论文部分内容阅读
本研究在总结前人经验的基础上,针对中国学生英译汉的特点,在三种文体、近1000篇译文中提取了20多个能够反映语言形式质量的文本特征。研究进一步采用多元线性回归方法考察了这些量化指标对每种文体内一半译文语言形式分数的预测力。研究结果表明,回归方程对同一题目另一半译文的评分取得了理想效果,人机评分的相关度和一致性良好。因此,本文提取的特征可用于构建自动评分系统。
On the basis of summarizing the experience of our predecessors, this study, based on the characteristics of Chinese students translating English into Chinese, extracted more than 20 text features from three literary styles and nearly 1,000 translations that reflect the quality of language forms. The research further uses multivariate linear regression method to examine the predictive power of these quantitative indicators for one half of the translational language form scores in each style of writing. The research results show that the regression equation achieves the ideal effect on the other half of the same title, and the correlation and consistency of the human-computer score are good. Therefore, the features extracted in this paper can be used to construct automatic scoring system.