Predicting literacy development and risk in Spanish-English bilingual first graders

Abstract

Purpose: There are well-established links between oral language and reading development in monolingual English-speaking children that are associated with literacy outcomes. Oral language, defined relative to lexical quality, provides key support for developing early reading skills. For bilingual children, the connection between oral language and reading development and risk for identification of having language or reading difficulties is assumed but has been less systematically studied. Our goal was to document the relationship between knowledge of semantics and morphosyntax and children's performance on letter and word identification in first grade in each of their languages so as to evaluate the risk for poor reading outcomes in the Spanish and English of bilingual first graders. Method: We examined data from 111 Spanish-English bilingual first graders on measures of Spanish and English semantics and morphosyntax in relation to letter-word identification, which indexes early reading development. We evaluated the within- and across-language relationships between skills and also whether the same children who performed in the at-risk range on the oral language tasks showed risk for future reading difficulties based on their letter-word performance. Results: Oral language skills and early reading skills were strongly related within languages, and cross-language relationships were also evident, though not for all domains. There were significant associations between risk for developmental language disorder (DLD) based on semantics performance and reading risk. Conclusions: We observe cross-linguistic relationships between letter-word identification and morphosyntactic and semantic skills in young Spanish-English bilinguals with and without risk for language and reading development. These findings indicate that connections between oral language and reading hold across differing levels of language and reading risk.

Keywords

bilingualism children developmental language disorders literacy Spanish English

The quality of children's lexical representations is foundational to both early and long-term literacy outcomes, whether measured by decoding or comprehension measures (Perfetti, 2007; Snowling and Hulme, 2021). Lexical quality refers to the richness, completeness, and flexibility of the phonological, orthographic, and semantic representations of words. Children with underspecified lexical representations, including children with risk for developmental language disorder (DLD) and bilingual children, may later have reading difficulties as indexed by measures of decoding such as letter identification (Snowling et al., 2016). Estimates of comorbidity of reading difficulties and DLD range from 43–55% up to 69% depending on the stringency of the cut point employed (Catts et al., 2001; Newbury et al., 2011).

Bilingual children from multilingual environments tend to have better reading outcomes when they learn to read in their home language first (e.g. Cummins, 1979). In the US context, where children are more likely to learn to read in English than their home language, nearly 68% of fourth graders classified as English language learners read below a basic level, and 91% could not read proficiently (National Assessment of Educational Progress, 2019). The 2018 Programme for International Student Assessment (PISA) results illustrate that this challenge is wide-reaching (Schleicher, 2019). In 28 of the 42 countries with over 5% immigrant populations surveyed in the PISA, children from immigrant backgrounds (who are more likely to be bilingual) had lower than expected reading scores than their non-immigrant peers even when adjusted for socioeconomic status (Schleicher, 2019). As a follow-up to these observations of lower than expected reading scores, it is important to consider what the rate of reading risk is when we consider both of a child's languages. By systematically studying the association between letter and word reading (a commonly used task on measures of reading achievement) and language skills in bilinguals with and without risk for DLD, we can gain insight into the nature of the association between semantic and morphosyntactic knowledge- both aspects of quality of lexical representations—and letter-word reading scores as an index of reading risk.

The simple view of reading posits that strong reading skills emerge when readers’ word recognition and language comprehension are high (Hoover and Gough, 1990), a principle that holds across languages (Verhoeven and Perfetti, 2017). High-quality lexical representations contribute significantly to reading development (Language Reading Research Consortium et al., 2016). This relationship has been observed between English (L2) vocabulary scores and reading in Spanish-English bilinguals (e.g. Goodrich and Namkung, 2019; Goodrich et al., 2022; Huang et al., 2022) but have not been evaluated in Spanish (L1). Prevoo et al. (2016) conducted a meta-analysis of 86 studies focusing on the relationship between oral language proficiency and school outcomes in bilingual children. Moderate to strong cross-language relationships occur between early reading development and oral proficiency in bilinguals, especially when the two languages are similar in their writing systems (Bialystok, 2007; Zhang and Koda, 2014). In particular, there were robust cross-language correlations found in phonemic awareness and morphological awareness (Chung et al., 2018; Ramirez et al., 2010). This means that accurate and robust representations of a word's form, or lexical quality (including its orthography, phonology, morphology, and syntax), its meaning (or semantic information), and its interconnectedness with other words are crucial for early reading development (Perfetti and Hart, 2001). Morphosyntactic and semantic knowledge are most closely related to reading comprehension (e.g. Muter et al., 2004; Share and Leiken, 2004). However, morphological awareness, which builds on grammatical skills, can help learners gain access to information about word forms and the interrelationship between words that supports efficient processing of written words (e.g. Apel and Diehm, 2014). Bilingual learners start with strong connections between words across their languages (e.g. Conboy and Thal, 2006) and strong connections between words and grammatical constructions within their languages. From this foundation, morphological awareness builds in both languages rather than directly transfers (Lucero, 2015; Simon-Cereijido et al., 2013). That said, languages differ in their orthographic transparency, and this affects the ease of decoding. In Spanish, grammatical markings provide reliable and salient cues to meaning (e.g. Bedore & Leonard, 2001). Analogous cues in English are less salient (Leonard, 2014) so it may require more exposure for children to fully leverage the affordances of these connections.

