摘要:本研究对国内被广泛使用的句酷批改网的反馈效果加以分析,以促进系统更好地为中国学生英语写作教学服务。研究使用个案分析方法,对两套CET4作文样卷共10篇作文进行人/机评阅对比,同时分析了高/低分档两组作文的网上反馈信息。结果显示,就同一分数档来看,人/机评阅一致性和相关性较高。从反馈结果来看,高分档作文总体评价较高,正面反馈较多;而低分档作文则相反。反馈能认可合理的语言表达,扩充同义词、近义词知识和区别容易混淆的词。但反馈在错误准确辨识和提供合理建议方面存在问题,有待进一步改进。
This study tries to analyze the effect of writing feedbacks provided by JUKU-an extensively-used online automated writing evaluation system. It discusses the merits and demerits of the feedbacks and gives due suggestions so that the system can better serve Chinese English learners in writing. Case study method was employed for the analysis of 2 groups of 10 writings from CET4 writing samples. Agreement and correlation of the human and machine scoring were calculated, and online feedback data of high-and low-score writings were compared and analyzed. The results show that high-score writings has better general and more positive analytic feedbacks than low-score writings while low-score writings have worse general and more negative analytic feedbacks. Besides, the system highlights the appropriate expressions enlarges knowledge of synonyms and distinguishes confusing words. But there are still problems in its ability of accurately identifying errors and providing appropriate suggestions. Therefore,there is still room for improvement in the future.