A Comparison Of The Test-taking Processes

This dissertation is a report of an empirical study that compared the process of taking multiple-choice (MC) listening comprehension test with that of taking listening summary translation (LST) test. Translation is generally regarded as an invalid testing tool, although recent quantitative studies on translation claimed that it is of high construct validity. The MC test format, in spite of criticisms on its validity, has long been popular with designers of large-scale tests because of its high reliability. The purpose of the present study is to compare the two test formats to see what are shared and what are different between their test-taking processes so as to deepen our understanding of them.The instrument for the study comprises two tape-recorded passages, which are named "student" and "actress" based on their contents, and two test papers - Test paper 1 and Test paper 2. Test paper 1 is made up of the MC task for "student" and the LST task for "actress". Test paper 2 consists of the MC task for "actress" and the LST task for "student". Each LST task has two parts: note-taking and summary writing.Four third-year English majors from Hunan University took part in the study. They were randomly divided into two groups to take the two test papers respectively. Retrospection task was done by the participants both after completing the MC task and after taking notes in the LST task. Then they thought aloud what came to their minds while writing the summary. Both the retrospection and the thinking-aloud were recorded, transcribed, coded, and analyzed.Three tentative conclusions have been drawn based on the coding and interpretation of the protocols. First, linguistic knowledge was crucial for comprehension and nonlinguistic knowledge played either a compensatory or backup role. Among the types of linguistic knowledge employed semantics played a determinant role for comprehension. Second, both MC and LST formats posed...