Children with DLD experience delays in vocabulary development and marked difficulties in grammatical development. By school age and the onset of reading instruction, early vocabulary delays often manifest in word finding difficulties associated with poor semantic development (McGregor et al., 2013; Sheng and McGregor, 2010). Monolingual and bilingual children with DLD are more prone to produce errors when asked to generate items in common categories (i.e. animals that live on a farm) and less likely to include high frequency or prototypical exemplars of these categories than their peers (e.g. Sheng et al., 2012; Shivabasappa et al., 2018). Deficits in grammatical morphology are associated with reduced morphological awareness and can interfere with representations of lexical forms (Apel, 2014). Employing measures of semantic and grammatical components of lexical quality may lead to a better understanding of the interplay between language and reading, especially in DLD.

With this in mind, our focus is on semantics and morphosyntax. Semantic knowledge comprises a child's knowledge of word meaning such as the category a word belongs to (e.g. taxonomic classification), the characteristic properties that distinguish the meaning of related words (e.g. shape, color, size) or the kinds of relations that a word has with other words such as analogical relationships. For bilingual children, semantic measures that capture underlying conceptual knowledge that converges across the child's languages are more informative than traditional receptive or expressive vocabulary measures that are more sensitive to the amount of time or experience children have in the languages of assessment (Peña et al., 2016). Morphosyntactic knowledge comprises information about the kinds of grammatical markers that can be used with words (i.e. which verb tenses are most commonly used with a verb, grammatical gender marking to establish concordance, or where a noun is a mass noun that requires no overt plural marking or a count noun that does). Robust knowledge of these properties of language is associated with high lexical quality.

DLD and bilingualism are both associated with reduced quality of lexical representations. DLD interferes with children's ability to learn new words and their ability to learn grammatical representations of words (see Bedore and Peña, 2008 for discussion). For bilingual children, lexical quality may be diminished when divided experience results in less developed representations of the words children know (Mancilla-Martinez and Lesaux, 2011; Proctor et al., 2012). Models of bilingual lexical representation suggest that both languages’ lexical networks may be linked (e.g. Dijkstra et al., 2019). Despite this interconnection of lexical representations across languages for bilinguals, less is known about their relationships for children with DLDs. Children with DLD are at particular risk for reading difficulties because of their widespread deficits in lexical representations (Bishop and Snowling, 2004); but when children are bilingual, cross-language associations in their lexical representations may affect their early reading acquisition.

In this study, we address two main research questions to better understand the relationship between knowledge of semantics and morphosyntax and letter-word identification as an indicator of reading risk:
How much variance in letter-word identification is accounted for by semantic and morphosyntactic performance and language experience?

What is the relationship between risk for low letter-word reading scores and semantic and morphosyntactic scores where risk is defined as performance below the 25th percentile?
Based on past research, we expected that semantic and morphosyntactic knowledge would account for significant amounts of variance in the children's letter-word reading skills and that language experience would impact performance in each language. In regard to our second question we expected that there would be a high degree of congruence between language and reading risk. Past work shows that language proficiency in bilingual children is associated with reading achievement and that between 17 and 71% of children with DLD have dyslexia (Ehrhorn et al., 2021).

Method

Participants

The study was approved by the Institutional Review Board at The University of Texas at Austin. Data from 111 Spanish-English bilingual first graders were selected from 147 participants in a study of language and reading development. The 111 children had complete data for both languages on all target measures and had a minimum of 20% exposure to their two languages. They attended six schools in two school districts in central Texas with large Latino populations. The schools provided dual language instruction during the early elementary grades gradually increasing the amount of English instruction with the goal of transition to English typically by second grade. The specific amounts of Spanish and English spoken varied by teacher. In first grade, most instruction was in English but many children received language arts and reading instruction in Spanish if that was considered by the school to be their stronger language. Children were tested in the fall semester of first grade when reading instruction begins to be systematically implemented (Texas Education Agency, 2020). Individual exposure to English and Spanish (input and output) was measured using the Bilingual Input Output Survey (BIOS) which is part of the Bilingual English Spanish Assessment (BESA) (Peña et al., 2018) via interviews with parents and teachers. We also gathered demographic information (race, ethnicity), socioeconomic status, history of language exposure, and parental perception of their children's language abilities during the interview. Reported average input and output in English was 40.1% (SD = 13.4%), and Spanish was the inverse. The average age of first exposure to English was 2.77 years (SD = 2.01). The children's mean age was 82.6 months (SD = 5.2), and 44% of the sample was female. Average maternal education was measured using the Hollingshead scale (1975). The average score for maternal education was 2.38 (median score = 2), consistent with a ninth-grade education.

Measures

Parents completed the Bilingual Input Output Survey (BIOS) by phone to describe children's English and Spanish exposure and use. Parents reported on the child's language exposure and use of each language on an hourly basis during a typical week and weekend day. Given their high correlation (>.9), exposure and use were averaged, and information is projected for a 7-day week to yield the percentage use for each language.

