A Latent Class IRT Approach to Defining and Measuring Language Proficiency

English language learner (EL) status has high stakes implications for determining when and how ELs should be evaluated for academic achievement. In the US, students designated as English learners are assessed annually for English language proficiency (ELP), a complex construct whose conceptualization has evolved in recent years to reflect more precisely the language demands of content area achievement as reflected in the standards of individual states and state language assessment consortia, such as WIDA and ELPA21. The goal of this paper was to examine the possible role for and utility of using content area assessments to validate language proficiency mastery criteria. Specifically, we applied mixture item response models to identify two classes of EL students: (1) ELs for whom English language arts and math achievement test items have similar difficulty and discrimination parameters as they do for non-ELs and (2) ELs for whom the test items function differently. We used latent class IRT methods to identify the two groups of ELs and to evaluate the effects of different subscales of ELP (reading, writing, listening, and speaking) on group membership. Only reading and writing were significant predictors of class membership. Cut-scores based on summary scores of ELP were imperfect predictors of class membership and indicated the need for finer differentiation within the top proficiency category. This study demonstrates the importance of linking definitions of ELP to the context for which ELP is used and suggests the possible value of psychometric analyses when language proficiency standards are linked to the language requirements for content area achievement. 1 A Latent Class IRT Approach to Defining and Measuring Language Proficiency Around the world, school and governmental systems are confronted with the challenge of measuring the language proficiency of students, employees, and citizens. This challenge has increased in recent years as globalization leads to increasing numbers of individuals who speak a language other than the societal language, but whose academic and economic fortunes are tied to their proficiency in the societal language. In the US, language proficiency assessments are most commonly used in public schools to assess the linguistic competencies of language minority students to ensure their ability to benefit from instruction conducted in English. Common descriptors used to characterize functional linguistic competence (i.e., ready to participate in regular instruction without linguistic support) focus on the ability to fluently interact with native speakers, to understand the main ideas (both concrete and abstract) when presented in complex texts and speech, and to produce complex written and oral arguments. Insofar as many native speakers of a language may struggle to achieve this level of linguistic competence because they lack declarative knowledge, or the ability to understand complex arguments involving abstract topics, measures of language proficiency can confound language competence with academic proficiency at higher levels of thinking. Put another way, there is marked variability in the verbal abilities of native speakers of a given language, and language proficiency assessments seek not to confound individual differences in verbal ability with individual differences in language proficiency. On the one hand, it is unrealistic to expect a person to display a level of language proficiency in a second language (L2) that they do not display in their first language (L1). Exceptions exist, of course, and certainly children acquiring L2 prior to full development of their L1 without continued emphasis on L1 development can achieve a level of competence in L2 that is unmatched in their L1. At the same time, there is clearly considerable variability in verbal ability among native speakers of a language, and this variability CONTACT: David J. Francis. dfrancis@uh.edu. Texas Institute for Measurement, Evaluation, and Statistics, 4849 Calhoun Rd, Mail Rm 373, Houston, TX 77204.