散文的机器翻译错误类型对比分析与统计

Comparative Analysis on Statistical Data of Error Types in Machine Translation of Prose

对当前市面上常见的机器翻译软件进行初步测评,根据散文特点、翻译软件的市场占有率和评价,选定谷歌、小牛、讯飞三款机器翻译软件进行机器翻译错误分析。结合已有的机翻错误类型,从《散文佳作108篇》中随机挑选30篇双语语料进行人机对比分析,归纳出三款机器翻译在词汇、句法、格式三大方面的错误类型,研究数据表明三款机器翻译软件在三大错误类型中都存在一定量错误,其中以讯飞表现最佳。

This paper chose three machine translation software including Google,NiuTrans and iFlytek to analyze machine translation errors according to the prose features and market share and evaluation of translation software.This paper randomly selected thirty pieces of bilingual corpus from Anthology of Chinese and English Prose for the comparative analysis on human translation and machine translation,and summarized machine translation errors in the vocabulary,syntax,and format.The result shows that iFlytek is the best performer,but three software commit some errors in vocabulary,syntax and format.