 20% rate of attrition. Children with missing and complete data were compared using t-test analyses and did not differ significantly on any of the study variables. It was concluded, missing data was completely at random, with no systematic relationship between the propensity of missing values and the observed data. With the goal of increasing the power of the analyses and the quality of the estimates, for each variable with missing data, we conducted multiple imputations including auxiliary variables, per expert recommendations in this area (Enders, Reference Enders2013; Graham, Reference Graham2009).
Descriptive findings (Table 1) include group sizes (N), unadjusted means (M), and standard deviations (SD) for each of the independent variable subgroups including demographic, language and literacy use at home, and early childhood language use. Children's non-verbal IQs were in the average range for the general population (M= 96, SD= 13). The mean score for the language proficiencies at kindergarten for both Spanish (M = 67, SD = 21) and English (M = 70, SD = 18) were both well below the normed mean standard score (SS) of 100. Below-average performance for dual language children on measures of oral academic language proficiency is typical when using standardized measures with monolingual norms and has been reported in studies of similar populations using the same measures (Oller & Eilers, Reference Oller and Eilers2002; Páez, Tabors & López, Reference Páez, Tabors and López2007). At second grade, mean scores were significantly higher and closer to the normed average performance in both English (M = 85, SD = 17), and Spanish (M = 72, SD = 23).
In regards to research question 1 on the individual effect of study variables, we first looked at child characteristics (gender and birth order). Girls had significantly higher Spanish proficiency than boys at kindergarten (F[1,228] = 7.016, p < .001) and second grade F[1,228] = 6.227, p < .05). No gender differences in English proficiencies were evidenced in either kindergarten or second grade. In regards to birth order, no significant differences in English or Spanish proficiency were demonstrated across children who were first-born, second-born, or third-born and beyond.
Next, we analyzed variables related to socioeconomic status. Poverty and maternal and paternal education levels were associated with differences in children's English, but not Spanish, proficiency in both kindergarten and second grade. Children not living in poverty had English proficiency scores over 10 SS higher than those living in poverty at second grade (F[1,228] = 7.126, p < .01). Children of mothers who had more education had higher levels of English proficiency at kindergarten (F[2,228] = 11.092, p < .001) and second grade (F[2,228] = 20.256, p < .001); similar patterns were found for father's level of education on children's English proficiency at kindergarten (F[2,186] = 4.758, p < .01) and second grade (F[2,186] = 9.985, p < .001).
Parents’ origin and the language use with their child were significantly associated with children's Spanish proficiency. Children with mothers from Dominican Republic and other Latin American countries had the highest levels of Spanish proficiency at kindergarten (F[3,228] = 12.638, p < .001) and second grade (F[3,228] = 9.986, p < .001), significantly higher than children with mothers from Puerto Rico or those who were born in the continental U.S. The same patterns were evidenced in relation to father's place of birth. Children whose mothers who spoke to them in mostly or only Spanish had significantly higher levels of Spanish proficiency at kindergarten (F[2,228] = 10.582, p < .000) and second grade (F[2,228] = 9.187, p < .000) than mothers who spoke both Spanish and English or mostly/only English to their child; which was also evidenced among fathers who spoke only Spanish with their child. Spanish proficiency at kindergarten (F[2,228] = 1.459, p < .05) and second grade (F[2,228] = 1.420, p < .05) were significantly higher for these children. There were no significant differences in children's English at kindergarten and second grade across families who spoke more or less English in the home.
Home literacy practices including the number of books at home and the languages parents read to their children were related to children's language abilities. The number of books at home had a significant effect on children's English proficiency at kindergarten (F[3,228] = 6.508, p < .001) and second grade (F[3,228] = 5.191, p < .01). Children with few (<10) to no books at home had significantly lower English proficiency at kindergarten. The number of books at home did not have a significant effect on children's Spanish proficiency. The frequency and language that children were read to in the home was related to their English and Spanish proficiency. Reading to the child every day in Spanish was significantly associated with higher Spanish at both kindergarten (F[3,228] = 3.921, p < .01) and second grade (F[3,228] = 10.917, p < .001). Reading to the child every day in English was also significantly associated with higher English at kindergarten (F[3,228] = 3.237, p < .05).
