A Quantitative Analysis of the Michigan English Placement Test-Form C

概要

論文の詳細を見る
There is a trend toward using commercially available English language proficiency tests in Japanese universities, to assess the language proficiency and development of individual students, and to assist language program evaluation. The Michigan English Placement Test (MEPT) is one such test. But is it an appropriate test for this context? The MEPT is a commercially available 100 discrete item, multiple-choice, receptive English language placement test. It is composed of four sections-Listening (L), Grammar (G), Vocabulary (V), and Reading (R) ; there is no oral communication or speaking section. It is used in a number of universities in Japan. Whereas this paper assesses the appropriateness of the MEPT fuom a quantataivc viewpoint, the accompanying paper is a qualitative analysis. Classical item analysis of the MEPT Form C (MEPT-C) was conducted, including the creation of a Difficulty Index and a Discrimination Index. For a sample of 342 Japanese university students, only 17% of items were appropriate, because most items were inappropriately difficult, and/or they did not discriminate adequately. For a sample of 223 high school students, only 26% of items were appropriate for the same reasons. However, when these two samples were combined (N=565), 57% of items discriminated adequately and were of appropriate difficulty. Slightly relaxing the Discrimination Index criteria increased the number of appropriate items to 79%. A number of conclusions were reached. Classical item analysis indicates that the MEPT-C may be an inappropriate English language proficiency test for Japanese university students, if the ability of the group of candidates is relatively homogenous. Moreover, a ceiling effect is apparent for fourth year students. This suggests the need to select a proficiency test that more appropriately discriminates for Japanese university students. There is an inverted u-shape relationship between discrimination and difficulty that is mitigated as the sample becomes more heterogeneous. Japanese students as a whole found the discrete skills sections-Grammar and Vocabulary-less challenging than the combined skills sections-Listening and Reading. Fourth year university students were markedly better than the first year students in the combined skills sections, but there was no significant difference in the discrete skills sections. Although there was no significant gender difference for first year university students, females were markedly better in the fourth year, principally because of their improvement in the combined skills sections. The implications of these results for language programs are discussed.
1998-11-25

A Quantitative Analysis of the Michigan English Placement Test-Form C

スポンサーリンク

概要

著者

関連論文

スポンサーリンク