概化理论在语言测试新进展中的应用

The Application of Generalizability Theory in the New Context of Language Testing

概化理论是一种发展较晚但功能全面的现代测试理论,能有效地设计、分析和评估测试项目。直到上个世纪八九十年代,RobertBrennan等人开发了GENOVA等一系列应用软件后,概化理论才在心理和教育测量领域的研究中逐渐受到重视,此后,运用概化理论研究实际问题的项目也逐渐多了起来。概化理论与项目反应理论属于不同的理论范畴,它们的基本假设也有很大的差别。

GT is a broadly defined measurement theory with a relatively short history. In the studies of language testing, Generalizability theory （GT） has been utilised as a powerful instrument and is especially useful in the new context of language testing research. In this article, the authors reviewed the key concepts of GT and the applications of GT in language performance test and criteria-referenced test. Comparisons between GT and Item response theory （IRT）, Many-facet Rasch Model （MFRM） in particular, are discussed, along with the prospect of GT development of application.