Earticle

“Test analysis of listening and reading item characteristics for the young Learners of English.” Studies in English Language & Literature. 37.2 (2011): 249-74. Currently, high-stake standardized language tests have been designed and administrated for the young learners of English at the elementary and secondary levels in Korea. The purpose of this study is to analyze the item characteristics of two tests which were intended to measure young learners’ English listening and reading ability. In order to identify the quality of test items, the three parameter Item Response Theory (IRT) was employed. Test takers were the young learners of English at the elementary and secondary levels, and they responded 60 items for each test. The quality of individual items was calculated with respect to the item difficulty level, discrimination power, and guessing parameter. The item parameters of sub-parts of tests were compared respectively on the basis of statistical indices. The findings suggest that the means of the difficulty, discrimination, and guessing parameters in two tests were quite similar. This study, however, identified the problematic aspects of item characteristics to be considered for enhancing the quality control of parallel test forms. (Chung-Ang University)

This study developed parallel picture forms for a narrative writing task, one version for an English writing task, and the other for a Korean writing task, to assess the extent to which elementary school students in a bilingual immersion program develop writing skills in both languages. One form was assigned randomly to a half of the total number of subjects and the other form to the other half. Across the groups both factor means and factor variances are found to be equal, providing the evidence that the two forms are parallel, namely equivalent. The results indicate that the superficial differences in the picture versions do not have a significant effect on students' performance. Put differently, all else being equal, the forms are found to generate comparable writing skills. Therefore, the picture pair is recommended as high-quality alternate writing prompts for those wanting to compare simultaneous writing skills in two languages or to measure growth of writing skills over time. The alternate measures are to preempt problems with practice effects and translation when repeated test administration takes place. The paper provides helpful guidelines for testers wanting to develop multiple versions of picture prompts and confirm the psychometric comparability of the versions.