Morphosyntactic and semantic knowledge and risk for DLD were measured using the Bilingual English Spanish Oral Screener for first graders (BESOS, Peña et al., 2010a). The BESOS includes semantics and morphosyntax subtests in Spanish and English. First-grade items were selected from a larger set of experimental items from the Bilingual English Spanish Assessment Middle Elementary (BESA-ME, Peña et al., 2010b) that were most sensitive to language disorder and development. The semantics subtests include expressive and receptive items (14 English and 16 Spanish items) that require children to retrieve and organize lexical information. Children can respond in Spanish or English on expressive items. Morphosyntax items (18 per language) tapped English and Spanish grammatical forms typically acquired by school age in monolingual speakers and English language learners (Peña et al., 2010). Normative data on the BESOS includes 349 typically-developing first graders. Preliminary data indicate an alpha level of .840 for English Morphosyntax; .890 for Spanish Morphosyntax; .648 for English Semantics; and .816 for Spanish Semantics. Preliminary classification analysis for 130 children (96 with typical development and 34 with DLD) indicates that a combination of the higher (between Spanish and English) semantics and higher morphosyntax scores has 91.1% sensitivity and 82.3% specificity.

Risk for reading difficulty was measured using the letter-word identification subtests (English and Spanish forms) of the Woodcock-Muñoz Language Survey-Revised (WMLS-R; Woodcock et al., 2005). The letter-word identification subtest includes a graded list of 75 items per language, starting with the upper and lowercase letters followed by words of increasing complexity. The test is sensitive to early reading difficulties in school-aged Spanish-English bilinguals (Vaughn et al., 2006), and the letter identification performance of kindergarteners is one of the best indicators of subsequent reading achievement (e.g. Catts et al., 2001). The test manual indicates high test-retest reliability (.92). Standard scores were calculated using the Woodcock-Muñoz Language Survey-Revised Normative Update software (WMLS-R NU; Schrank et al., 2010).

Procedure

Test Administration and Scoring. Children completed all four BESOS subtests and the English and Spanish letter-word identification subtests over one or two sessions. For the BESOS, a ceiling of five no-responses was employed per subtest to minimize frustration over testing in a language in which a child had minimal competence. For the WMLS-R, standard protocols were followed. The order of language screening was randomized, though all single-language tests were administered during a single block. Screening in both languages took approximately 20−30 min total. Screening took place in a quiet classroom at children's schools by trained bilingual research assistants or licensed bilingual speech-language pathologists. All testing took place at the beginning of the school year and was completed by early October.

The BESOS was scored in real-time by the examiner based on a list of acceptable and unacceptable responses on the test form. Item-level responses were entered into Excel to calculate a total raw score, and standard scores were then generated for each subtest. Responses for the WMLS-R letter-word identification test were scored following standard protocols: raw scores were entered into the WMLS-R NU proprietary software, which generated standard scores; these were then entered into an Excel spreadsheet. All responses to the BIOS were entered into an Excel spreadsheet, which was programmed to weight input and output by day of the week (e.g. school or weekday versus weekend day). Calculations generated a measure of percentage of exposure to and use of each language in a week-long period. We identified the year of first exposure to English based on reported yearly exposure to each language. All data were imported into SPSS version 26 for statistical analysis.

Determination of Risk. Children scoring in the 25th percentile or below on either the semantics or morphosyntax subtest of the BESOS in their higher scoring or better language for each subtest were considered to be at-risk for DLD (Peña et al., 2011). Children who scored in the 25th percentile or below on their better language on the letter-word identification task were considered to be at-risk for poor reading outcomes (Vaughn et al., 2003).

Results

On the letter-word identification test, children's mean standard score was 94.95 (SD = 17.0) in English and 111.41 (SD = 23.5) in Spanish, suggesting greater overall dominance, and slightly more variability, in Spanish. Their BESOS syntax and semantics scores showed similar advantages for Spanish, the L1 and dominant language of the majority of the group. For semantics, the mean score was 89.47 (SD = 16.4) for English and 94.66 (SD = 17.9) for Spanish. For morphosyntax, the mean standard score was 83.89 (SD = 25.1) for English, and 95.98 (SD = 19.8) for Spanish. (Table 1 displays the range of scores).

Table 1.
Descriptives and correlations for participants’ age, language experience, BESOS, and letter-word reading standard scores.

Mean SD Range 2 3 4 5 6 7 8 9

1 Age (mos) 82.59 5.27 (75–105) .19* 0.09 −.24* −.24* −0.00 −0.06 −.28 −.39

2 AOE 2.77 2.01 (0–6) ∼ −.19* 0.15 −0.13 .26** −0.00 0.06 −0.12

3 Eng use 0.40 0.13 (0.2–0.76) − ∼ -.16* .42 −0.08 .39 −.20* 0.19

4 Sp Sem 94.66 17.86 (49.33–134.44) − − ∼ .49 .69 .49 .35 .45

5 Eng Sem 89.47 16.43 (30–129.38) − − − ∼ .36 .69 0.12 .54

6 Sp MS 95.99 19.80 (42.3–123.31) − − − − ∼ .55 .36 .34**

7 Eng MS 83.89 25.12 (34.84–119.98) − − − − − ∼ .21* .5

8 Sp LW 112.02 22.00 (69–169) − − − − − − ∼ .38

9 Eng LW 94.95 17.03 (53–155) − − − − − − − ∼

Note. AOE = age of first English exposure; Eng use = percentage of English use at home and school; Sp Sem = standard score BESOS semantics in Spanish; Eng Sem = standard score BESOS semantics in English; Sp MS =standard score BESOS morphosyntax in Spanish; Eng MS = standard score BESOS morphosyntax in English; Sp LW = standard score Woodcock-Muñoz Language Survey Letter-Word in Spanish; Eng LW = standard score Woodcock-Muñoz Language Survey Letter-Word in English.

