The Effects Of L1 Thinking On L2 Writing Test Performance

The study tries to integrate SLA (Second Language Acquisition) studies on L1 thinking into LT (Language Testing) field. SLA researches (e.g. Lado, 1957; Corder, 1967; Martlew, 1983; Ellis, 1994; Pica, 1994) have found that L1 thinking exerts certain influence on L2 writing. However, in LT researches, according to Bachman's model on the factors affecting test scores, L1 thinking is not mentioned at all. Focusing on L1 thinking in testing situations may help test constructors more specifically describe the potential factors that affect test performance, more precisely define construct, and thus improve validity. The present study adopts the think-aloud method to elicit the L2 writing processes of 16 Chinese university students. Quantitative and qualitative analyses of their writing products and writing processes are reported in order to investigate whether L1 thinking imposes certain effects on L2 writing test performance.Firstly, this study aims to investigate the relationship between total L1 thinking amount and the general L2 writing quality. On the basis of previous studies e.g. Lay (1982), Zamel (1982), Cumming (1987), Wang Wenyu & Wen Qiufang (2002), etc., we assume that test-takers' L1 thinking positively or negatively affects L2 writing test performance. Through correlation analysis, a negative relationship is proved.Secondly, in the light of a process research method, correlation analyses are conducted again to test whether L1 thinking amount in different composing processes affects writing quality differently. Together with further regression analyses, a significantly negative relationship is found between L1 thinking in text-generating and language use scores.Thirdly, with the hypothesis that students apply L1 thinking differently, protocols are analyzed to describe the characteristics of L1 thinking by the High and Low L2 proficiency students (High students and Low students). These characteristics cast some light on the cause of the negative correlation between L1 thinking in text-generating and language use quality. The Low students' frequent dependence on Chinese leads to word-for-word translation, simplification, or other strategy uses which result in linguistic errors and obscure expressions. These partly account for their poor writing quality.Fourthly, the findings in the present study are discussed. The common L1 thinking among all the subjects suggests L1 thinking is a potential factor affecting L2 writing test performance. The characteristics of L1 thinking in text-generating implies its strategic role in L2 writing tests.Finally, the implication of these findings for language testing and language teaching, especially the issues of construct definition and validity, are evaluated. At last, the suggestions for further studies are presented.