Home literacy environment and existing knowledge mediate the link between socioeconomic status and language learning skills in dual language learners

Highlights

• Spanish–English Dual Language Learners’ language learning skills vary by SES.

• The magnitude of the SES effect does not vary by child age.

• Home literacy environment and children’s existing language knowledge mediate the SES effect.

Abstract

Children learning two languages (Dual Language Learners; DLLs) represent a rapidly growing population in the United States. DLLs are disproportionately more likely to live in families of low socioeconomic status (SES), which places many of them at risk for poor dual-language outcomes. To date, most studies on SES and dual language development have relied on static measures of vocabulary and syntactic skills, without examining the language learning processes – children’s ability to acquire new language items. The current study used a newly developed language measure, the Quick Interactive Language Screener: English and Spanish (QUILS:ES), to assess 3- through 5-year-old Spanish-English DLLs’ language learning processes. We also examined the association between SES (as measured by primary caregivers’ education) and language learning processes, and further explored mechanisms underlying the association. DLLs from higher-SES families showed better language learning skills than those from lower-SES families. The size of the gap did not vary by child age. Home literacy environment (i.e., access to books, book-reading frequency) and children’s existing knowledge (i.e., vocabulary and syntactic knowledge) mediated the SES effect. Together, these findings highlight the need to better prepare DLLs from low-SES families for learning from a dual-language environment. Supporting DLLs’ language environment and knowledge through learning materials and language and literacy activities in both languages is crucial for ameliorating the SES gap in language learning processes.

Introduction

Dual Language Learners (DLLs) – children exposed to a language other than English at home – represent 32% of the U.S. population under age 8; and the majority of them are from Spanish-speaking families (McCabe et al., 2013; Park, O’Toole, & Katsiaficas, 2017). Experiences and abilities in both languages are foundational for DLLs’ future success (Hammer, Lawrence, & Miccio, 2008; Iglesias & Rojas, 2012; Tamis-LeMonda et al., 2014). While early English skills predict DLLs’ academic trajectory throughout the school years (Kieffer, 2008; Halle, Hair, Wandner, McNamara, & Chien, 2012), strong home language development benefits English learning and promotes positive family relationships and socioemotional adjustment (Hammer et al., 2014, Oh and Fuligni, 2010). Despite the importance of early dual language skills, contextual factors such as poverty or low socioeconomic status (SES) place many DLLs at disproportionate risk for poor language outcomes. Indeed, approximately 28% of DLLs from Spanish-speaking households are living below the federal poverty rate, compared to an overall childhood poverty rate of 19% in the United States (Child Trends, 2018, Koball and Jiang, 2018). Childhood poverty relates to increased risk for a range of negative developmental outcomes, including language delays and poor academic achievement (Hair, Hanson, Wolfe, & Pollak, 2015). There is a need to understand how SES relates to dual-language development, and how the effect of SES changes over preschool years, a critical time for DLLs to gain school readiness skills.

To date, studies on SES and dual-language acquisition have almost exclusively focused on language product – what children know about the two languages (Calvo & Bialystok, 2014; Gathercole, Kennedy, & Thomas, 2016). However, little is known about whether SES relates to DLLs’ language learning processes – their ability to gain new words and grammatical structures in one language or both of their languages. Do children from different SES backgrounds vary in their ability to learn from their language environment? If so, why? These are important questions, because language learning processes are a prerequisite for language acquisition, which in turn is foundational for long-term language and academic growth (Maguire et al., 2018). Answers to these questions will reveal new opportunities to narrow the language gap. If SES differences in DLLs’ language learning processes do exist, language interventions may be more effective if language learning skills are targeted in addition to the quantity and quality of children’s language environment (Maguire et al., 2018). Additionally, understanding the contextual contributors to children’s language learning processes will help reduce bias in the diagnosis of language disorders in DLL and low-income populations (Peña, Iglesias, & Lidz, 2001; Weismer & Evans, 2002).