Within and cross-language correlations between language use, reading, and oral language

To understand the relationships between language use, language, and reading performance, we examined within- and across-language correlations between reading, semantics, and morphosyntax scores (see Table 1). English letter-word scores were highly correlated with morphosyntax and semantics scores within and across languages. Generally, within-language correlations were stronger than cross-language correlations. The strongest correlation with English letter-word scores was found with English semantics (r = .54, p < .0001), followed by English morphosyntax (r = .50, p < .0001), Spanish semantics (r = .45, p < .0001) and finally Spanish morphosyntax (r = .34, p < .0001). Percentage of current English use did not correlate with English letter-word scores (.19, p = NS). Spanish letter-word scores were significantly correlated with Spanish morphosyntax (r = .36, p < .0001), and semantics (r = .35, p < .0001) as well as English morphosyntax (r = .21, p = 0.025); however Spanish letter-word and English semantics scores were not correlated. Percentage of English use was negatively associated with Spanish letter-word reading (−.20, p < .0001). Recall that English use is the inverse of Spanish use.

Oral language predictors of reading performance within each language

Our first question addressed variance explained in children's reading performance in each of their languages. Predictor variables included performance on oral language measures (morphosyntax and semantics) and language experience as indexed by percentage of English use. We ran one stepwise regression analysis per language to determine which combination of language and experience measures best predicted letter-word identification performance. Results are reported in Tables 2 and 3. For English letter-word performance, the overall model was significant at R²= .336, F(1, 109) = 27.307, p < .001. English and Spanish semantic knowledge together accounted for 28.5% of the variance. Neither Spanish nor English morphosyntax nor current language use contributed to the variance accounted for in the model. For Spanish letter-word performance, the overall model was also significant at R² = .161, F (1, 109) = 16.248, p < .001. Here, Spanish morphosyntax scores accounted for a significant amount of the variance (13%), and inclusion of language experience, again indexed by English use, accounted for an additional 3.1% of the variance. English morphosyntax and semantics scores were eliminated in the stepwise model as they did not contribute significantly to the variance accounted for in the model.

Table 2.
Predictors of English letter-word identification.

Model R R ² Adj R² SE of the estimate Change statistics

R² change F change df1 df2 p

1—Eng Sem .540 .291 .285 14.4069 .291 44.757 1 109 .000

2—Eng Sem + Sp Sem .580 .336 .324 14.0091 .045 7.278 1 108 .008

Note. Eng Sem = English semantics; Sp Sem = Spanish semantics.

Table 3.
Predictors of Spanish letter-word identification.

Model R R ² Adj R² SE of the estimate Change statistics

R² change F change df1 df2 p

1—Sp MS .360 .130 .122 20.6185 .130 16.248 1 109 <.001

2—Sp MS + Eng use .401 .161 .145 20.3406 .031 3.999 1 108 .048

Note. Sp MS = Spanish morphosyntax; Eng use = percentage of English use.

Overlap of reading and language risk

Our second question addressed the relationship between language and reading risk. Language risk was defined by scores below the 25th percentile on the BESOS Semantic and Morphosyntax subtests based on the child's best performance across their two languages. Risk for future reading difficulties was defined as performance below the 25th percentile in the better language, just as for language risk. We used logistic regression to evaluate the odds of reading risk based on language scores. Low semantics performance was associated (B = −2.054) with increased reading risk, but morphosyntax scores were not. Because there were only two predictors in the model, we entered them in two steps and repeated the analyses entering the predictors in reverse order. There were no differences in the odds associated with the variables in the two iterations of the analysis, so we report only the first iteration. These results are shown in Table 4.

Table 4.
Association of language and reading risk based on better language scores.

B SE Wald df p Exp(B)

MS risk .468 .895 .273 1 .601 1.597

Sem risk −2.054 .918 5.011 1 .025 .128

Constant −1.612 .548 8.666 1 .003 .200

Note. MS risk = >25th percentile in the better language in morphosyntax; Sem risk = >25th percentile in the better language in semantics. Reading risk = >25th percentile in the better language in letter-word identification.

Risk for language and reading impairments

Our second research question investigated the degree of overlap between language and reading risk in our population. A summary of this overlap is shown in Table 5. Of the 111 children included in the present study, 39 showed risk for DLD based on scoring below the 25th percentile on either the morphosyntax or semantics subtest in their strongest language, for a risk rate of 35%, which was in line with our expectations based on the design of the BESOS oral language screener. In contrast, only eight children in our sample were identified as having reading risk based on their letter-word performance, for a rate of 7%, which is less than the expected 25% rate. While the majority of the children showing reading risk were also identified as having risk for DLD (6 of the 8), a large number of the children flagged as having risk for DLD performed in the normal range on the letter-word test in at least one of their two languages. This result was surprising given the documented high rate of overlap between oral language and reading risk reported in the literature on DLD (Snowling et al., 2016).

Table 5.
Overlap between children with reading risk and DLD risk based on at least one linguistic domain (morphosyntax or semantics) in child's better language.

