摘要
本研究旨在利用多元线性回归统计模型,构建面向中国英语学习者的写作计算机辅助评分模型,并应用大规模语料验证其可靠性。研究发现:在考虑语法及基本写作风格等因素的条件下所构建的评分模块,辅以事后人工复查,可相当准确地预测写作的总体评分。因此在特定条件下,该评分模块可应用于计算机辅助写作评卷。
This research has constructed a computer - assisted essay scoring system for Chinese learners of English through the statistical model of multiple linear regressions, and tested the reliability of the system with a large-scale corpus. The conclusion is drawn that the essay scoring module thus derived, examining grammar and basic writing styles, is capable of making reliable prediction of the overall scores for writing samples, when complemented with post-hoc human inspection. Thus the scoring module is applicable to computed-assisted essay scoring practices under designated circumstances.
出处
《外语与外语教学》
CSSCI
北大核心
2012年第6期61-65,共5页
Foreign Languages and Their Teaching
基金
陕西省社会科学基金项目"中西部条件下面向ESL/EFL的计算机化任务型测试探索"(项目编号:07J009S)的资助
为其阶段性研究成果
关键词
英语写作
自动评分
语法判断
辅助评分
English writing
automated scoring
grammaticality evaluation
computer-assisted scoring