Lastly, we investigated the effect of language use in early childhood settings outside the home. Children who attended an early childhood education setting where only or mostly Spanish was used had significantly higher Spanish proficiency at both kindergarten (F[5,228] = 4.456, p < .001) and second grade (F[5,228] = 4.246, p < .001) compared to settings where some or only English was used as well as for children who did not have early childhood education. English proficiency at kindergarten and second grade were not different when a child had attended an early childhood education setting in which Spanish was used (even when it was the only language used).
In regards to research question 2, we first investigated the association of the predictor variables and outcome variables with Pearson bivariate correlations. Socioeconomic predictors (poverty and paternal education) were significantly associated with children's English proficiency but for the most part not with children's Spanish proficiency. Whereas, sociocultural predictors (parents’ place of birth, home language use) were significantly associated with children's Spanish proficiency but for the most part not with children's English proficiency (see Table 2).
Pearson Bivariate Correlations of Study Variables.
Note. * indicates p < 0.05, ** indicates p < 0.01
Next, we used hierarchical multiple regression to test the shared and unique contributions of the study variables to children's English and Spanish proficiencies at kindergarten and second grade. We regressed each of the study variables in separate models for each of the outcome of Spanish and English proficiency at kindergarten and second grade. Results of the final models are presented in Table 3. Predictors were entered in seven separate blocks controlling for previously entered variables on each of the outcomes. After controlling for non-verbal IQ, gender and birth order were entered in the second block of the model. Gender significantly predicted children's Spanish proficiency at kindergarten (ß= -.139 p. <.05; ΔR2=.03). Next socioeconomic status variables were entered: father's education was significant (ß= .149 p. <.05) but only after controlling for parent origin and home language use. Children whose mothers were born in the Dominican Republic or other Spanish speaking country had significantly higher Spanish at kindergarten (ß= .326 p. <.01; ΔR2=.13) than children with mothers born in the continental U.S. or Puerto Rico. Home language use was entered in the fourth block of variables and included mother to child language use, father to child language use, maternal Spanish proficiency, and maternal English proficiency. Controlling for each of these variables, mother to child language use significantly predicted children's Spanish at kindergarten (ß= .255 p. <.01; ΔR2=.11). Home literacy practices were entered in the next block but did not significantly contribute to children's Spanish at kindergarten. Lastly, preschool language was entered and significantly predicted children's Spanish at kindergarten (ß= .201 p. <.01; ΔR2=.04). Regressions results predicting children's Spanish proficiency at second grade were closely aligned with predictors of children's Spanish proficiency at kindergarten. In addition to the significant predictors at kindergarten, birth order significantly predicted children's Spanish at second grade (ß= -.132 p. <.05; ΔR2=.04). Also, father to child language use (ß= .166 p. <.01; ΔR2=.11) and reading to children in Spanish had a significant effect at second grade (ß= .179 p. <.01; ΔR2=.03) that was not present in kindergarten.
Hierarchical Multiple Regressions of Study Predictors of English and Spanish Proficiencies at Kindergarten and Second Grade
Note. * indicates p < 0.05, ** indicates p < 0.01
Next predictors were regressed on children's English proficiency at kindergarten. After controlling for non-verbal IQ, mother's education (ß= .177 p. <.01) significantly accounted for 5% of the variance of children's English proficiency. Home literacy practices were significantly associated with children's English proficiency (ß= .205 p. <.01; ΔR2=.03). Regressions results predicting children's English proficiency at second grade were aligned with predictors of children's English proficiency at kindergarten. In addition to the significant predictors at kindergarten, birth order (ß= -.150 p. <.05; R2=.03), poverty (ß= -.140 p. <.05; R2=.10) and maternal education (ß= .236 p. <.05; R2=.10) had a consistent significant effect. Home literacy practices were significantly associated with children's English proficiency (ß= .161 p. <.05; ΔR2=.02) at second grade.
The study variables accounted for moderate amounts of shared variance in each of the models, with 37% of the variation of Spanish proficiency at kindergarten and 42% at second grade and 35% of the variation of English proficiency at kindergarten and 46% of English proficiency at second grade significantly predicted. Overall, Spanish was significantly predicted by gender, birth order, mother's origin, mother and father's language use, the frequency of reading to child in Spanish at home, and the language use in early childhood settings. Whereas, English was predicted by other variables, including: birth order, poverty, mother's education, and home literacy practices. Simultaneously controlling for the effect of the multivariate predictors in the regression models demonstrated a consistent effect of certain predictors; although, when controlling for the other variables in the model, paternal education was also related to children's Spanish proficiency. After controlling for the role of mothers’ origin and home language use the effect of fathers’ origin was no longer a significant predictor of children's Spanish proficiency at kindergarten and second grade as seen in the ANCOVA models. The unique role and influence of maternal and paternal backgrounds suggests that both mother's place of birth and father's education influence their children's dual language development in different ways.
