The classroom language context and English and Spanish vocabulary development among dual language learners attending Head Start

Highlights

• Examined associations between the classroom language context and vocabulary skill.

• Analyzed the language of instruction and the proportion of dual language learners.

• Bilingual instruction was the same for English vocabulary skill as English only.

• Mostly Spanish instruction related to higher Spanish, lower English vocabulary skill.

• Higher proportion of DLLs related to lower English but unrelated to Spanish scores.

Abstract

Using a nationally representative sample of dual language learners (DLLs) attending Head Start, this study investigated how the language used for instruction and the proportion of DLLs in the class was associated with English and Spanish receptive vocabulary development between the fall and spring (n = 531). Based on teacher report of the language or languages used for instructional activities in the classroom, teachers were categorized as using (1) English only, (2) a mix of English and Spanish, or (3) mostly Spanish. Three-level hierarchical linear models showed that children in classrooms using a mix of English and Spanish had English vocabulary scores that were no different than children in English-only classrooms. Children in mostly Spanish classrooms, however, had significantly lower spring English scores than children in English-only classrooms. In addition, children in English-only classrooms had significantly lower Spanish vocabulary scores than children in the other two categories of classrooms, which did not differ from each other. The higher the proportion of DLLs in a class the lower were spring English scores, but not Spanish vocabulary scores. Findings suggest that using bilingual instruction, and sharing classrooms with English-dominant peers can promote English vocabulary development without a cost to Spanish vocabulary development.

Introduction

Over the past few decades, schools in the United States have increasingly become home to a large and diverse population of children whose first language is not English. In 2013, 4.5 million language minority children were enrolled in schools in the United States. Most of these children speak Spanish at home, and are enrolled in the early elementary grades (National Center for Education Statistics [NCES], 2016). An area of debate in research and policy has concerned the best means of supporting achievement for young Dual Language Learners (DLLs), who are still developing their first language as they are learning English (Goldenberg, Nemeth, Hicks, Zepeda, & Cardona, 2012).

DLLs are at higher risk for long-term difficulties with language and literacy proficiency, as well as lower academic attainment, in part because they are likely to enter kindergarten having never been exposed to formal English vocabulary (Gándara, Rumberger, Maxwell-Jolly, & Callahan, 2003; Kieffer, 2012, NCES, 2003). Research has shown, however, that attending preschool may particularly benefit this at-risk group of students (Buysse, Peisner-Feinberg, Páez, Hammer, & Knowles, 2013; Gormley, 2008; Magnuson, Lahaie, & Waldfogel, 2006). The purpose of this paper is to examine the association between DLLs’ vocabulary development and one aspect of the preschool experience–the classroom language context.

DLLs are a diverse group with wide variation in familial country of origin, socioeconomic status (SES) and language proficiency (Calderón, Slavin, & Sánchez, 2011; Halle, Hair, Wandner, McNamara, & Chien, 2012). Despite this variability, DLLs are more likely to come from families that are lower-income with few years of formal education, and with limited access to high-quality educational resources (Calderón et al., 2011; Gándara et al.,2003). Since there is robust evidence of class-based differences in vocabulary knowledge between high- and low-SES children (Farkas & Beron, 2004; Fernald, Marchman, & Weisleder, 2013; Hoff, 2013), low-income DLLs may face particular difficulties with developing oral language proficiency. Indeed, studies have found that DLLs substantially lag behind monolingual norms in their word production in both languages (Boyce, Gillam, Innocenti, Cook, & Ortiz, 2013; Páez, Tabors, & López, 2007), in some cases by as many as two standard deviations (Hammer, Lawrence, & Miccio, 2008). Most DLLs experience vocabulary growth over preschool and kindergarten in both English and Spanish (Páez et al., 2007), and given that DLL children are developing two languages, it is perhaps not surprising that they would lag behind monolingual children in both. Studies have found, however, that even when summing DLLs’ vocabulary knowledge in both languages, a lag behind their same-age monolingual peers persists (Boyce et al., 2013).

Such findings are troubling, since early oral language skill is a critical component of later literacy achievement, which many researchers view as the foundation for academic success (Durham et al., 2007, National Early Literacy Panel, 2008; NICHD Early Child Care Research Network [ECCRN], 2005). One longitudinal analysis of DLLs using nationally representative data found that kindergarten vocabulary skill in both English and Spanish predicted levels of English reading in third through eighth grade (Kieffer, 2012). These findings mirror those of nationally representative studies with monolingual English-speaking children (NICHD ECCRN, 2005). Early difficulties with language skill in both English and Spanish may contribute to achievement gaps in both reading and math between DLLs and their language majority peers (Han, 2012, Reardon and Galindo, 2009). Such gaps remain after controlling for SES, and indicate that DLLs with limited English proficiency may fall as far as two grades behind by eighth grade (Halle et al., 2012).

