Content Validity Of The Proofreading Subtest In NMET:a Corpus Linguistic Approach

This thesis is carried out with the purpose to research and explore the content validity of the proofreading subtest in NMET by adopting the corpus linguistic approach. This study focuses on investigating whether the error types designed in NMET are representative or not or whether they are in accordance with the typical ones that students frequently make in their daily writing or not.An analysis and a statistic of the error types in CLEC’s sub-corpus ST2are made to figure out the error types that are frequently made by senior high school students in their daily writing. Then according to the criteria of tagging errors in Chinese Learners’ English Corpus (CLEC), the errors designed in the47proofreading materials which are collected from the past NMET papers of Liaoning, Shaanxi, Sichuan, Zhejiang, Chongqing, and the National Test I and the National Test Ⅱ from the year2007to2013are firstly classified (Chongqing07-12, the National Test Ⅱ07-12). Then a statistic of the error types in NMET is made as well to check what kinds of error types that are frequently designed in NMET. The results show that29error types are included in NMET, and in ST2these29error types take up85percent among the49error types that can be designed in NMET, which indicates that the error types designed in NMET are representative.Finally, a comparison is made between the frequency of the29error types in NMET and those in ST2with the assistance of the statistical software SPSS17.0and the Pearson Correlation analysis is conducted. The statistical result shows that their correlation coefficient reaches0.678, which indicates that the error types designed in NMET are in consistence with the error types that students frequently make in their daily writing, thus it can be said that the error types included in NMET and those in ST2are highly correlated.