Abstract:This paper gives a review of MT in terms of four aspects, namely, the definition of MT, the linguistic techniques in MT, the historical development of MT, the positive and negative views toward MT. In the part of linguistic techniques, three kinds of techniques: “direction and indirection”, “interlingua and transfer” and “local and global scope” are employed. In the chapter concerning the historical development of MT, it briefly introduces three generations of MT. This paper also studies different genres of texts translated by Google Translate to testify the existing opinions, i.e. machine translation works more efficiently but needs post-human editing to a large extent. In the case study of Google, the texts to be translated have been carefully elected based on several metrics, including sentence types, amount of pragmatic information and semantic domain. Based on the analyses of sample texts, the error genres frequently appearing in MT can be classified into three types: lexical genre, grammatical genre and pragmatic genre.

Key words: MT, linguistic techniques, Google translate.

摘要:本文回顾了机器翻译的历程，包括四个方面：机器翻译的定义，机器翻译所采取的语言学知识，机器翻译的历史发展以及人们对机器翻译所持的看法等。在语言学知识的部分，本文介绍了语言学的三个方面，第一方面是直接或间接手段，第二方面是中介语或转移，第三方面是本地或全球范围。在机器翻译的历史发展部分，本文简要地介绍了三代机器翻译。本文也进行了谷歌翻译案例研究。通过对不同文体机器翻译的研究证实现有的结论，即机器翻译虽然效率高但是还是很大程度上依赖人为修改。在谷歌翻译案例分析中，被译文章选择的标准主要有三个维度：句子长度，语用信息的多少以及语义领域。机器翻译中常出现的错误主要包括词汇，句法，语义三种。

关键词：机器翻译，语言学知识，谷歌翻译