Growth Mixture Modeling of Bilingual Language Development and Literacy Outcomes

Growth Mixture Modeling of Bilingual Language Development and Literacy Outcomes

First Author: Jason Lon Anthony -- University of South Florida
Additional authors/chairs: 
Matthew Foster; Yi-Jui Iva Chen ; Jeffrey Williams
Keywords: Bilingualism, Vocabulary, Oral Language, Literacy development, Growth Modeling
Abstract / Summary: 

Purpose. This study identified subgroups of dual language learners (DLL) from English and Spanish growth trajectories and then compared their literacy outcomes. Monolingual English speakers (MES) from the same schools were included to contextualize the findings.
Method. At four equidistant time points across kindergarten, we administered the Expressive One Word Picture Vocabulary Test (EOWPVT) to 551 DLL and 562 MES and we administered a Spanish-only adaptation of EOWPVT Spanish-Bilingual Edition to DLL. At the end of kindergarten, we administered Word Attack, Letter Word Identification, and Passage Comprehension (WJ III) to all participants and like subtests from Woodcock Munoz Bateria III to DLL.
Results. Growth mixture modeling identified four latent growth profiles of DLL (Entropy= .90, VLMR p=.05, posterior probabilities= .89 to .97). Spanish Dominant profile demonstrated average proficiency and growth in Spanish but low proficiency and no growth in English. English Dominant profile had average proficiency and growth in English but low proficiency and no growth in Spanish. Balanced Average profile exhibited average proficiencies in both languages, which developed at average paces. Balanced Low profile had underdeveloped Spanish and English that failed to improve. English Dominant profile had English language trajectory and English literacy equivalent to most MES. Balanced Average group developed English at the same rate as MES, but their English proficiency and English literacy lagged slightly behind (Hedges g= .15 - .18). Spanish Dominant and Balanced Low profiles evidenced English vocabulary and English literacy far behind MES (g= .84 - 2.7). The Spanish Dominant and Balanced Average profiles achieved much higher Spanish literacy than remaining DLL profiles (g > 4.0; ps < .001). Implications. Substantial differences in acquisition of English and Spanish exist among DLL. These differences are predictive of Spanish and English literacy achievement.