探讨克服语言交流障碍的翻译器.docx

1、探讨克服语言交流障碍的翻译器 探讨克服语言交流障碍的翻译器核算机翻译一般叫机器翻译(Machine Translation或MT)，即全自动高质量机器翻译(Fully Automatic High Quality Machine Translation,简称FAHQMT或MT)，就是人类运用核算机进行天然言语间的彼此翻译，运用软件完成从一种天然言语文本到另一种天然言语文本的翻译。Accounting machine translation is generally called machine translation ( Machine Translation or MT ), the aut

2、omatic high quality machine translation ( Fully Automatic High Quality Machine Translation, referred to as FAHQMT or MT ), is the human natural language between translation using computer, using software from a natural language text to another natural language text translation.3.机器翻译的前史与现状The histor

3、y of the former and current situation of 3 machine translation在古希腊年代就有人提出运用机械设备来进行言语翻译的主意，其中之一是如何用机械手法来剖析天然言语。17世纪，大家初次提出运用机械字典战胜言语妨碍的想象。1903年，古图拉特(Couturat)和洛(Leau)在通用言语的前史一书中指出，德国教授里格(W.Rieger)初次运用了“机器翻译”(Machine Translation)这个术语。1952年在美国麻省理工学院(MIT)召开了第一届世界机器翻译会议，标志着机器翻译正式迈出了第一步。1978年在中科院核算机所的一台64

4、K容量的核算机上成功地进行了20个标题的机器翻译测验。1987年在日本箱根举办了第一届机器翻译峰会(MT Summit)，并决议今后每两年轮流在亚、欧、美定时举办。20世纪90年代后，跟着微机的遍及，相继呈现了多种翻译软件，例如金山公司的词霸系列，实达铭泰的东方快车系列，Trados 翻译软件等。到当前为止，因为核算机科学、言语学研讨的开展，特别是核算机硬件技能的大幅度前进以及人工智能在天然言语处置上的使用，机器翻译现已取得了相当大的前进，不过，机器翻译的水平距抱负的“全自动高质量（FAHQ）”的方针还有很远的间隔。In ancient Greece, people proposed the

5、use of mechanical equipment for speech translation idea, one of which is how to use the mechanical approach to analysis of natural language. In seventeenth Century, we first proposed the use of mechanical dictionary to overcome language barriers imagination. In 1903, the ancient Tula special ( Coutu

6、rat ) and the Los ( Leau ) pointed out that in the common language of history in one book, the German professor Rigg ( W.Rieger ) the first use of the machine translation ( Machine Translation ) this term. In the United States in 1952, Massachusetts Institute of Technology ( MIT ) held the first ses

7、sion of the World Conference on machine translation, machine translation officially marks the first step. Accounting machine in 1978 in the Chinese Academy of Sciences accounting machine 64K capacity by successfully for the machine translation test 20 title. In 1987 the first MT summit held in Japan

8、 Hakone ( MT Summit ), and decide the future every two years in Asia, Europe, America turns are held regularly. Nineteen ninties, along with the microcomputer throughout, have presented many kinds of translation software, such as Kingsoft Corp PowerWord series, Shida Mingtai Orient Express series, T

9、rados translation software. So far, because the computer science, language development research, especially the use of greatly advancing hardware computer technology and artificial intelligence in natural language processing, machine translation has made considerable progress, however, fully automat

10、ic high quality machine translation from the level of ambition of ( FAHQ ) policy has a long interval.4.机器翻译面对的疑问4 machine translation in the face of doubt4.1影响机器翻译质量的中心是歧义的处置。在词汇的层面上，词汇的歧义主要是一词多义，如英语单词“note”能够指“笔记”、“短信”、“注释”、“纸币”等。以下是词汇歧义的比如：Center machine translation quality of 4.1 impact is ambig

11、uous disposal. At the lexical level, lexical ambiguity is mainly polysemy, such as the English word note can refer to notes , text , notes , notes . The following is the example of lexical ambiguity:原文：Thank-you notes are heart-warming.Source: Thank-you notes are heart-warming.机器译文：谢谢笔记温暖人心。人工译文：谢谢信

12、温暖人心。Machine translation: Thank you notes warm. Artificial translation : Thank you letter warm.在布局层面上，常见的布局歧义有 and(和)布局，如 nice girls and boys, 它既可表明“好女孩和男孩”，也可表明“好女孩和洽男孩”，这种歧义只要人工翻译才干消除，核算机无法辨认此歧义，可见歧义的处置是影响机器翻译质量的要害。At the layout level, ambiguity of common and layout ( and ) layout, such as nice gi

13、rls and boys, it can show good girls and boys , also can show good girls and boys , as long as the human translation can eliminate this ambiguity, accounting machine cannot identify this ambiguity, ambiguity disposal is visible the key of machine translation quality.4.2人类翻译方针集中于方针言语，若是有必要的话，译者会选用灵敏的

14、办法以使翻译传情达意，有的时分会运用意译的办法，这是机器翻译所无法抵达的。4.2 human translation policy focused on policy speech, if necessary, the translator can choose sensitive approach to make the translation to express emotions, sometimes will use free translation approach, this is machine translation cannot arrive.以文学翻译为例，文学翻译是“传达

