Cognitive, Sociolinguistic, and Psychometric Perspectives in Science and Mathematics Assessment for English Language Learners

Wested, San Francisco CA

Investigators

Abstract

This two-year proposed project will investigate the cognitive and sociolinguistic influencesthat shape the performance of English language learners [ELLs] on science and mathematicsassessments. It will address the fact that performance of ELL students in science andmathematics is inconsistent both across test items and across languages. The knowledge gainedfrom this project will inform practitioners, researchers, and policy makers about new approachesfor attaining equitable testing and obtaining valid measures of the achievement of ELLs in science and mathematics. It will provide valuable information relevant to both classroom-basedtesting and the inclusion of linguistic minorities in large-scale testing. We will compare the inferred cognitive activities of ELL students, the problem solutionstrategies they use, and the scores they obtain on the same set of science and mathematics itemsadministered both in English and their native languages. In addition, we will investigate whether and how performance differences across languages and items are reduced when students are tested with locally-adapted versions of tests that reflect their first languages dialects. We will examine the performance of ELLs across languages and items from three perspectives: cognitive, psychometric, and sociolinguistic. These perspectives will allow us to: (1) determine how the language in which ELL students are tested influences their cognitive activities and the problem solution strategies they use on science and mathematics items; (2)determine whether ELL students perform better on local dialect than standard dialect versions of tests; and (3) examine the inconsistency of ELL student performance across items and across languages as an interaction of three sources of score variability: student, item, and language (i.e., a given ELL student performs better in his/her native language than in English for some items but better in English than in the native language for other items). We expect to improve our understanding of the cognitive and sociolinguistic processes that operate below the surface as ELL students interpret and solve problems. We also expect to gain knowledge on how these processes are reflected in the performance of these students and the variability of their scores across languages and items. This knowledge will contribute to improving both classroom-based and large-scale assessment practices. Goals Ultimately, our goal is to contribute to attaining more valid measures of ELL academic achievement and more equitable testing practices by offering a fresh approach to the testing of linguistic minorities. This approach recognizes the complex relationship between language and cognitive phenomena, and the social nature of language. From the knowledge we gain in this investigation we expect to be able to formalize a set of recommendations for improved testing policies for ELLs.