Research has consistently shown that DLL children benefit from the same high quality instructional environments as language majority children (Goldenberg, Hicks, & Lit, 2013; Slavin, Madden, Calderón, Chamberlain, & Hennessy, 2011). Nevertheless, there are special considerations for children who enter school with relatively weak English skills. Pragmatically, DLL children in the United States need skills in English to succeed in school and beyond. Developing or maintaining skills in their first language, however, hold notable cognitive benefits (Barac, Bialystok, Castro, & Sanchez, 2014), is important for maintaining connections to their family and culture (Wong-Fillmore, 2000), and can expand career opportunities later in life (Proctor, August, Carlo, & Barr, 2010; Rumbaut, 2014). Consequently, the appropriate balance of the use of the first and second languages in the classroom to support skills in both languages is an important consideration for early childhood educators of DLLs.

The classroom language context includes both the language of instruction and the peer composition of DLLs in the classroom. Both factors may exert a significant impact on DLL children’s development in both English and Spanish. Bilingual instruction has long been a politically contentious topic, but the weight of the evidence from the early elementary grades indicates that providing instruction in the child’s first language (L1) promotes the maintenance of the L1 at no cost to development of the child’s second language (L2). A meta-analysis by Cheung and Slavin (2012) found a modest average effect size of 0.21 in favor of educational programs in elementary school that give children opportunities to develop and use their L1. Further, two recent studies offering evidence on the long-term effects of bilingual instruction found that children who received bilingual instruction in the early grades of elementary school experienced faster literacy and math growth, were more likely to be reclassified to mainstream education, and had higher English proficiency at the end of high school than children who received English-only instruction (Umansky & Reardon, 2014;Valentino & Reardon,2015).

Fewer studies have examined the language of instruction in the preschool years. A few randomized control trials (RCTs) have evaluated bilingual programs such as two-way immersion – in which teachers use the home language for half the day and English for the other half – and transitional bilingual – in which teachers initially use a higher proportion of the home language and eventually transition to a mix of both, and then to a higher proportion of English. These studies have found that relative to English-only programs, preschoolers in the bilingual programs enjoyed an advantage in their Spanish language skills, with no difference for their English skill (Barnett, Yarosz, Thomas, Jung, & Blanco, 2007; Durán, Roseth, Hoffman, & Robertshaw, 2013; Farver, Lonigan, & Eppe, 2009). For example, one study comparing the effects of a two-way Spanish immersion program and a monolingual English immersion program on preschoolers’ Spanish and English vocabulary found no significant differences between the two treatment groups’ English development, but the bilingual program resulted in substantial gains in Spanish skill (Barnett et al., 2007). A similar study comparing the effects of transitional bilingual and monolingual Head Start programs found that the positive effect of bilingual instruction on Spanish vocabulary, and the null effect of bilingual instruction on English vocabulary was sustained through a three-year follow-up (Durán et al., 2013). Taken together, these studies suggest that when children are instructed in their L1 in addition to their L2 in preschool, they are able to develop their L1 abilities, while also promoting, or at least not undermining, the development of their L2 abilities.

Many DLL preschoolers, however, do not have access to their home language in the preschool classroom (Figueras-Daniel & Barnett, 2013;Tabors & Snow,2003), and most preschools do not have formal bilingual programs, such as two-way immersion or transitional, instead using the L1 in an ad-hoc manner (Figueras-Daniel & Barnett, 2013). Head Start, for example, does not prescribe language use; the language of instruction is left to individual programs to decide, and programs serving DLL Spanish speakers vary considerably in the proportion of English and Spanish used by teachers. The effects on language and literacy development found in studies of classrooms where Spanish is used at the teacher’s discretion rather than within a structured program as evaluated by the RCTs discussed above, are not necessarily seen (Burchinal, Field, López, Howes, & Pianta, 2012; Hindman & Wasik, 2015). One observational study, for example, found no evidence of a main effect of the proportion of Spanish used in the classroom on English literacy skill (Burchinal et al., 2012), and in a study with Head Start preschoolers, the association between using any Spanish for instruction and vocabulary skill in Spanish or English was not significant (Hindman & Wasik, 2015).

In summary, relatively few studies have examined the language of instruction in preschool. Evidence from RCTs suggests that using both English and Spanish results in similar language development in English, and stronger development in Spanish relative to monolingual English instruction. In contrast, findings from descriptive studies examining natural variation in English- and Spanish use in preschool are mixed. Some have not found significant associations between the language(s) used in the classroom and child outcomes, and others have found a negative association between Spanish use and English language proficiency.