15、作者的悉数目的，即作者对在读者思想感情上发生艺术效果的悉数目的”。即便关于专业翻译工作者来说，文学翻译也不是件简单的事，更何况是关于没有任何认知才干的机器。笔者曾让机器翻译了下面几句：以下是美国19世纪出名作家爱默森长篇散文Beauty中的原句。Literary translation as an example, literary translation is to convey the author of all, all is the author of artistic effect in the thoughts and feelings of the reader . . . Ev

16、en a professional translator, literary translation is not a simple thing, especially on the machine without any cognitive ability. I have let the machine translation the following sentences: here is the United States in nineteenth Century, the famous writer Emerson s prose Beauty in the original sen

17、tence.原文：But this beauty of Nature which is seen and felt as beauty, is the least part.Original: But this beauty of Nature which is seen and felt as beauty, is the least part Beauty Ralph Waldo EmersonBeauty Ralph Waldo Emerson机器译文：可是这哪个看见而且作为秀丽感到的天然的秀丽，是最小有些。Machine translation: but that which is s

18、een and felt as the beautiful natural scenery, is the smallest some.夏济安译文：可是但凡耳目所能辨认出来的美，仅仅天然之美的最低有些。Xia Jian: but all eyes and ears can distinguish beauty, only the lowest some natural beauty.以下是美国出名作家马克.吐温早期创造的一则优异短篇小说竞选州长中的原句。 原文：I hauled down my colors and surrendered.The following is the famous

19、 American writer Mark. Twain created an excellent early short stories running for governor in the original sentence. Original: I hauled down my colors and surrendered Running For Governor Mark Twain 机器译文：我沿着我的色彩拖而且屈服。Running For Governor Mark Twain machine translation: I along my colors drag and yie

20、ld.张有松译文：我消声匿迹，自愧不如。Zhang Yousong: I disappear, feel ashamed of ones inferiority.以上机器译文荒诞可笑。因为机器没有思想、推理、判断才干，缺少人工译者的归纳常识和长时间沉淀下来的文化常识等，无法对原文发生全部的晓得，做的仅仅机械转码，无法打破思想妨碍。要取得地道完好的译文，有必要对机器译文进行人工修正、加工。Above the absurd. Because the machine without thinking, reasoning, judgment ability, lack of artificial p

21、recipitation down the translators inductive knowledge and long time culture common sense, can not occur all know of the original text, do mechanical transcoding, unable to break the ideological prejudice. To achieve the perfect translation, it is necessary through artificial modification of processi

22、ng, machine translation.5.机器翻译的开展趋势5 the development trend of machine translation机器翻译的质量尽管不能和人工翻译混为一谈，但有许多优势是人工翻译所不具备的。我国出名核算言语学与机器翻译教授董振东说：“美国要开发一种白话翻译机，专门用来审问与美国人言语不通的塔利班俘虏。若是用人来做翻译，因为言语不通，即便翻译被塔利班收购，对美国人信口胡说，美国人也不会晓得，而机器就不能够呈现这种状况。这机器如今有没有创造出来，能到啥翻译水平还不得而知，但这至少是一个相当好的开展范畴。” 因而，发扬机器翻译的本身优势是决议机器翻译开

23、展趋势的重要因素之一。机器翻译的开展趋势是处理人工智能的技能疑问。“假定要从人工智能上有所打破，那只能在神经网络和含糊核算上寻觅未来。”为了探究人工智能，使电脑仿照人脑，进行更多的智力劳作，含糊核算便应运而生。神经网络和含糊核算都是经过对人脑的布局和推理办法的模仿来完成核算机的智能化。“只要人脑才干和人脑比较”这看似废话，关于机器翻译来说，却是技能上真实意义上的打破口。Machine translation and human translation quality although cannot be confused, but there are many advantages are a

24、rtificial translation has not. Chinas famous accounting linguistics and machine translation professor Dong Zhendong said: the United States to develop a vernacular translation machine, designed to interrogate and American language of Taliban prisoners. If a person to do the translation, because of t

25、he language barrier, even if the translation is the acquisition of Taliban, the Americans talk nonsense, Americans will not know, and the machine will not be able to show this situation. This machine now have created, to what level of translation also can make nothing of it, but at least it is a fai

26、rly good development category. Therefore, developing machine translation itself superiority is one of important factors resolution machine translation development trend. The development trend of machine translation is the process of artificial intelligence skill question. If want to break from the a

27、rtificial intelligence, it is only in the neural network and fuzzy accounting for the future. In order to explore the artificial intelligence, the computer imitates the human brain, more intellectual labor, vague accounting will emerge as the times require. Neural network and fuzzy accounting are in

28、telligent after on human layout and method of reasoning to imitate to complete accounting machine. As long as the comparison of human talents and human - this seemingly useless talk, about machine translation skills, is the true meaning of breach.6.结语6 Conclusion当前，机器翻译的水平较50年前有了很大的晋升，但要完全战胜言语妨碍，使MT

29、投入实践使用，还需数学家、核算机教授、人工智能教授、言语学家、心理学家、认知学家、逻辑学家等通力合作，共同努力，从理论研讨和使用开发上有所打破，信任机器翻译会抵达光亮的对岸。The current machine translation, the level of 50 years ago has been greatly promoted, but to completely overcome the language barrier, make MT into practice use, still need to mathematicians, computer Professor, Professor of artificial intelligence, linguists, psychologists, cognitive scientist, logician, act with united strength, joint effort, the break from the theory research and application development on trust, machine translation will arrive at the bright side.