Discussion
In the present study, family background and language use at home and early childhood settings accounted for substantial variance in children's Spanish and English language proficiency at kindergarten and second grade. However, these associations tended to follow a surprising pattern: while some of these predictors were associated with English proficiency, others were associated with Spanish proficiency, with little to no overlap. Overall, socioeconomic predictors (poverty, parental education, and home literacy) were associated with children's English proficiency. On the other hand, gender, parent origin, and use of the Spanish language at home and early childhood education settings were associated with levels of children's Spanish proficiency. It is important to try to understand in which way, and potentially why, these differences in environments and experiences relate to children's dual language development. Moreover, there is need to understand how these early predictors from home continue to relate to children's dual language development once they are in school, with potentially critical implications for their academic achievement.
Parents in our sample and their socioeconomic status for the most part were representative of the makeup of Latino immigrant groups in the Northeast, and included a large proportion of Dominicans, many living in poverty (U. S. Bureau of the Census, 2000). Children in our study who were living in poverty had lower levels of English proficiency at second grade. Living in poverty could have limited children's exposure to learning experiences and opportunities in English. For those children who are not living in poverty it is understandable they would have higher English abilities; however, it is not clear why they do not also have higher Spanish proficiency, which would follow the pattern seen among monolingual children (Mistry et al., Reference Mistry, Benner, Biesanz, Clark and Howes2010; Spencer, Clegg & Stackhouse, Reference Spencer, Clegg and Stackhouse2012). Our findings also demonstrated mother's level of education as a significant predictor of children's English proficiency at kindergarten and second grade. Immigrant parents with higher levels of education have been associated with higher levels of children's English language skills by kindergarten (Bohman, Bedore, Peña, Mendez-Perez & Gillam, Reference Bohman, Bedore, Peña, Mendez-Perez and Gillam2010). Other research has demonstrated that maternal level of education is especially important for children of poor, young, or single-mothers, and is associated with young children's language skills and the home language environment (Magnuson, Sexton, Davis-Kean & Huston, Reference Magnuson, Sexton, Davis-Kean and Huston2009). Interestingly, after controlling for the effect of maternal education, our findings demonstrated that children with more educated fathers have higher levels of Spanish, strongly suggesting a unique contribution of fathers to children's dual language development. Overall, parents’ levels of education accounted for the substantial amount of variance in English proficiency at kindergarten and second grade. A potential explanation is that living in poverty offers these children fewer resources and less exposure, making the people around them all the more valuable in terms of educating them, and as such parent's level of education becomes more important. Moreover, families with lower levels of education may have fewer resources and less exposure to English, as parents’ level of education is often connected to different career opportunities that often require higher levels of English proficiency and necessity to interact in English.
Gender differences were found for Spanish but not for English. Girls had 0.5 SD higher standard scores in Spanish proficiency than boys, but there were no gender differences in English proficiency. Studies of monolingual children have shown that girls tend to develop stronger language abilities before boys (Locke, 2002), which is reflected in our study participants’ proficiency in their first language, Spanish. Our findings demonstrate gender differences in children's Spanish language proficiency that may stem from early capacities and home language influences. Furthermore, our results document that these gender differences happen very early on, in kindergarten (if not earlier), and suggest that they may deepen as time goes by. As mentioned, in contrast, there were no gender differences in English at either kindergarten or second grade. It is possible that, in learning the school language, boys and girls have similar learning opportunities in terms of language exposure and instruction. On the other hand, if girls’ stronger language learning capacities extend to learning a second language, achieving only equal ability may result from having fewer learning opportunities than boys at school and other settings. While gender differences among dual language learners have been reported in adolescents (Lutz, Reference Lutz2006; Portes & Rivas, Reference Portes and Rivas2011), our study is the first to document, to our knowledge, this phenomenon in younger dual language children. Furthermore, our study uniquely focuses on a mostly Latino-Caribbean (as opposed to Mexican or Central American) sample using more precise direct child psychometric assessments (versus self-reports) of language proficiency.
