| The present study aims to investigate the construct validity of the Computerized Oral test (COT) of NMET II in Guangdong Province in 2004. It contained two major analyses. Both qualitative analyses and quantitative analyses were employed. The qualitative analyses consisted of experts' responses to the questionnaire on whether COT subtests could test oral communicative competence, a content analysis of the test items and a discourse analysis of test takers' performances. Twenty experienced English teachers in FELC and MA students who majored in SLA or Language Testing in Guangdong University of Foreign Studies were asked to rate the questionnaire carefully and independently. These 20 respondents had all rated COT in June 2004. The results of the questionnaires revealed that COT could cover most of the oral communicative competence. The results from the questionnaire were confirmed by the analyses of the test items and students' recorded performances. According to the expert judgments and the analyses of test items and test takers' performances, though Part A in COT could only elicit students' pronunciation and intonation, the two parts could measure more competences of students, such as linguistic competence, part of pragmatic competence and strategic competence, discourse competence and fluency. It actually covered all the 5 major competences proposed by Hedge (2001).In the quantitative analyses, 30 recorded performances were sampled from the population on COT and were re-rated. Correlation analyses and factor analysis were conducted on these performances. In the internal correlations, students' scores on each subtests correlated significantly, generally within 0.3-0.5, indicating that the subtests of COT measures the same construct but each of them focuses on a different aspect of the construct. Also, each subtest correlated more highly with other parts of COT than with tests of other constructs, say, listening or reading in the present study. All the...
|
PDF Full Text Request