Ambiguities And Disambiguation In The English-Chinese Machine Translation

ln the era of information explosion and economic globalization, especially after China's entry. to WTO, international information exchange is surging tremendously. Prevalent access to theInternet makes the exchange surpass the geo8raphical boundary but lan8ua8e diffeencesbecome the biggest barrier. The most effective way to solve the problem of the translation ofmass information quickly and inexpensively is machine translation (MT), since traditionalhuman translation is far inadequate.The term machine translation (MT) is normally taken in its restricted and precise meanin8 offully automatic translation. The importance of MT research is self evident. However the processof MT rcsearch is arduous. MT technology has undergone a history of over 60 years since itsdebut in the United States fOr collecting intelli8ence in l940s. It has experienced ups and downsin its course of development and has a mixed reception. Though many MT systems come into themarkat, their quality of translation remains far from satisfaction. Owning to the charaCteriStics ofcomputer-and immaturity of lin8uistics and al8orithm, one of major bOttlenecks MT confronts isambiguity and disambiguation. Based bn the exierience the author gained in debu88in8 theE-way En8lish-Chinese lnachine translation system at the Be6in8 Chinese Information ResearchLaboratory for last yeaf, the paper discusses some.phenomena ofambi8uities in English-Chinese~ MT system and puts forward some reso1utions of ambiguities.In principle, the translation process of MT system imitates what a human translator does. The- second section of this thesis introduces the histmp development and nature of MT A humantranslator must have bilingual vocabulary and the knowIed8c of 8rammap 8cncral knowled8eand common sense and contextual jud8ment. A MT system involves two patlera systemdictionary or corpus and a translation en8ine that includes three modules; the lexical analysismodule, syntactic analysis module and transforming and generating module. It is a set of rulesbuilt on the theory of Phrase Structure and recursiveness of languages, written in a kind of lo8icprogramming language. A syntactic parsing tree will be generated through all these rules. And inthis section, the process of transforming is also demonstrated.In light of the principle of MT system, most machine translation systems operate on asingle-sentence level. By tl1is token, computer has no knowledge of the subject matter and theculture, social conventions and contextual judgment. Ambiguities are so pervasive that humanbein8 can resolve them easily but it is a hard nut to crack for the MT system. The third seCtion ofthe thesis expounds diversities of ambiguous categories that can be classified into the semanticand syntactic ambiguity. The semantic qmbiguity tefers to determining the sense of words. FirStcomes the problem of adding quantifiers, for the MT system does not know what kind quareifiersshould be added to the responding Chinese when translating numbers. Polysemous wordsambiguity refers to the failure of MT in determining the word sense used in the same part of~ speech due to shortage of semantic constraint. Part of Speech Conversion refers to theincapability of MT in determining the part of speech in the similar syntactic structure because ofstrong generatin8 capability of the syntactic rules. The Ambi8uity from references of pronounsstands for the incapability of MT system to refer back to the correct reference for lack ofcontextual judgment. Syntactic ambiguitics involve syntactic structure ambiguities, includingIv //6bambi8uities from attributive clauses, "of' phrase, prepositional phrases, parallel StruCture and-- long sentence. What is more, the thesis also discusses the influence of ambi8uities on arenamely the disambiguated source language and disambi8uated tar8et lan8ua8e, disambigUatedsource lan8uage and ambiguous tar8et language and ambi8uous source langua8e andd...