机器学习视域下《三国演义》三译本翻译风格对比研究

A ML-based Comparative Study on the Translation Styles of the Three English Versions of Three Kingdoms

本文使用机器学习中的分类和聚类方法,对中国四大名著之一的《三国演义》三译本进行翻译风格考察,从特征集中筛选出20个显著特征,并结合语篇进行阐释和总结。研究结果表明,筛选后的特征能够有效区分三译本的风格差异,分类器和k-means聚类分析能够实现的平均准确率均达到95%以上。在篇章层面,各译本均在词汇、符号上显示出不同的风格特征;在关键词特征方面,使用频次差异足以显示出译者的个人偏好。本研究以期对翻译风格创新对比研究提供一定实用方法。

Classification and clustering methods in machine learning are used to investigate the translation style of thethree English versions of Three Kingdoms . 20 salient features are chosen and analyzed. The research shows that the chosenfeatures can effectively distinguish their stylistic differences with an average accuracy more than 95%. At the discourselevel, each version shows its stylistic features on vocabulary and punctuation;in terms of keywords, the discrepancy inword frequency shows the translator's personal preference. This study aims to provide a practical path for the comparativestudy of translation style innovation.