The development and validation of a cloze test of english structure

Abstract

Purposes of the study The purposes of this study were (1) to develop and validate a cloze test of English structure by correlation its scores with the scores on a multiple-choice structure test, (2) to find out if there are any order effects on the scores of both tests, (3) to find out if the differences between sexes affect the score of either test, and (4) to provide a frame of reference for interpreting cloze test scores in relation to multiple-choice test scores. Procedure Two 80-item multiple – choice structure tests, Form A and Form B, were pre-tested with 100 second – year students in Nakornpathom Teachers’ College. On the basis of item analysis. 75 good items were selected and re-arranged according to their difficulty levels to form a final form of the test. A cloze passage was constructed, based upon all 75 grammatical points, which were deleted. Both tests were administered twice, with a week interval, to two groups of second-year students (100 students per group) in Nakornsawan teachers’ College in reversed order. The tests were scored, and after the reliability coefficients were calculated, the data were then entered into product-moment correlations, Fisher’s Z transformations, regression analysis and t- tests. Findings The reliability coefficients of the multiple-choice and the cloze tests were .80 and .92 respectively. The correlation coefficients between the scores of both tests in both groups were .56 and .70 (significant at p<.01) and the pooled correlation was .64. There were no significant differences between the correlation coefficients of both groups, and their were no order effects on the scores of both tests. However, differences between sexes were noted on the multiple-choice scores taken before and after the cloze test, and on the cloze scores taken before the multiple-choice test. The female subjects scored significantly higher than the male subjects. However, differences between sexes had no effect on cloze test scores taken after the multiple-choice test. In addition, regression analysis indicated that multiple-choice percentage scores of 30, 40, 50 and 60, for example, were comparable to percentage scores of 21, 32, 40 and 52 on the cloze test. Recommendations Since cloze procedure has been proved as a reliable method for measuring proficiency in English as a foreign language, it should be used more and more in Thailand, specially during our national printing material shortage. Morever, value of cloze already proves for measuring reading comprehension, a unified battery of cloze passages could be used for measuring both structure and reading comprehension.