With few exceptions (e.g. Burchinal et al., 2012) studies on the effects of the languages used for instruction largely do not capture the wide variation in language contexts outside of prescribed bilingual education programs, as evaluated through RCTs. Though it has been established in the literature that using the L1 has value, and that increased exposure to a language is correlated with gains in that language (Boyce et al., 2013, Gámez, 2015; Pearson, Fernandez, Lewedeg, & Oller, 1997), it is unclear whether there is an optimal balance for teachers’ use of the first and second languages. Furthermore, the bilingual programs that have been evaluated with RCTs, such as two-way immersion and transitional bilingual, are intended to use English and Spanish equally, or with greater initial use of Spanish that transition to greater use of English over the school year. Such programs do not necessarily reflect the range of language use in the classroom, in which teachers may use more Spanish than English for instruction. Consequently, it is unclear how the balance of using more Spanish than English in the classroom relates to children’s vocabulary development in their L1 and L2. Furthermore, it is unclear from studies that evaluate specific bilingual programs whether the effects identified are due to the specific instructional program or to the language used for instruction (Buysse et al., 2013).

The present study takes advantage of the natural variation in teachers’ use of Spanish and English in Head Start programs to examine differences in classroom language use on children’s Spanish and English vocabulary development. Based on evidence from evaluations of bilingual programs, I expect that children in classrooms that use a mix of English and Spanish for instruction will have similar English vocabulary scores to children in classrooms using English only, while children in classrooms using mostly Spanish may have slower English vocabulary growth, relative to English-only classes. I would expect the reverse to be true for Spanish vocabulary. Specifically, I hypothesize that children’s Spanish vocabulary scores will not be different when they receive instruction mostly in Spanish versus in both English and Spanish, but that children in English-only classrooms will have lower Spanish vocabulary growth than children in mostly Spanish classrooms.

The classroom language context is not limited to the languages used by teachers; peers may also influence language development. A small but growing number of studies suggest that English-speaking monolingual children’s language skills benefit from sharing early childhood classrooms with classmates who have relatively high language abilities (Justice, Petscher, Schatschneider, & Mashburn, 2011; Mashburn, Justice, Downer, & Pianta, 2009; McGregor, 2000). Peers’ English language skill may be particularly important for DLL children developing English as a second language. In two studies with preschool DLL children that examined the language of peers, the proportion of total interactions that occurred with peers in English was positively associated with DLL children’s year-end English expressive vocabulary knowledge, controlling the teacher’s use of English, and children’s total language use and language proficiency in the fall (Palermo and Mikulski, 2014, Palermo et al., 2013). These findings suggest that exposure to peers’ English may support DLL children’s English word production.

Extant research has not provided evidence on whether the composition of DLLs in classrooms is related to DLLs’ native language and English language development. One study with kindergarteners and first graders, however, found that higher proportions of English learners in the classroom was associated with lower English reading test score gains for native English speakers (Cho, 2012). Furthermore, since prior research has found that teachers use the child’s L1 or L2 in response to children’s own language preferences (Stipek, Ryan, & Alarcón, 2001), it is likely that teachers use more Spanish than English in classrooms with higher proportions of DLLs. Prior research has not, however, examined the independent contribution of these two aspects of the classroom language context. In the present study I reasoned that DLLs would have more opportunities to hear and to practice English if they shared classrooms with relatively more native English-speakers, which would be associated with more rapid English vocabulary development than if they were in classrooms with mostly other DLLs. A higher proportion of native English speakers, however, may conversely limit children’s exposure to Spanish vocabulary, and relate to lower growth in Spanish vocabulary. Higher proportions of DLLs, on the other hand, may give children more opportunities to hear and speak Spanish, and thus support their Spanish vocabulary ability, while perhaps limiting their English vocabulary ability.

A number of studies have found that a range of factors impact the academic development of DLL children living in poverty (Halle et al., 2012; Kim, Curby, & Winsler, 2014). In this study, I examine a key part of DLL children’s preschool experience: the language context of the classroom. I build on prior studies of teacher language with Head Start DLLs (Hindman & Wasik, 2015) by including more nuanced indicators for the amount of English and Spanish instruction used in the classroom. I also analyze a less examined element of the classroom language context, the proportion of children in the class who are DLLs. I thus provide evidence on the effects of differing degrees of bilingual instruction, as well as the classroom peer composition, using a nationally representative sample of low-income preschoolers. My research questions are as follows:

• Is using English only, a mix of both English and Spanish, or mostly Spanish associated with DLLs’ vocabulary growth in English and Spanish?

• Is the proportion of DLLs in the classroom associated with DLLs’ vocabulary growth in English and Spanish?