For Inflected Language Context-sensitive Lexical Analyzer

Morphological analysis, the first step of machine translation, is basic for further analysis and convert. The multiple inflection makes it difficult for morphological analysis to achieve concise result. But the multiple inflection also carries some useful information of natrual languages. So extracting morphological features concisely and effectively is crucial to the whole effectiveness of machine translation system.The characters of inflecting languages are studied in this thesis. An inflecting language oriented context-sensitive morphological analysis approach is presented. In this approach, not only formation, but also syntactic and semantic features are distinguished. Further more, the context in which a word appears is also considered. Thus a more reliable result will be attained. To relax the burden of syntactic analysis, inflection relations between connected words are used to remove unreasonable results.Considering the characters of inflecting languages, a morphological analyzer was acheived on the base of intelligent machine translation system IMT. The analyzer has been used in German to Chinese and Russian to Chinese MT system and its effectiveness is attested.