Abstract
This paper reports on the comparison of two types of scale descriptors in a questionnaire on language learning strategies. The main purpose of this study was to investigate which type of the two scale descriptors, frequency-based or trueness-based, is better for language learning strategy research. With a few weeks' interval, a questionnaire on learning strategies was administered to 408 EFL learners twice with frequency-based scale descriptors and trueness-based scale descriptors alternately. First, mean differences in the responses obtained from the two different scale descriptors were examined. Second, confirmatory factor analysis was applied to see which scale descriptor of the two shows better fit to the hypothesized model. Finally, equidistance between the categories was checked. Results show that trueness-based scale descriptors elicited slightly higher mean values of responses, which resulted in a better fit to the model. The distances between the categories were almost identical for both scale descriptors. The findings of the current study partially provide supportive evidence for the claim made by Dornyei and his colleagues that trueness-based scale descriptors are preferable to frequency-based counterparts in a questionnaire on learning strategies. But they also show that the latter descriptors are not totally 'flawed' as was claimed.
