The direct testing of academic writing in a second language

Abstract

In recent years, second-language proficiency testing has become increasingly concerned with testing the ability to actually use a language. Despite widespread interest, however, there has been surprisingly little discussion of how to estimate the validity of a language-use test, how to construct one without losing control over what is being tested, or how to evaluate one after it has been constructed. The present paper is an attempt to derive, use and partially validate, for one small area of language use, a theoretical approach to the design of language-use tests, which will be called explicit direct testing.