Children who had a parent born in Puerto Rico or the continental US had lower Spanish proficiency at kindergarten and second grade compared to children with parents, particularly mothers from the Dominican Republic or other Spanish speaking countries. Puerto Rican households tend to be more mixed in terms of generation of immigration with back and forth travel and frequent stays between countries (Durand & Massey, Reference Durand and Massey2010) and may use more English in the home. Furthermore, English is more widely spoken in Puerto Rico compared to other countries included in our sample. Children of Puerto Rican origin more often come from homes where less Spanish is spoken compared to those of Dominican Republic, Mexican or other Latin American descent (Arriagada, Reference Arriagada2005; Alba et al., Reference Alba, Logan, Lutz and Stults2002). Children of mothers born in the Dominican Republic had higher levels of Spanish at kindergarten and in second grade. It is important to note that variations in language development across ethnic and cultural groups are sometimes reduced when controlling for socioeconomic status, language use, and proficiency of primary caregivers (Portes & Hao, Reference Portes and Hao1998). In our study, after controlling for these other variables, maternal origin had a persistent impact on child Spanish language proficiency. This finding does align with other studies where language differences have persisted even with these controls (Arriagada, Reference Arriagada2005; Cabrera et al., Reference Cabrera, Shannon, West and Brooks-Gunn2006).
In terms of language used in the home among mothers and fathers, children's Spanish abilities were associated with parents speaking to them in only or mostly Spanish, but children's English abilities were not associated with parents’ speaking to them in English (or Spanish) at home. As other studies have demonstrated, the amount of exposure and input of both the home language and the community language influences children's language preference, dominance, ability and usage (Bohman et al., Reference Bohman, Bedore, Peña, Mendez-Perez and Gillam2010; Genesee, Paradis & Crago, Reference Genesee, Paradis and Crago2004). In our study, parents’ use of Spanish at home was associated with higher Spanish, but demonstrated no loss of English, an important finding consistent with the literature that home language use is not detrimental to their development of English skills (National Academies of Sciences, 2017; Winsler, Diaz, Espinosa & Rodriguez, Reference Winsler, Diaz, Espinosa and Rodriguez1999). Children whose parents more often spoke to them in Spanish had higher Spanish proficiency levels at both kindergarten and second grade. In these homes, children's exposure to conversations, interactions, and dialogues in Spanish is more likely which in turn results in children having a higher Spanish proficiency. Language differences in home environment in relation to input quantity and quality have been demonstrated to collectively predict children's later language and academic skills (Dickinson & Tabors, Reference Dickinson and Tabors2001; Payne, Whitehurst & Angell, Reference Payne, Whitehurst and Angell1994; Weigel et al., Reference Weigel, Martin and Bennett2006).
Exposure to print in the home, including the number of books, frequency of parents reading, and the language which parents read to their children, was also related to children's dual language abilities. Overall, children from homes with more frequent home literacy practices had higher levels of English. Furthermore, having books in both Spanish and English had a positive effect on children's English. Children from homes with more than 25 books had higher English abilities at kindergarten and second grade. Overall, families had fewer books in Spanish than in English. As previously discussed, Latino immigrant families in general have limited access to books, and this is particularly so for books in Spanish for young children (Reese et al., Reference Reese, Thompson and Goldenberg2008).
Of interest, the more fine-grained analyses of literacy practices considering the frequency of parents’ reading, and whether children were read to in Spanish or English, showed that frequent reading in Spanish was related to children's Spanish language proficiencies. However it is important to note that reports of reading practices may be a better indicator of literacy practices at home than the number of available books and may have a more proximal impact on children's dual language proficiencies. Reading in Spanish to children involves a deliberate and intentional strategy that demands concerted efforts from the parents. The language in which adults and older siblings read to children was associated with later proficiency, especially in Spanish. Parents’ literacy practices also had a large effect on children's English abilities which is consistent with previous studies (Lewis et al., Reference Lewis, Sandilos, Hammer, Sawyer and Méndez2016).
Having attended early childhood settings in which Spanish was used strongly predicted children's Spanish language abilities at kindergarten and second grade, compared to not having attended an early childhood program or attended one where only English was used. Early childhood experiences seem to play a role in further developing children's home language during the early school years. Attending Spanish speaking early childhood settings was not associated with children's English proficiency, replicating in this population the well-documented finding that maintaining and supporting the home language in preschool settings is not detrimental to learning the new language (National Academies of Sciences, 2017).
