A Validation Study Of The Listening Test Of ESL Placement Test

This study was conducted to validate the listening part of ESL Placement test (ESLPT), a homemade placement instrument presently used in Meiya College of International Studies. Hunan University, by addressing three test method effect-related research questions. Besides, an overall evaluation of the listening test was carried out with reference to Bachman and Palmer's Test Usefulness Framework. To answer the research questions, three kinds of data, including test performance data, retrospective verbal reports, and questionnaire feedbacks, are collected and analyzed in a triangulated fashion.The main findings of this study include: 1) The task of cloze dictation has successfully tested the ability to recognize words and suffixes from aural materials. Besides word and suffix recognition, there are other skills are tapped by cloze dictation, as revealed in the analysis of verbal reports and questionnaire feedbacks. Local understanding plays an important role in accomplishing this task, together with comprehension monitoring; it is an obvious distinction between the performances of low and high-level learners. 2) Spoken Statements has some poor psychometric qualities, at least in this test version, vocabulary knowledge and sound recognition are important in answering these items correctly. The fact that too many lexical overlaps exist among spoken statements and options turns this task into another vocabulary test via aural channel. 3) The SAQ does have some method effects on test-takers' performance, especially the extra writing burden and its resulting shortage of time. Since test-takers are allowed to construct their own answers, interpretation ambiguity is another short answer question related issue, even though this problem is found only in few items in this test. As for the text and item type, the findings indicate that the narrative is the more difficult selection in this test; its text characteristic and topic familiarity may account for that. However, the fact that 3 out of 5 items attached are inferential ones complicates this issue. What we can conclude is that both factors may contribute to its high difficulty. 4) The test shows itself an acceptable psychometric quality. However, in the overall evaluation of listening test, we can see that it cannot be regarded as a real communicative language test, furthermore, in this test, the quality of authenticity is problematic in a way. Most materials used are not authentic, and many tasks are not real-life activities at all.The significance of this study lies in the direct contribution to the ESLPT listening test development. The validity information about listening test scores is of great importance for test development and evaluation. The findings of this study are informative for other similar tests. Particularly, the findings of the test formats used in ESLPT are...