Reading risk Total

No risk risk

DLD risk No risk 70 2 72

Risk 33 6 39

Total 103 8 111

Note. DLD = developmental language disorder. Italicized values represents totals by category. Total number of children with language and reading risk is bolded.

Discussion

Cross-language relationships often occur between early reading development and oral proficiency in bilinguals, especially when the two languages are similar in terms of their writing systems (Bialystok, 2007; Prevoo et al., 2016); however, a relatively small range of oral skills has been studied, with the bulk of research focusing on phonological awareness and single-word vocabulary. Here we sought to evaluate lexical quality using semantics and morphosyntax measures—in relation to letter-word knowledge, a highly predictive early reading measure. We examined correlations and predictive relationships within and across languages. We then turned to the degree of overlap in DLD and reading risk in our population of bilingual first graders when evaluating risk via performance in their strongest language in each domain.

Relationships between oral language and reading skills

We take as the foundation for our current study the results of Prevoo et al.'s (2016) meta-analysis that provides evidence for robust relationships between oral language and reading skills both within and across languages among bilingual immigrant children. A gap in the studies included in the meta-analysis was that they overwhelmingly operationalized oral language using either single-word vocabulary measures or general oral language proficiency tests rather than measures that tap lexical quality. Here, we focused on semantic and morphosyntactic measures that contribute to lexical quality. We asked if measures of semantics and morphosyntax were associated with letter-word identification in first grade and what the association was between language and reading risk in bilingual first graders. Thus, our findings extend our understanding of the relationship between oral language and early reading both within and across languages to semantic depth and morphosyntax in several important ways.

First, we replicated the robust correlations among oral language and reading measures within languages using measures of semantic depth and morphosyntax (Prevoo et al., 2016). These correlations were more robust in English than in Spanish. One possible reason for this is that children's early strengths in Spanish help them leverage their acquisition of English (see Bedore et al., 2020, for further discussion). For example, knowledge of Spanish vocabulary seems to support English vocabulary, which in turn supports the learning of English morphosyntax and prereading skills. In addition, children's exposure to language and reading in English may be more similar than their Spanish experiences because much English exposure occurs in school. Second, we found moderate to strong cross-language correlations in letter-word identification and morphosyntactic and semantic measures. These findings converge with robust cross-language correlations found in phonemic awareness and morphological awareness, vocabulary, and reading comprehension strategies (Chung et al., 2018; Prevoo et al., 2016). Further, they support the linguistic interdependence hypothesis (Cummins, 1979), which posits that second language abilities are dependent upon first language competence, and the view that common cognitive components underlie language skills in both languages (Geva and Ryan, 1993). This is evident in the finding that Spanish semantic performance contributed significantly to children's English letter-word performance. Semantic tasks tap generalized knowledge of concepts that can support the development of both languages rather than meaning of language specific vocabulary items.

Stepwise regression analysis results suggest that morphosyntax and semantic knowledge play a role in predicting letter and word reading performance but that the relationship differs for English and Spanish. For Spanish, but not English, current use makes a difference. Spanish morphosyntax scores accounted for most of the variance in Spanish letter-word identification though language experience adds slightly to the variance accounted for in the model. For English semantic knowledge in both languages accounts for most of the variance in letter-word scores. Here, current language use does not increase the variance accounted for by the model.

Spanish-English differences may relate both to different contexts of acquisition for each language at home and school and to differential outcomes. Recall that children had substantial exposure to English in school through dual language instruction and systematic transition to English in the school programs in which they were enrolled. Given the nature of school based language exposure, the quantity and quality of input may have been more homogeneous because it is tied to educational lesson plans and the opportunities to progress in both oral and literate domains are more closely linked. In contrast, because Spanish was first acquired in the home, there was likely more variability between families in the degree of exposure to both oral and written language, including perhaps parental literacy for joint book-reading behavior. These may impact Spanish reading knowledge and mitigate the influence of oral linguistic proficiency. Another source of the differences may lie in the orthographic transparency of Spanish and English. At this early stage of reading, it may be that oral linguistic proficiency plays more of a role in the development of English reading than Spanish reading because learning to decode in a highly transparent orthography (like Spanish) may not require as fine-grained phonological and lexical representations (Ziegler and Goswami, 2005). Another way that school experience may affect these outcomes is that instruction is needed for decoding but not for early oral language development. This is especially apparent in the English model, where it is word knowledge that is most associated with letter-word performance, and English experience does not contribute to the model. Recall, however, that English experience is moderately associated with English semantic knowledge. In contrast, Spanish letter-word decoding is more accurate.

Risk for language and reading impairments

The relatively low rate of reading risk observed in our population (see Table 5) may stem from crucial differences between our population and the sample used in the development of the WMLS-R NU, which was selected to be representative of the US population as a whole (Schrank et al., 2010). As a group, our participants scored higher than expected: in Spanish, their mean standard score of 112 on the letter-word test put them in the ‘high average’ range, while their English standard score of 95 placed them squarely in the average range, despite their reduced exposure to each language stemming from divided input. This may have been due to the greater likelihood of children in our sample to have participated in preschool, compared to the national average, which is around 64% of the 3–5-year-old population according to the 2017 Current Population Survey (US Census Bureau, 2017). In the current sample, 81.3% of the children had attended preschool. This experience likely gave our participants better reading outcomes than might otherwise have been expected in first grade, as they would have had several extra years of schooling in which prereading skills like letter recognition are often taught. Prior research with ELLs has found that preschool (Head Start) enrollment can make a crucial difference in early reading development (Hammer et al., 2007). It is also worth noting that testing took place early in the school year but not before reading instruction commenced for the year. The participants all received reading in Spanish which included sight word practice and phonics activities (e.g. sound blending at the syllable level and sound letter identification) that provide exposure to the kinds of tasks tapped in the Woodcock Johnson (see Texas Knowledge and Skills, 2020). Early exposure to these skills provides a reading foundation that helps children make gains in Spanish. Given that all children had the same amount of exposure we expect this had minimal impact on the presence of reading risk.

