Examining Heterogeneity Among Latino Dual Language Learners' School Readiness Profiles at the End of Prekindergarten

• In Event: Dual Language Learners’ school readiness skills, classroom experiences, and parental influences in early childhood settings

Abstract

Emerging research shows there is substantial heterogeneity in the English and Spanish language and literacy proficiencies of dual language learners (DLLs) in U.S. preschools (NASEM, 2017). DLLs comprise approximately 23% of the preschool population in the U.S. The present study uses latent profile analysis to examine the heterogeneity of Spanish-English DLLs. While previous studies (Kim et al., 2018; Lambert et al., 2017; Lonigan et al. 2018) have focused on language, language and literacy, or language and mathematics, the present study identifies the varying profiles of DLLs' cognitive, linguistic, literacy, and mathematics skills at preschool. Including multiple measures of school readiness across both languages provides a comprehensive look at the school readiness abilities of DLLs.

We examined within-group variability in 320 Spanish-English speaking DLLs' cognitive, linguistic, literacy, and mathematics skills at the end of prekindergarten (M = 5.22 years old). All DLLs in the present study were Latino (53.7% female) and each DLL spent at least one year in a Head Start classroom. All families reported speaking Spanish as the primary language in their home. Approximately half of the families were from Mexico (52.8%), Cuba represented the second largest group (14.2%). Children in the present study were assessed individually in English and Spanish on 18 subtests (9 per language) of the Woodcock-Johnson III/ Batería 3 during the spring prior to entry into kindergarten. WJIII subtests included Spatial Relations, Visual Matching, Picture Recognition, Picture Vocabulary, Oral Comprehension, Spelling, Letter-Word Identification, Quantitative Concepts, and Applied Problems. Given that classroom characteristics are associated with DLLs' emerging cognitive, linguistic, and academic skills, classroom covariates, as measured by the CLASS (i.e., emotional support, classroom organization, and instructional support), are controlled. To empirically identify profiles of cognitive, linguistic, literacy, and mathematics skills, a series of mixture models were estimated in Mplus (Version 8.2), specifically using latent profile analysis. All latent profile analyses utilized W scores. All models were estimated using full information maximum likelihood estimation (FIML) with robust standard errors using the expectation-maximization (EM) procedure.

The results of latent profile analyses are reported in Table 1. Four profiles of DLLs are identified. Most DLLs were classified as English Dominant (38.75%; i.e., stronger skills in English than Spanish on average), followed by Balanced Average (30.63%; i.e., average skills are above normal in both English and Spanish), Spanish Dominant (17.81%; i.e., stronger skills in Spanish than English on average), and Balanced Low (13.12%; i.e., average skills are below normal in both English and Spanish), respectively.

In general, the Balanced Average profile outperformed the other profiles in English and Spanish, and their norm-referenced standard scores provide additional evidence that bilingual development is not associated with educational risk. Additional comparisons of achievement by profile will be provided in the final paper. This information has important implications for curricular development and classroom instruction. The most critical implication stemming from this work is that one should not make assumptions about children's abilities based solely on the DLL label.

Authors

• Lisa Maria López, University of South Florida
  Presenting Author

• Matthew E Foster, University of South Florida
  Non-Presenting Author