论文部分内容阅读
The study bases its theoretical framework on Bachman’s model on the factors affecting test scores, in which personal trait and test method facet are both the factors imposing influences on language test performance and leading to test bias.First, this study aims to investigate the relationship between Field Independence / Field Dependence and cloze test performance. On the basis of previous studies e.g. Witkin, Hansen etc., we assume that test-takers’ FI, as an import personal trait, positively correlates with their cloze test performance. The correlation analyses are conducted to test the hypothesis.Second, the cloze procedure, as a test method, has 5 facets, i.e. Exact Word Cloze, Acceptable Word Cloze, Banked MC Cloze, Separate Item MC Cloze and C-cloze. In the light of Bachman’s model, we hypothesize the power of influence imposed by FID on cloze test performance varies across the 5 cloze test method facets. The cross-sectional comparison is performed to test the hypothesis.Third, the findings in the previous studies indicate that there exists a cross-gender difference. Then, we hypothesize there exist interactive effects between FI and the sex difference, and the sex difference affects cloze test performance through a roundabout way, i.e. via the third party-FI. To test this hypothesis, we make a factorial design and conduct a Two-way ANOVA.Another purpose of the present study is to explore the cognitive strategies taken by FI test-takers and those by FD test-takers in the process of taking the cloze tests. We take the introspection method to trace and describe the cognitive strategies.In result, the three hypotheses were proved to be true. Then, the findings in the present studies are discussed within the conceptual frameworks of FID and cloze procedure. Finally the implication of these findings for language testing theories and practices, especially the issues of construct validity and test bias, are evaluated. At last, the suggestions for further studies are presented.