The current study examined the association between SES and Spanish-English DLLs’ language learning processes. We assessed 3- through 5-year-old children’s language knowledge (i.e., vocabulary and syntax) and language learning processes in both languages using a new language measure, the Quick Interactive Language Screener: English and Spanish (QUILS:ES; Iglesias, de Villiers, Golinkoff, Hirsh-Pasek & Wilson, in press). In particular, we investigated whether SES related to language learning processes; and if so, how the strength of the SES effect changed over preschool years and what factors mediated the SES effect.

Language learning processes refer to children’s ability to gain the meaning of new language items from language environment. In the current study, we examined children’s language learning processes using two well-studied approaches: Fast mapping and syntactic bootstrapping. Fast mapping is the process in which children learn and extend new words with minimal exposure (Carey & Bartlett, 1978). For instance, when presented with an unfamiliar, previously unnamed object and a familiar one (e.g., an apple), a young child can rapidly match a novel word (e.g., “dax”) with the unfamiliar object. Syntactic bootstrapping allows children use syntactic structures to infer the meaning of words (Gleitman, 1990, Naigles, 1990). An example would be when hearing the sentence “he is praving something”, a child can understand that the novel word “praving” refers to an action rather than an object. Both fast-mapping and syntactic bootstrapping are important mechanisms that support efficient language acquisition.

Language learning processes are thought to involve both experience-independent and experience-dependent mechanisms. Unlike vocabulary and syntactic knowledge that are highly associated with language input, language learning processes might be less dependent on prior experiences. Some researchers have proposed that language learning processes such as those utilized during fast mapping may be relatively “experience-independent” (Campbell, Dollaghan, Needleman, & Janosky, 1997). However, recent evidence has shown that language learning processes are not fully static, but rather are malleable and influenced by experience-dependent factors, such as children’s existing language skills and language environment (e.g., Aravind et al., 2018; Bion, Borovsky, & Fernald, 2013; Byers-Heinlein & Werker, 2009).

The notion of “language begets language” emphasizes the role of existing language knowledge and skills in language learning (Hemphill & Tivnan, 2008). Evidence suggests that children’s existing vocabulary and grammatical knowledge relates to their fast mapping skills (Aravind et al., 2018; Bion et al., 2013; Gray, 2004; Wilkinson & Mazzitelli, 2003), and their ability to learn nouns and verbs through reading and instruction (Hill, Wagovich, & Manfra, 2017; Kelley, 2017). Knowing more words and grammatical structures allows children to better target the referent of a new word in ambiguous situation, form a more nuanced and deeper understanding of words and grammatical structures, access and retrieve their linguistic knowledge more efficiently, and integrate semantic information more successfully (Bion et al., 2013; Cain, Oakhill, & Elbro, 2003; Hill et al., 2017; Huang, Leech, & Rowe, 2017; Maguire et al., 2018; Weisleder & Fernald, 2013), all of which are critical for language learning processes (Bion et al., 2013; Hurtado, Marchman, & Fernald, 2008; Jackson, Leitao, & Claessen, 2016; Kan, Sadagopan, Janich, & Andrade, 2014). For instance, strengthening the phonological representations of novel words through speech practice has been found to successfully improve fast mapping performance (Kan et al., 2014). Notably, the association between existing language skills and language learning processes is likely bidirectional. Children with better language learning skills early on show better language outcomes years later (Bion et al., 2013; Venker, Kover, & Weismer, 2016). As such, language learning processes and existing language skills may influence each other reciprocally as an integrated language learning system.

Environmental input can also support the development of language learning processes. Language environment with high quantity and quality, characterized by diverse, complex, and contingent language input, question use, inferential talk, and rich literacy experiences (e.g., books and book-reading activities), promotes children’s vocabulary and grammatical knowledge (Farver, Xu, Lonigan, & Eppe, 2013; Hirsh-Pasek et al., 2015; Hoff, 2003; Rowe, 2012; Rowe, Leech, & Cabrera, 2017; Tompkins, Bengochea, Nicol, & Justice, 2017), which may in turn support successful language learning processes. Alternatively, home language environment may directly facilitate language learning processes by offering children the opportunities to hone skills that are essential to language learning. For instance, the quantity, lexicon diversity, and syntactic complexity of language input have been found to predict word recognition speed (Hurtado et al., 2008). An experimental study suggested that asking questions about new words during book-sharing enhanced children’s ability to form word-referent associations (Blewitt, Rump, Shealy, & Cook, 2009). Repeated and varied exposures to words in multiple ways increased children’s ability to detect recurring linguistic structures (Schwab & Lew-Williams, 2016) and allowed them to gain more refined and deeper word knowledge (Coyne, McCoach, Loftus, Zipoli, & Kapp, 2009).

