Association for Public Policy Analysis & Management

This study examines the effects of translating assessments from English into the student’s native language, an understudied accommodation. English language learning (ELL) students represent the fastest growing segment of the U.S. student population. In 2015, 4.8 million children were ELL students, and most ELL children were enrolled in elementary school grades. More than 77 percent of ELL students spoke Spanish at home (NCES, 2018). Because most U.S. schools instruct only in English, many Spanish-speaking students struggle to learn once they are in school due to limited English proficiency. One of the factors responsible for test performance is the language-based accommodations making assessments linguistically accessible to ELL students (Salend 2008, Abedi 2014).

This study uses data from the Early Childhood Longitudinal Study, Kindergarten (ECLS-K: 2010), a nationally representative dataset of more than 15,000 children who attended kindergarten in 2010. In this study, we specifically examine a subsample of children whose home language is Spanish (n=2,400) because those were the only ones who could be assessed in Spanish or English. In kindergarten, all children took a language screener test, the preLAS. Children living in Spanish speaking households and who were below the cutoff point in the preLAS test were administered math, reading and executive function assessments in Spanish, while children who were at or above the cutoff point, or whose households did not speak a foreign language, were administered the test in English. I used a regression discontinuity design, which resembles a randomized experiment because children who were just below the cutoff point and who were administered the tests in Spanish would be similar to children who were just above the cutoff point and who were administered the tests in English. Around the vicinity of the cutoff point, this method allows to answer the causal question of whether translation accommodations for young ELLs have an effect on test scores.

Our findings show that in the spring of kindergarten, students tested in Spanish performed significantly better on mathematics, cognitive functioning, and working memory assessments than did students tested in English. We did not find effects for the fall of kindergarten though. Together, these results suggests that accommodations are effective when children have to answer linguistically challenging and difficult questions, and that occurs in the spring, after children were taught something new. Our results are robust to different specifications and bandwidths.

Under the Every Student Succeeds Act (ESSA), signed into law in 2015, states receive additional support for ELL children. While some states provide language accommodation for ELL children, there is a lot of variation in the type of accommodation as well as the subjects. According to Tabaku, Carbuccia-Abbott & Saavedra (2018), not all states offer standardized assessments in languages other than English. For example, only five states offer language arts in Spanish, twelve states offer science in Spanish, four states offer social science in Spanish, and thirty-one states offer mathematics in Spanish. Rethinking state policies for ELL children is timely as ELL children represent 10% of all children attending public schools.