Effects Of Response Format On Performance In Computerized Oral Composition

With advances in science and technology and its wide application, the innovations in language testing have brought about the popularity of computerized oral test. A huge number of studies have been prompted to investigate various aspects of the oral test that might affect test takers' performance. However, research into the effect of performance condition is far from abundant in a testing context. To explore this field fully and thoroughly, this thesis undertakes an empirical work to find out whether performance condition has effect on computerized oral composition test by comparing two different response formats: video-present narration and video-absent narration.The test task was extracted from Part C of Paper A in the NMET2 Computerized Oral Test in Guangdong, in 2007. The original form remained intact in one condition while the video clips were removed in response time in the other.Sixty third-year senior high school students participated in the study. They were two homogeneous groups in terms of their Oral English Proficiency. One group took the oral composition test with video-present response format and the other with video-absent one. At the end of the tests, the students were required to complete a questionnaire about their perception of and reflections on task processing. The data were gathered both by subjectively rating the speech samples and by counting discourse features based on the spoken discourse analysis.Results showed that response format had no effects on the students' overall performance in computerized oral composition tests. However, the test takers under the video-absent narrating condition achieved significantly lower content but higher fluency scores than those under the video-absent narrating condition. Furthermore, the results of transcribtion and coding demonstrated that, compared with the video-present narrating condition, removing visual support in response iime was beneficial to test takers' Speech Rate and led to statistically more inferential but less factual information conveyed in storytelling.This study sheds some enlightening insights to teachers and testers engaged in computerized oral composition test about the influence of response format on examinees' test performance and task processing. It also comes to a fresh conclusion that response format has no effects on test-takers' ovall performance but affects individual aspects of their performance and the process of task transaction.