Language learning processes are supported by existing language skills and the language environment, both of which are experience-dependent and therefore subject to contextual influences. The current study focused on SES as a critical contextual factor that can shape children’s language skills and experiences, and potentially language learning processes.

In both monolingual and DLL samples, SES, often indicated by parental education and/or household income, predicts children’s English vocabulary, syntactic knowledge, and language processing skills (Buac, Gross, & Kaushanskaya, 2016; Calvo & Bialystok, 2014; Fernald, Marchman, & Weisleder, 2013; Gathercole et al., 2016; Gonzalez et al., 2017; Huttenlocher, Waterfall, Vasilyeva, Vevea, & Hedges, 2010; Pace, Luo, Hirsh-Pasek, & Golinkoff, 2017). The SES-related gaps in language product can be partially explained by differences in the home learning environment, including the quality of parent-child language interactions, the availability of literacy resources, and the frequency of language learning activities (Farver et al., 2013, Gonzalez et al., 2017, Hoff, 2003, Huttenlocher et al., 2010, Pace et al., 2017).

Fewer studies have examined the link between SES and DLLs’ Spanish outcomes, which tends to be language-specific. That is, mothers’ highest education achieved in Spanish predicts children’s Spanish vocabulary, whereas their highest education achieved in English relates to children’s English vocabulary (Hoff, Burridge, Ribot, & Giguere, 2018). Some studies with Spanish-dominant families in the United States found no association between maternal education and DLLs’ Spanish vocabulary (Deanda, Arias-Trejo, Poulin-Dubois, Zesiger, & Friend, 2016; Hammer et al., 2012; Place & Hoff, 2016). Yet, there is also evidence that an association between parental education and DLL’s Spanish vocabulary exists in families where parents use both English and Spanish (Stadthagen-González, Gathercole, Pérez-Tattam, & Yavas, 2013). Perhaps it takes effort to maintain a bilingual home environment. Parents who use both languages tend to highly value bilingualism, and those with higher-SES may have more resources to achieve their goal to support children’s bilingual development (Stadthagen-González et al., 2013). However, some (not all) Spanish-dominant parents may value English learning but predominantly use Spanish due to their low English proficiency.

Building on previous work on the relation between SES and language knowledge, the current study examined SES differences in language learning processes. Despite the considerable individual variation in children’s performance on language learning tasks (Kelley, 2017), little is known about the extent to which contextual factors such as SES contribute to the development of language learning processes, especially in DLL populations. The few studies on this topic have yielded mixed findings.

A few studies have found that, even though children from low-SES families tend to receive less rich language input and have less developed vocabulary and grammar, they might be as successful as children from higher-SES families in acquiring language when learning opportunities are presented. A study showed that African American toddlers from low-SES and mid-SES families performed similarly on a task of fast mapping nouns, despite the robust SES differences in their existing knowledge of vocabulary (Horton-Ikard & Weismer, 2007). However, this study only included a small sample (n = 30) with a restricted age range (30–40 months). Evidence suggests that the SES-related gap in vocabulary widens during infancy (Fernald et al., 2013). It is plausible that SES differences in language learning processes start to emerge in preschool years when the gap in language knowledge becomes large.

Similarly, another study used a dynamic assessment to measure African American kindergarten children’s ability to learn novel nouns through story reading, and did not find any SES differences, although this might be due to the low performance across the two SES groups (Burton & Watkins, 2007). Additionally, children were classified into low-risk (i.e., relatively higher SES) or high-risk (i.e., relatively lower SES) groups based on whether they received free lunch, whether their mothers attended at least some college, and whether their school had a concentration of children from low-income families. The relatively low variability in SES, in combination with the small sample size (n = 24), could result in limited power to detect the SES effect.

