DLL Outcomes varies by Bilingual Education Model: Two-Way Immersion and Home Language Support

• In Event: 1-187 - Bilingual Education Program Models & Classroom Practices for DLLs: Processes and outcomes from preschool through elementary school

Integrative Statement

Dual Language Learners (DLLs) are a rapidly growing segment of the U.S. student population, making up almost 15% of the K-12 population and predicted to be 40% by 2030. A concern among U.S. educators and scholars is how best to reduce or close the achievement gap that persists between White economically advantaged students whose native language is English and DLLs who are more likely to be in poverty. Different “models” of bilingual education have been attempted over the years with much heated political and scholarly debate about which model works best, for whom, and even toward what desired end (i.e., how much to value the home language). Recent evidence tends to favor programs that teach content though some degree of home language instruction, with strong, cumulative support for two-way immersion (TWI) models – programs that teach core content areas in English and a partner language, and that integrate both traditional DLLs (language other than English at home) and native English-speaking children learning a second language.

Using data from the Miami School Readiness Project (MSRP), we followed a large sample of DLLs (N = 19,650 - 47% female, 80% in poverty (FRL); 86% Latino, 10% Black, 4% White/other) from pre-K through 5th grade. We control for many covariates (demographics, cognitive, language, motor, social, and behavior skills at age 4) associated with both academic performance and speed of English acquisition, and use student records to collect the ecologically valid outcomes of GPA, standardized math and reading test scores in 5th grade, and the grade at which each DLL attained English proficiency (exited ESOL). The independent/predictor variable of interest was the type(s) of language program in place at the DLL’s public elementary school in Miami (i.e., one-way immersion vs. TWI vs. sheltered ESL vs. mainstream ESL) with a focus on the amount of instruction provided (according to the models, not observed in the classroom) in the child's home language.

Multiple regressions controlling for important variables associated with speed of English acquisition (gender, poverty, ethnicity, cognitive, language, motor, social, and behavioral skills at age 4) showed that DLLs enrolled in two-way immersion learned English faster than those in other program models. Similar regression analyses predicting academic performance in 5th grade, controlling for the above covariates and speed of English acquisition, showed that DLLs who attended a school whose bilingual education models included some degree of home language instruction (50-90%) performed better on standardized math and reading test scores and had higher GPAs in 5th grade compared to those whose programs were limited to more traditional ESOL models (i.e., sheltered and mainstream English language content-based instruction). DLLs who went to TWI programs also had higher 5th grade GPAs than those in other program models and this effect was fully mediated by children’s faster English language acquisition. This study provides robust longitudinal evidence that bilingual models that support the use of DLL students’ home language in the classroom are effective in not only speeding up the acquisition of English but also enhancing achievement in the long term.

Authors

• Adam Winsler, George Mason University
  Presenting Author

• Ellen Serafini, George Mason University
  Non-Presenting Author

• Nadine Rozell, George Mason University
  Non-Presenting Author