An Investigation of EFL Learners' Diagnostic Assessment Capabilities for a Classroom-Based Speaking Test

Abstract

　　This study aims to examine the quality of peer assessment of a classroom-based speaking test. Peer-assessed test scores were compared with teacher-assessed test scores using a diagnostic rubric developed for the test. Both peer- and teacher-assessed scores were also compared with those of an external speaking test, the Telephone Standard Speaking Test (TSST), to ensure the reliability of the teacher assessment. The results of the comparison not only showed that the teacher assessment of student performance was nearly as accurate as that of the TSST but also that students could not distinguish between the two assessments. In addition, it was found that students assessed their peers (particularly less-proficient ones) more leniently than did teachers. Regarding the students' diagnostic assessment capabilities, they were as accurate as the teachers on non-linguistic aspects, such as the opinion exchange criterion, but far less so when assessing linguistic accuracy. The results suggest that it may be efficient for teachers to use peer-assessed test scores for non-linguistic aspects and concentrate on assessing their grammatical accuracy for formative assessment.