A latent trait method for measuring a dimension in second language proficiency

Abstract

In this paper it is argued that classical test theory and correlational techniques such as factor analysis have been unable to deal with many of the measurement problems that arise in the language area. As an alternative, the Partial Credit model, a latent trait model for the analysis of data scored in ordered categories, is used as the basis for the construction and analysis of an oral interview test of English as a second language. The test is designed to measure a dimension of oral proficiency that can loosely be called grammatical competence. The advantages of applying such a model are mentioned along with a discussion of its implications for future research and testing methodology in the language area.