Lexical Profiles of Reading Texts in High-Stakes Tests: Where are the Benchmarks?

摘要

In language testing literature, the lexical profiles issue has been extensively discussed when examining the quality of reading texts in high-stakes tests. The interpretation and use of lexical profiles, however, have been lacking a point of reference (i.e., benchmarks). Therefore, this study attempts to establish benchmarks for lexical profiles of reading texts in a high-stakes test in China - the National Matriculation English Test (NMET). To elicit sufficient samples, a corpus of 909 NMET reading texts was constructed. Based on the corpus, two stages were employed. Firstly, the 909 texts were screened through a text coverage analysis and representative text samples were selected. Secondly, two sets of benchmarks were established based on the text samples. Overall, this study contributes empirical evidence to evaluating the lexical profiles of NMET reading texts, and has practical implications for developing reading texts in high-stakes tests.