Another possible explanation for the lower than expected rate of reading risk within our sample might be our methods of determining risk – using the letter-word reading measure and making decisions about low performance in both languages (rather than English alone). While letter-word identification has been successfully used as an identifier of reading risk, it must be recognized that Spanish's orthographic transparency may make it a less robust indicator of reading risk than its English counterpart. It is now widely accepted that diagnosing DLD in bilinguals must take into account language dominance and consider performance in a child's strongest language, so that true risk of impairment is not confused with limited proficiency due to exposure (Bedore and Peña, 2008); however, this practice may not be as widespread in the field of reading, where poor performance in English alone may be sufficient to warrant special services in schools or to classify bilingual children has having reading risk for research purposes (e.g. Swanson et al., 2006). Future research should evaluate if focusing on better language scores provide a better estimate of true reading risk. Evaluating language and reading in larger groups of bilingual children with more extensive language and reading measures could help disentangle sources of language and reading risk that can be addressed in classrooms.

Implications

Our study contributes valuable data to an underexplored research area, showing within- and cross-linguistic relationships between oral language and reading skills in developing bilinguals. Our discovery that Spanish morphosyntactic awareness and semantic ability are strongly related to English letter-word performance adds further support to the linguistic interdependence hypothesis (Cummins, 1979) and prior research reporting better educational achievement in English for ELL's with stronger skills in Spanish (Cobo-Lewis et al., 2002). Indeed, our participants’ relatively high levels of letter-word performance in both their L1 and L2 English show that bilingual children, though at higher risk of poor reading outcomes based on national achievement test data, may well exceed expectations when given sufficient support and education in both languages from early ages. This is valuable information for clinicians and educators who are working with bilingual children and families to establish educational plans and interventions. These findings suggest that providing continued support for the child's first language development can positively contribute to early reading skills such as letter-word reading.

		Mean	SD	Range	2	3	4	5	6	7	8	9
1	Age (mos)	82.59	5.27	(75–105)	.19*	0.09	−.24*	−.24*	−0.00	−0.06	−.28**	−.39**
2	AOE	2.77	2.01	(0–6)	∼	−.19*	0.15	−0.13	.26**	−0.00	0.06	−0.12
3	Eng use	0.40	0.13	(0.2–0.76)	−	∼	-.16*	.42**	−0.08	.39**	−.20*	0.19
4	Sp Sem	94.66	17.86	(49.33–134.44)	−	−	∼	.49**	.69**	.49**	.35**	.45**
5	Eng Sem	89.47	16.43	(30–129.38)	−	−	−	∼	.36**	.69**	0.12	.54**
6	Sp MS	95.99	19.80	(42.3–123.31)	−	−	−	−	∼	.55**	.36**	.34**
7	Eng MS	83.89	25.12	(34.84–119.98)	−	−	−	−	−	∼	.21*	.5**
8	Sp LW	112.02	22.00	(69–169)	−	−	−	−	−	−	∼	.38**
9	Eng LW	94.95	17.03	(53–155)	−	−	−	−	−	−	−	∼

Model	R	R ²	Adj R²	SE of the estimate	Change statistics
1—Eng Sem	.540	.291	.285	14.4069	.291	44.757	1	109	.000
2—Eng Sem + Sp Sem	.580	.336	.324	14.0091	.045	7.278	1	108	.008

Model	R	R ²	Adj R²	SE of the estimate	Change statistics
1—Sp MS	.360	.130	.122	20.6185	.130	16.248	1	109	<.001
2—Sp MS + Eng use	.401	.161	.145	20.3406	.031	3.999	1	108	.048

	B	SE	Wald	df	p	Exp(B)
MS risk	.468	.895	.273	1	.601	1.597
Sem risk	−2.054	.918	5.011	1	.025	.128
Constant	−1.612	.548	8.666	1	.003	.200

	Reading risk	Total
DLD risk	No risk	70	2	72
	Risk	33	6	39
Total	103	8	111

Footnotes

Acknowledgements

The authors thank all of the interviewers and testers for their assistance with collecting the data and the school districts for allowing us access to the participants.

Declaration of conflicting interests

Drs. Bedore and Peña are co-authors of the Bilingual English Spanish Assessment which includes the Bilingual Input Output Scale (BIOS). They receive royalities for its sale.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by National Institute on Deafness and Other Communication Disorders Grant 1R21DC011126-01 to Lisa M. Bedore.

ORCID iDs

Lisa M Bedore

Elizabeth D Peña

References

Apel

(2014) A comprehensive definition of morphological awareness. Topics In Language Disorders 34(3): 197–209.

