부산대학교 도서관

This study investigated the content validity of proofreading tasks in newly reformed high-stake Gaokao English tests in China from 2016 to 2021. The tests specifications and Readability analyzer were jointly used to examine the topics, genres and difficulty level of the input of Gaokao proofreading tasks, Chinese Learner English Corpus (CLEC) was employed to check the appropriateness of test items, and iWeb corpus was used specifically to observe the representativeness of collocational errors designed in proofreading tasks. With above examinations, the content validity of Gaokao proofreading tasks was considered not very high in terms of possibly inadequate coverage of genres and topics in specifications, and the tested errors’ relatively low correlation with the error types categorized in CLEC st2 group (with Pearson correlation r = .353). However, the difficulty level of task input was suitable for students’ level (with all Readability Ease scores > 60) and collocational errors in proofreading tasks were revealed to be highly representative since their frequencies in iWeb corpus were significantly fewer than those of expected forms (with Chi-square x² = 12,619,260.6). Some implications on modification of corpus tools for their better adaptation to language testing and multimedia-based English teaching were further proposed.