Linking China's Standard of English Language Ability with the Automated Essay Assessment

Abstract

The purpose of this two-stage study is to examine the potential of China's Standard of English Language Ability (CSE) for Automated Essay Scoring (AES) development. In the first stage of the study, 657 descriptors from 24 writing-related CSE scales were inspected to generate categories representative of EFL learners' academic writing ability. In the second stage, the findings were compared to e-rater construct composites. Similarities between the CSE and e-rater categories were evident in terms of the overall emphases on the grammatical and textual dimensions as well as the coverage from the basic concepts like mechanics to the more abstract terms such as style. The comparison also pointed to great difference. The CSE categories presented positive descriptions of language use for learning/teaching purpose, while e-rater focused on language errors out of scoring concern. E-rater offered more explicit and operational interpretations of surface linguistic features. This paper not only supplies validation evidence of the CSE scales, but also demonstrates their usefulness for developing a customized AES system for EFL writers in China.