Another study did not find SES differences in Spanish-English DLLs’ ability to learn novel English nouns (Buac et al., 2016). Children heard a list of novel words paired with referents and then selected pictures corresponding to the novel words in a recognition task. Notably, the study tested children’s memory of word-referent associations rather than their ability to infer the meaning of words. Additionally, the study only examined English word learning and did not control for children’s relative language exposure at home (e.g., English only, Spanish only, or both English and Spanish), a potential confounder of the SES effect.

The Eco-Bio-Developmental Model posits that environmental factors such as SES can alter brain architecture, which in turn impacts basic learning and cognitive functioning traditionally thought to be independent from experiences (American Academy of Pediatrics, 2014). Indeed, children from different SES backgrounds vary in brain structure and brain activation associated with language and literacy (Farah, 2017, Hackman and Farah, 2009, Kuhl, 2011) and their ability to process vocabulary and syntactic input during real-time comprehension (language processing efficiency; Fernald et al., 2013, Huang et al., 2017). Likewise, there are SES differences in other processing-based tasks, such as non-word repetition and sentence repetition (Buac et al., 2016, Chiat and Polišenská, 2016, Meir and Armon-Lotem, 2017). Given that the ability to process language input is foundational for language acquisition, SES might also play a role in language learning processes. Additionally, compared to other processing skills, the language learning processes examined in the current study might more closely depend on children’s existing language knowledge.

Only two studies to our knowledge have documented the association between SES and language learning processes. One study assessed 3- through 5-year-old, monolingual children’s learning of nouns, verbs, and adjectives using the English monolingual version of the Quick Interactive Language Screener (Levine et al., 2020). Results revealed an SES gap throughout preschool years, the magnitude of which did not vary by child age. Another study tested school-age children’s ability to learn novel English nouns through reading sentences. Children from low-SES households learned fewer novel words than their more affluent peers (Maguire et al., 2018). The SES effects on English word learning did not vary by children’s DLL status. Nonetheless, the findings were limited to English learning processes and were based on a mixture of children who are English-speaking monolingual and DLLs from diverse linguistic backgrounds.

The inconsistent findings reviewed above might be due to differences in the age groups studied and tasks used, the limited variability in SES, as well as the lack of control of DLLs’ relative language experiences. Previous work with toddlers did not find significant SES differences (Horton-Ikard & Weismer, 2007), yet some studies found SES differences in preschooler and school-age children (Levine et al., 2020; Maguire et al., 2018). The SES effect in language learning processes is likely to emerge in preschool years when SES differences in children’s language experiences and skills become large. Thus, the current study focused on 3- through 5-year-old preschoolers from a wide range of SES backgrounds (i.e., parental education ranging from lower than high school to graduate degree), while controlling for children’s dual-language home experiences. We also explored developmental changes in the strength of SES effect over the preschool years. Based on previous work, we expected the SES effect to be stable over time. Additionally, the few studies of children who are DLLs exclusively focused on the learning of English nouns, without examining children’s ability to learn other word categories (e.g., verbs, adjectives) and their learning processes in the home language. Thus, the current study examined SES differences in DLLs’ learning of nouns, verbs, and adjectives in both English and Spanish.

Another aim of the study was to explore the underlying mechanisms of the likely association between SES and language learning processes. Here we examined two potential mediators: Existing language knowledge and home literacy environment.

The substantial SES gap in children’s existing vocabulary and syntactic knowledge can place children from low-SES families at a disadvantage in language learning. As reviewed above, the notion of “language begets language” highlights the role of existing language skills in language learning. Vocabulary and syntactic knowledge are essential components of language ability and are related to other language skills, such as language processing efficiency (Hurtado et al., 2008). Indeed, a study found that children’s existing vocabulary mediated the association between SES and word learning skills (Maguire et al., 2018), although these findings were primarily based on school-age monolingual children. However, the few studies examining language knowledge and language learning processes with DLLs have yielded mixed findings. A study with Hmong-English DLLs found no significant correlation between receptive vocabulary and fast-mapping ability within each language (Kan & Kohnert, 2008), whereas other studies with Spanish-English DLLs found an association between English vocabulary and English novel word learning (Buac et al., 2016). Additionally, typically developing DLLs outperformed those with primary language impairment in a word learning task (Kapantzoglou, Restrepo, & Thompson, 2012). Thus, more evidence is needed to understand the role of existing knowledge in the link between SES and language learning processes in DLL populations.