Apel

Diehm

(2014) Morphological awareness intervention with kindergartners and first and second grade students from low SES homes: a small efficacy study. Journal of Learning Disabilities 47(1): 65–75.

Bedore LM and Peña ED (2008) Assessment of bilingual children for identification of language impairment: current findings and implications for practice. International Journal Of Bilingual Education And Bilingualism 11(1): 1–29.

Bedore LM, Peña ED, Fiestas CW, et al. (2020) Language and literacy together: supporting grammatical development in dual language learners with risk for language and learning difficulties. Language Speech and Hearing in the Schools, 51 (2):282–229.

Bedore LM, Peña ED, Gillam

, et al. (2010) Language sample measures and language ability in Spanish-English bilingual kindergarteners. Journal of Communication Disorders 43(6): 498–510.

Bedore LM and Leonard

(2001) Grammatical morphology deficits in Spanish-speaking children with specific language impairment. Journal of Speech, Language, and Hearing Research 44(4): 905–924.

Bialystok

(2007) Cognitive effects of bilingualism: how linguistic experience leads to cognitive change. International Journal Of Bilingual Education And Bilingualism 10(3): 210–223.

Bishop

DVM

Snowling

(2004) Developmental dyslexia and specific language impairment: same or different? Psychological Bulletin 130(6): 858–886.

Catts

Fey

Zhang

, et al. (2001) Estimating the risk of future reading difficulties in kindergarten children: a research-based model and its clinical implementation. Language, Speech, & Hearing Services in Schools 32(1): 38.

10.

Census Bureau (2017) Current Population Survey (CPS). Washington DC: US Department of Commerce.

11.

Chung

Chen

Geva

(2018) Deconstructing and reconstructing cross-language transfer in bilingual reading development: an interactive framework. Journal Of Neurolinguistics 50: 149–161.

12.

Cobo-Lewis

Pearson

Eilers

, et al. (2002) Effects of bilingualism and bilingual education on oral and written Spanish skills: a multifactor study of standardized test outcomes. In: Oller

Eilers

(Eds) Language and Literacy in Bilingual Children. Bristol, Blue Ridge Summit: Multilingual Matters, 98–117.

13.

Conboy

Thal

(2006) Ties between the lexicon and grammar: cross-sectional and longitudinal studies of bilingual toddlers. Child Development 77(3): 712–735.

14.

Cummins

(1979) Cognitive/academic language proficiency, linguistic interdependence, the optimum age question and some other matters. Working Papers on Bilingualism Toronto 19: 197–202.

15.

Dijkstra

TON

Wahl

Buytenhuijs

, et al. (2019) Modelling bilingual lexical processing: a research agenda and desiderabilia. Bilingualism: Language and Cognition 22(04): 703–713.

16.

Ehrhorn

Adlof

Fogerty

, et al. (2021) Probing phonological processing differences in nonword repetition for children with separate or co-occurring dyslexia and developmental language disorder. Scientific Studies of Reading 25(6): 486–503.

17.

Geva

Ryan

(1993) Linguistic and cognitive correlates of academic skills in first and second languages. Language Learning 43(1): 5–42.

18.

Goodrich

Namkung

(2019) Correlates of reading comprehension and word-problem solving skills of Spanish-speaking dual language learners. Early Childhood Research Quarterly 48: 256–266.

19.

Goodrich

Fitton

Thayer

(2022) Relations between oral language skills and English reading achievement among Spanish–English bilingual children: a quantile regression analysis. Annals of Dyslexia. 1–23. https://doi.org/10.1007/s11881-022-00257-1

20.

Hammer

Lawrence

Miccio

(2007) Bilingual children's language abilities and early reading outcomes in head start and kindergarten. Language, Speech, and Hearing Services in Schools 38(3): 237–248.

21.

Hoover

Gough

(1990) The simple view of reading. Reading And Writing 2(2): 127–160.

22.

Huang

, Bedore LM, Ramírez

, et al. (2022) Contributions of oral narrative skills to English reading in Spanish–English Latino/a dual language learners. Journal of Speech, Language, and Hearing Research 65(2): 653–671.

23.

Language and Reading Research Consortium, Murphy

, et al. (2016) Investigating profiles of lexical quality in preschool and their contribution to first grade reading. Reading And Writing 29(9): 1745–1770.

24.

Leonard

(2014) Children With Specific Language Impairment (Language, Speech, And Communication). Cambridge, Massachusetts: A Bradford Book.

25.

Lucero

(2015) Cross-Linguistic lexical, grammatical, and discourse performance on oral narrative retells among young Spanish speakers. Child Development 86(5): 1419–1433.

26.

Mancilla-Martinez

Lesaux

(2011) The gap between Spanish speakers’ word reading and word knowledge: a longitudinal study. Child Development 82(5): 1544–1560.

27.

McGregor

Oleson

Bahnsen

, et al. (2013) Children with developmental language impairment have vocabulary deficits characterized by limited breadth and depth. International Journal of Language & Communication Disorders/Royal College of Speech & Language Therapists 48(3): 307–319.

28.

Muter

Hulme

Snowling

, et al. (2004). Phonemes, rimes, vocabulary, and grammatical skills as foundations of early reading development: evidence from a longitudinal study. Developmental Psychology 40(5): 665.

29.

National Assessment of Educational Progress (2019) National Achievement-Level Results. Washington DC: nationsreportcard.gov.

