Lexical Bundle Investigation for Automated Scoring of Business English Writing

摘要

Business English (BE) writing teaching and assessment are important in college level English teaching. Automated essay scoring (AES) is already widely accepted in English writing evaluation and should be applied in business English teaching. One of the key steps of AES is the selection of scoring features. In order to explore the effect of lexical bundles applied in AES research, the relationship between lexical bundle amount and business English writing quality is investigated. Two business English corpora were constructed: BE writing textbook corpus and college students' composition corpus. Target lexical bundles were extracted from the textbook corpus. The correlation coefficients between writing scores and target lexical bundles from students' writing were calculated. The results show that the amount of three-word bundles and the number of total bundles have a moderate correlation with writing scores, which means these two features should be covered in AES study of BE writing.