基于语料库的人机译本语言计量特征对比研究——以Peter Pan四译本为例

Research on the linguistic features of human translation and machine translation based on corpus:Taking Peter Pan's four translations as an example

机器翻译如今深刻影响着翻译的各个领域。为观察文学领域中人工译本与机器译本的语言特征差异,基于自建Peter Pan英文原文及其两个人工译本、两个机器译本的英汉平行语料库,运用Python、CRIE、Emeditor和ABBYY Aligner等工具,从词法与句法两个层面对人机译本的17个语言计量特征进行对比研究。研究结果表明:在文学作品翻译中,机器翻译尚不如人工翻译,机器翻译处理词汇与句子不灵活,主要表现为代词“显化”明显和行文缺乏逻辑。两个人工译本相比,任溶溶译本语言富有韵律,阅读难度较低;杨静远译本语言丰富,流畅自然。两个机器译本相比,谷歌译本较有道译本词汇表达更多样;有道译本更贴近中文表达。

Machine translation has a profound impact on various fields of translation today.To observe the differences in linguistic features between human translation and machine translation in the field of literature,the authors built Peter Pan's English-Chinese parallel corpus of the original English text and its two human translations and two machine translations.Tools such as Python,CRIE,Emeditor and ABBYY Aligner were used to conduct comparative research on 17 linguistic structures of the four translations from the lexical and syntactic levels.The research results show that in the field of literature,machine translation is not as good as human translation,with the former being less flexible than the latter in processing words and sentences,mainly inflected in features of evident"explicitation"of pronouns,and lack of logic in writing.It is found from the comparison of the two human translations that Ren Rongrong's version is full of rhythm and easy to read,while Yang Jingyuan's version is rich in language and fluent in text.It is found from the comparison of the two machine translations that the Google version has more diverse vocabulary expressions than the Youdao version,while the Youdao version is closer to Chinese expressions.