How Crucial Is English Language Learner Support in Kindergarten and First Grade for Spanish-Speaking Students' Math Learning?

• In Event: 26.048 - Which Policies and Practices Promote Improved Outcomes for English Learners? Evidence From Large-Scale Quantitative Studies

Abstract

This study focuses on assessing the contribution of ELL services in kindergarten and first grade to the math learning of Spanish-speaking students throughout the elementary grades. ELL students, especially those from a Spanish-speaking background, lag in math achievement from kindergarten entry through the later school years (Fry 2007; Reardon & Galindo 2009). Although early math skills predict later achievement in reading and math, there has been scarce research evaluating the impact of ELL services on math learning. Researchers of second language acquisition have argued that it takes four or more years (Collier, 1987, 1989; Cummins, 1981) to reach the academic English proficiency necessary for learning in content areas (August & Hakuta, 1997). Yet in practice, the actual provision of ELL services varies across schools, which makes it possible to empirically examine the above argument against alternative positions. For example, one may reason that, once language minority students have acquired the system of basic mathematical symbols, the level of academic English proficiency required for math learning is perhaps different from that required for learning in other content areas.

We drew data from the nationally representative Early Childhood Longitudinal Study- Kindergarten cohort (ECLS-K). The students were followed from the beginning of kindergarten in 1998 through the end of fifth grade in 2004. We identified 2,205 students whose primary home language was Spanish. Many of these students displayed limited English proficiency at kindergarten entry. We have developed and applied a nonparametric marginal mean weighting through stratification (MMWS) strategy to remove selection bias associated with baseline and time-varying covariates including a student’s oral English proficiency, literacy skills, math skills, approaches to learning, family resources, and school characteristics. Our key identification assumption is that, given the past observed covariate history, treatment history, and outcome history, a student’s current treatment assignment is assumed independent of all the future potential outcomes. When this assumption holds, the weighted data approximate data from a sequential randomized experiment in which students of similar backgrounds are assigned at random to ELL services in kindergarten and again in first grade after having received the treatment in kindergarten.

Analyzing a weighted growth model, we have found that, despite the fact that students displaying lower math skills were more likely to be assigned to ELL services, at the end of each wave through fifth grade, Spanish-speaking students who received ELL services in kindergarten only stayed ahead in math achievement of those who were never treated by at least one standard deviation. The benefit of receiving two or more years of ELL services was considerably less and dissipated after first grade. Additional adjustments for potential remaining bias through sensitivity analysis corroborate these findings.

The results suggest the undeniable importance of providing language support to Spanish-speaking students when they start formal math instruction in kindergarten. We do not find value added by providing additional years of ELL services for these students’ math learning. Yet, the continuation of ELL services may enhance student learning in other content areas, a topic that we will investigate in future research.

Authors

• Guanglei Hong, University of Chicago

• Josh Gagne, University of Chicago