SES might also exert its impact through home literacy environment. Here we focused on home literacy environment (e.g., the availability of books and the frequency of book reading activity) as a proxy of high-quality language experiences, because book-related experiences expose children to rich, diverse, and complex language input and questions (Peterson & McCabe, 1994; Soderstrom & Wittebolle, 2013; Tamis-LeMonda, Custode, Kuchirko, Escobar, & Lo, 2018), which have been found to be important for language learning. According to the parental investment model, parents from low-SES background have limited financial resources and human capital to invest in their children’s development (Becker and Tomes, 1986, Haveman and Wolfe, 1994). Consequently, children from lower-SES families tend to have fewer books and less frequent book-reading experiences than their peers from higher-SES families, which may further lead to disadvantages in vocabulary, oral language, and emergent literacy skills (Farver et al., 2013; Froiland, Powell, Diamond, & Son, 2013; Mol & Bus, 2011). Notably, children can also actively shape their home literacy environment, such that children with stronger language skills may seek out more literacy experiences and elicit higher-quality language input from their parents (Harste, Woodward, & Burke, 1984; Park, Tsai, Liu, & Lau, 2012; Sy, Gottfried, & Gottfried, 2013). However, there is some evidence that the influences from parents to children tend to be more robust than those from children to parents (Leigh, Nievar, & Nathans, 2011). Therefore, the current study focused on the influence of home environment on language knowledge and learning processes rather than vice versa.

Few studies have examined whether the home literacy environment affects children’s language learning processes though it might relate either directly or indirectly through the mediation of existing knowledge (i.e., home literacy environment → language knowledge → language learning processes). Recently, a growing body of work reveals the impact of the home literacy environment on children’s language processing skills (Hutton et al., 2015, Hutton et al., 2017). For instance, when listening to a story, preschool children who had richer literacy materials and more frequent book-reading at home showed greater activation in brain areas related to semantic language processing (Hutton et al., 2015). Thus, children with richer home literacy environments might be better prepared to learn from their linguistic environment. Again, most studies were based on monolingual samples. Whether these findings apply to DLL populations in one or both languages is still unknown.

The ability to learn language is fundamental for language growth and academic success. Understanding the nature and mechanisms of SES differences in language learning processes is critical for addressing the language gap and improving the development of children living in adverse contexts. This is especially important for DLLs, who are disproportionately likely to live in poverty and are at risk for having academic delays (Hoff, 2013, Park et al., 2017). The current study examined SES differences in Spanish-English DLLs’ language learning processes by asking three questions.

• Does SES relate to DLLs’ English and Spanish learning processes? We expected SES to be related to DLLs’ language learning skills in both English and Spanish.

• Does the strength of the SES effect vary based on child age in a cross-sectional design? Based on prior findings with monolingual children (Levine et al., 2020), we expected the SES effect to be stable over the preschool years.

• Which pathways explain the effect of SES on language learning processes? We tested home literacy environment and existing language knowledge as two sequential mediators of the association between SES and language learning processes. Specifically, there might be three potential pathways: (1) SES → home literacy environment → language learning processes; (2) SES → language knowledge → language learning processes; and (3) SES → home literacy environment → language knowledge → language learning processes (see Fig. 1).

Notably, the relationship between language knowledge and language learning processes can be bidirectional, such that early language learning processes may support children’s language knowledge later (Bion et al., 2013; Venker et al., 2016). Although the cross-sectional nature of the study precludes a test of the directionality of the relationship, in supplementary analyses we explored two additional pathways: (1) SES → language learning processes → language knowledge, which is an alternative to the second pathway described in the previous paragraph; and (2) SES → home literacy environment → language learning processes → language knowledge, which is an alternative to the third pathway described above.