概化理论在口语考试设计中的应用研究

An Application of Generalizability Theory in the Design of a Speaking Test

本研究将概化理论应用于一项针对职前中学英语教师的口语考试设计中,对这项考试的试测数据进行了概化和多元概化分析,确定了成绩误差的来源,分析了不同考试设计模式的信度,研究结果对口语考试任务的选取、合成总分的方法、评分员人数的确定和评分任务的分配,以及评分模式和方法的选择等具有重要的参考意义。

This study applies Generalizability theory into the design of a speaking test for pre-service secondary school English teachers and investigates the trial test results by using GENOVA and mGENOVA. By identifying the sources of the errors, the study analyzes the reliabilities of different designs. The results are significant in selecting test tasks and rating methods, giving composite scores, selecting raters and allocating rating tasks among raters in speaking tests.