The Effect Of Stem-input Modes On Performance On A Computer-based Multiple-choice Listening Comprehension Test

This paper reports on an empirical study that compared the performance andprocesses of taking computer-based English listening comprehension MCQ test(CBELCT) in audio stem-input (AI) mode, the visual stem-input (VI) mode and theaudio and visual stem-input (AVI) mode. The present study is set out to find the effectof stem-input modes on test performance of students at different proficiency-level onCBELCT and the processes of doing MCQ listening comprehension test in differentstem-input modes.The current study consisted of a20-minute CBELCT and a10-minute recordedradio interview immediately after the test for each interviewees. The test papers forthe CBELCT were designed into three versions, each of which contained threestem-input modes.90Grade3high school students took part in the CBELCT. Theywere three homogenous groups in terms of their English proficiency, which wasverified by their performance on two mock NMETs. These participants were dividedinto high, medium and low proficiency-level respectively and3at eachproficiency-level were invited to participate in the immediate retrospective interviews.After the experiments, both quantitative and qualitative data were analyzed. Thestatistical results showed the significant effect of stem-input modes on performanceon the computer-based English listening comprehension MCQ test and items withaudio stem-input mode were shown to be the highest in its difficulty anddiscrimination. Qualitative data obtained from immediate retrospective interviewsindicates that the reason lies in Chinese students’ relatively weak performance inEnglish listening, the different process of answering items with different stem-inputmodes and the limited capacity of working memory.In addition, it is also found that the stem-input mode has no significant effect onhigh proficiency students while it exerts effect on medium proficiency students mostsignificantly. The information processing and test-taking strategies of students atdifferent proficiency-levels are probed into. The finding is that high proficiency students did items in three different stem-input modes by integrating various listeningstrategies and using a schema processing, making more use of meta-cognitivestrategies than medium and low proficiency students.The above suggests that English teaching of listening should involve the teachingof information processing, listening strategies and training of working memory. Thepresent study also has implications for computer-based English listeningcomprehension test.