English Ambiguity And Disambiguation In English-Chinese Machine Translation

Ambiguity is a common phenomenon in natural language. In the course of communication between human beings, the ambiguity problem is not a great trouble because of the help of human intelligence. Therefore, it's not a problem to human translation of natural language. But for the machine which contains little intelligence, the ambiguous sentences begins to appear in a large scale and to be a fundamental obstacle to machine translation (MT). Statistics shows that a MT system with well-defined syntactic rules produces 2.8 parsing trees per sentence when English 900 is processed, and the average lexical ambiguity is 2.5. Even though the property that language distribution is comparatively uneven has been taken into consideration, the probability to generate a correct daily-used sentence (not readable sentence) is below 40%. Therefore, it is quite obvious for us to tackle with the ambiguity problem before we get any practical result in MT.As a matter of fact, the ambiguity phenomenon in natural language has so complicated a background that every approach to eliminating ambiguity can solve only a part of the problem. Now not a few of researchers have listed some specific sentences to illustrate the ambiguity problem. The author thinks that several methods should be employed in the solution of ambiguous problems. And different methods will work at different levels.Ambiguity appears at many levels, such as phonological ambiguity, lexical ambiguity, grammatical ambiguity, semantic ambiguity and pragmatic ambiguity. According to the demand in MT, ambiguity is divided into lexical ambiguity and structural ambiguity in this paper. And the disambiguation method that is discussed here is on the basis of context, and focused on the elimination of lexical ambiguity. At the end of this paper, the author lists some important subjects which are the difficulties of MT and need to be explored in the future. It is hoped that this study can provide some help for the ambiguity resolution in present machine translation research. This thesis consists of the following parts:Chapter one: introduction: English ambiguity, its present state of disambiguation in English-Chinese MT and the objective of the study.Chapter two: ambiguity and machine translation: this part will look at the definition of ambiguity, its features, its discrimination from fuzziness and vagueness and its sources; the general procedures of machine translation; and ambiguity categorization in machine translation.Chapter three: theoretical background and some strategies of disambiguation in MT: mainly include the psychological, pragmatic and semantic background for disambiguation; and the strategies of ambiguity remain, domain limitation and semantic analysis.Chapter four: disambiguation method based on contexts: discussion of the definition, representation, organizing and processing mechanism of contexts, and the algorithm for word sense disambiguation method based on contexts.Conclusion...