The meaning and uses of language test scores: An argument-based approach to validation

Abstract

In this article, I discuss how the quality of language tests can be determined by means of a validation process. In the past, the quality of language tests was often determined by examining their reliability, content validity and reflection of real-world tasks. There have also been attempts to define the language ability construct, but this has proved to be a divisive issue. Attempts at validation were often unsystematic and ad
hoc, reflecting a “toolkit” approach. Recent work in validation suggests an argumentbased approach, which focuses on both the interpretation and uses of test scores. One of the main proponents of this approach is Michael T. Kane. I outline and assess his approach to validation and discuss and evaluate the most common inferences in language testing, such as sampling, scoring, generalization, extrapolation, explanation and utilization/decision-making. This approach allows for a systematic approach to the evaluation of tests, but requires further refinement in language testing and assessment.

Keywords: validity, validation, argument-based approach, test interpretation, test usefulness, test inferences