30.

Newbury

Paracchini

Scerri

, et al. (2011) Investigation of dyslexia and SLI risk variants in reading- and language-impaired subjects. Behavior Genetics 41(1): 90–104.

31.

Peña, ED, Bedore, LM, Gutiérrez-Clellen

, et al. (2010a) Bilingual English Spanish Oral Screener (BESOS) .

32.

Peña ED, Bedore LM, Gutiérrez-Clellen

, et al. (2010b) Bilingual English Spanish Assessment - Middle Extension (BESA-ME) .

33.

Peña ED, Bedore LM and Kester

(2016) Assessment of language impairment in bilingual children using semantic tasks: two languages classify better than one. International Journal Of Language & Communication Disorders/Royal College of Speech & Language Therapists 51(2): 192–202.

34.

Peña ED, Gillam

, Bedore LM, et al. (2011) Risk for poor performance on a language screening measure for bilingual preschoolers and kindergarteners. American Journal Of Speech-Language Pathology 20(4): 302–314.

35.

Peña ED, Gutiérrez-Clellen

, Iglesias A, et al. (2018) Bilingual English Spanish Assessment (BESA). Baltimore, MD: Brookes.

36.

Perfetti, C (2007) Reading ability: Lexical quality to comprehension. Scientific Studies of Reading 11(4): 357–383.

37.

Perfetti

Hart

(2001) The lexical basis of comprehension skill. In: Gorfein

(ed) On the Consequences of Meaning Selection: Perspectives on Resolving Lexical Ambiguity. Washington: American Psychological Association, 67–86.

38.

Prevoo

MJL

Malda

Mesman

, et al. (2016) Within- and cross-language relations between oral language proficiency and school outcomes in bilingual children with an immigrant background: a meta-analytical study. Review of Educational Research 86(1): 237–276.

39.

Proctor

Silverman

Harring

, et al. (2012) The role of vocabulary depth in predicting reading comprehension among English monolingual and Spanish–English bilingual children in elementary school. Reading And Writing 25(7): 1635–1664.

40.

Ramirez

Chen

Geva

, et al. (2010) Morphological awareness in Spanish-speaking English language learners: within and cross-language effects on word reading. Reading & Writing: An Interdisciplinary Journal 23: 337–358.

41.

Schleicher

(2019) PISA 2018 Insights and Interpretations. Paris: Organization for Economic Cooperation and Development.

42.

Schrank FA and Woodcock RW (2010) Wood-Muñoz Language Survey-Revised Normative Update Scoring and Reporting Program. In Riverside Publishing.

43.

Leikin

(2004). Language impairment at school entry and later reading disability: connections at lexical versus supralexical levels of reading. Scientific Studies of Reading 8(1): 87–110.

44.

Sheng

McGregor

(2010) Lexical–semantic organization in children with specific language impairment. Journal of Speech, Langauge, and Hearing Research 53: 146–159.

45.

Sheng L, Peña ED, Bedore LM and Fiestas CE (2012) Semantic deficits in Spanish-English bilingual children with language impairment.

46.

Shivabasappa

, Peña ED, and Bedore LM (2018) Core vocabulary in the narratives of bilingual children with and without language impairment. International Journal of Speech-Language Pathology 20(7): 790–801.

47.

Simon-Cereijido

Gutiérrez-Clellen

Sweet

(2013) Predictors of growth or attrition of the first language in latino children with specific language impairment. Applied Psycholinguistics 34(6): 1219–1243.

48.

Snowling

Hulme

(2021) Annual research review: reading disorders revisited–the critical importance of oral language. Journal of Child Psychology and Psychiatry 62(5): 635–653.

49.

Snowling

Duff

Nash

, et al. (2016) Language profiles and literacy outcomes of children with resolving, emerging, or persisting language impairments. Journal of Child Psychology and Psychiatry, and Allied Disciplines 57(12): 1360–1369.

50.

Swanson

Sáez

Gerber

(2006) Growth in literacy and cognition in bilingual children at risk or not at risk for reading disabilities. Journal Of Educational Psychology 98(2): 247–264.

51.

Texas Education Agency (2020) Texas Essential Knowledge and Skills for Grade 1 . (accessed 11-23-2020).

52.

Vaughn

Cirino

Linan-Thompson

, et al. (2006) Effectiveness of a Spanish intervention and an English intervention for English-language learners at risk for reading problems. American Educational Research Journal 43(3): 449–479.

53.

Vaughn

Linan-Thompson

Hickman

(2003) Response to treatment as a means for identifying students with reading/learning disabilities. Exceptional Children 69(4): 391–410.

54.

Verhoeven

Perfetti

(2017) Learning To Read Across Languages And Writing Systems. Cambridge, UK: Cambridge University Press.

55.

Woodcock

Muñoz-Sandoval

Ruef

, et al. (2005) Woodcock-Muñoz Language Survey-Revised. Itasca, IL: Riverside.

56.

Zhang

Koda

(2014) Awareness of derivation and compounding in Chinese–English biliteracy acquisition. International Journal Of Bilingual Education And Bilingualism 17(1): 55–73.

57.

Ziegler

Goswami

(2005) Reading acquisition, developmental dyslexia, and skilled reading across languages: a psycholinguistic grain size theory. Psychological Bulletin 131(1): 3–29.