Finally, in our study, differences in either Spanish or English proficiencies among first-born, second-born, or children who are third-born and beyond were only found after controlling for other study variables. When controlling for other variables in our study, first-born children demonstrated higher levels of Spanish at kindergarten and second grade and higher levels of English at second grade which differs from other studies (Bridges & Hoff, Reference Bridges and Hoff2014). Although in our study we compared language abilities across children with different birth-order positions from separate families: that is, comparisons across (and not within) families. As first-born children in our analyses include “only child” families which may obscure differences between children with siblings. Our findings may also suggest that the effect of birth order on English may not be fully evident at school entry (kindergarten) and differences become more significant over time.
Limitations
The present study was a community-based study which did not permit for an experimental design and would benefit from a larger sample of children particularly for the subgroups analyses. In addition to a larger sample size, additional data points would have allowed us to examine long-term trajectories of the effect of home language variables over time. Given the narrow developmental window (K-2) of outcomes considered in our study, the effect of some variables may not be as noticeable until later years. Additional time points would also be amenable to more robust analytical strategies such as growth-curve modeling, possibly providing confirmation of specific language profile trajectories. While most studies have focused on immigrant adolescents and adults, research on younger children is lacking, making it difficult to piece together developmental trajectories. More research is needed to better understand relationships among the early language experiences of dual language Latino children and how these experiences relate to later dual language abilities and school performance. Further studies are needed to investigate how specific linguistic domains of oral language and literacy develop among dual language children from diverse backgrounds. Furthermore, additional measures – related to acculturation, children's own perceptions, and gender roles – could further support our interpretations of the findings. The current study analyzes the association between early language experiences and environments in relation to children's dual language proficiencies over time. While the present study is strengthened with its use of direct assessment of mother and children's Spanish and English proficiency, home language use reported by mothers based on direct observations would strengthen the study. Additionally, it would be important to replicate this study in other ethnic groups of children of immigrants, as well as in other geographic areas. Nonetheless, there is a great need for data on the home and school language environments of young dual language children whose first significant contact with English occurs at kindergarten during this important developmental transition, which the present study provides.
Conclusions
This study directly responds to calls to contribute to greater understanding to our increasingly diverse student populations in the U.S. and around the globe. A number of scholars (Garcia-Coll, Lamberty, Jenkins, McAdoo, Crnic, Wasik & Vazquez, Reference Garcia-Coll, Lamberty, Jenkins, McAdoo, Crnic, Wasik and Vazquez1996; González, Reference González2001; National Academies of Sciences, 2017) have urged the field of education to consider multiple factors and processes that affect outcomes among linguistically and culturally diverse students. The findings from the current study provide evidence for the differential effect of socioeconomic and sociocultural predictors prior to beginning school on respectively English and Spanish language proficiencies. Our findings demonstrate that learning and maintaining the home language is not detrimental to learning English. Children's Spanish proficiency was associated with parents’ country of origin, gender, and use of Spanish at home, during literacy practices, and in early childhood settings which may act as sociocultural indicators of children's Spanish skills. In contrast, indicators of socioeconomic status such as absence of poverty, higher parental education and home literacy practices are largely predictive of higher English skills. These findings demonstrate that many variables that support the development of each language do not conflict with each other and it is fully possible to acquire dual language proficiencies.
The present study has the potential to inform broad educational practices and support interventions that target dual language children of immigrants. Government agencies and organizations, including professional associations whose members work directly with young dual language children, should also promote practices in families and programs that support the development of children's bilingualism. In addition, these agencies should conduct public informational campaigns to provide information about the capacity of infants, toddlers, and preschoolers to learn more than one language. These campaigns should include information on the communicative, social, cognitive, emotional, and employment advantages of bilingualism and the absence of evidence of harmful effects. Lastly, educators can use these findings to help identify dual language learning opportunities and intervention efforts. An increase in dual language proficiencies of young Latino children of immigrants at school entry may result in higher school readiness and achievement. Understanding dual language development of Latino children of immigrants is an essential first step in implementing appropriate services for children who, although growing rapidly in numbers, continue to be widely underserved.
Acknowledgements
This study was supported by the National Institute of Mental Health (Scientist Development –K01—grant K01MH01947 to C. Toppelberg, PI). The authors would like to thank the children, families, teachers, and schools that participated in the study, as well as Jiali Xu for her helpful comments and edits in a prior version of this manuscript.
