The Effects Of Expected Response Formats On Translation Performance

The characteristics of expected response is one of the most important part of Test Method Facets (Bachman,1990) and Task Characteristics Theory (Bachman,1996), so the effects of different expected response formats on test takers'performance have been given growing attention. Most studies at home and abroad focused on the effects of selected response, especially different reading, oral, and listening response formats, on test takers' test scores. Some researchers also wanted to study the effects of different response formats on writing, but were limited by few response formats of writing test. In recent years, an increasing number of scholars began to study the effects of different constructed response formats on test scores. A new translation format, completing-sentence C-E translation, is added to CET 4, which has slight difference with the whole sentence C-E translation. Hence, this paper aims to study the effects of different response formats on translation performance.This paper selected the sophomore students of International Trade 081,082, and Countryside Development 081 in Shanxi Agricultural University as subjects.40 of them were selected at random to participate in the pilot test, and 37 papers were thought effective. All the 100 students participated in the main research,60 effective papers and 99 effective questionnaires were collected. The reliability of all the questionnaires used in this paper was calculated by Cronbach coefficient, and the result showed that the reliability coefficients of the questionnaires were all acceptable. Simultaneously,6 college English teachers, who have rich experience in college English teaching, were also chosen as raters to participate in the study. The rating scale of CET 4 format was designed based on the rating scale of CET 4 rating scale in June,2006, while the rating scale of the whole sentence C-E translation was set based on the think aloud data of the 6 raters. Rating reliability was all calculated by Cronbach coefficient and the rating difference was tested by ANOVA to make sure the scores can reflect the test takers'real translation competence as much as possible. After the testing, rating scores of Rater 1, Rater 3, and Rater 4 in the two formats of main study showed no significant difference, and their rating reliability coefficients were all higher than 90%. Hence, the average score of these 3 raters are adopted as the final scores of these two formats in main study. By qualitative and quantitative analysis, the present paper analyzed the research question from three aspects:do these two expected response formats have similar nature, i.e., do they have similar test function? Do the scores of these two expected response formats have significant difference? If they do, which kind of response format does elicit test takers performance better? What perception do raters have of rating scale and rating process in practical rating, i.e., what effect does rating have on translation performance.From the quantitative analysis of raters'and test takers'perception of response formats, it is found that the nature of these two response formats are rather similar. In other words, the function of completing-sentence C-E translation is to test vocabulary and grammar, while the function of the whole sentence C-E translation is to test not only vocabulary and grammar but also sentence structure. Although completing-sentence C-E translation gave test takers many hints and its test points were fewer, the quantitative analysis of the test scores showed that the scores of completing-sentence C-E translation were significantly lower than those of whole sentence C-E translation. The result of the quantitative analysis of raters'perception of rating scales and rating process in practical rating shows that, rating remains one of the most important elements affecting translation scores. Simultaneously, from the quantitative analysis it is also found that although all the raters think the test function of CET 4 is to test vocabulary and grammar, and the test function of whole sentence C-E translation is to test not only vocabulary and grammar but also sentence structure, they all rank information expression as the most important element in practical rating. Consequently, it is further proved that the nature of CET 4 format and whole sentence C-E translation is rather similar.