A Semantic Study Of Lexical Conversion Based On Predicate-Argument Structure

The classification of lexical categories (also called parts of speech or word classes) is the center of traditional grammar of Indo-European languages and closely related to the grammatical analysis of sentences. At the syntactic level of grammatical analysis, each constituent of a sentence is assigned to a lexical category, reflecting the syntactical function of each constituent to predicate of the sentence. Chinese grammar studies modeled after the grammar models of Indo-European languages, which are not actually applicable to Chinese. Chinese lexical categories have their own characteristics varying considerably from those of English in terms of morphological, syntactic and semantic properties.According to Croft (2000:13),"the main problematic categories for cross-linguistic identification are the fundamental grammatical categories:noun, verb and adjective, subject and object, head and modifier...,"i.e., the lexical categories and related categories. Differences in the properties of lexical categories between English and Chinese make English-Chinese translation a challenging operation. To confirm the usage of the target language, some conversions should be made in English-Chinese translation instead of following the original lexical categories in source language mechanically.Traditionally, translation by conversion is normally operated at the grammatical level, but the constraints of syntactic rules and grammatical forms for translators may lead to awkward and unnatural translations. Therefore, the approach of conversion adopted in this thesis is semantic-oriented as opposed to grammatically-based. Beyond the syntactic level, there are semantic levels including the argument categories level and the predicate-argument relations level. At the argument categories level, each constituent of the sentence is described as a semantic concept. The predicate-argument relation reveals the semantic role of each argument related to its predicate. At this higher level, argument-predicate representation neglects the grammatical properties of the surface structure and contains more abstract and precise information. Actually, the meaning of a sentence relates in complex ways to the syntax of the sentence; therefore, in spite of having different forms, these levels actually describe the same event. As a result, in English-Chinese translation, the predicate-argument structure can be helpful to overcome the syntactic variation problem and break through the constraints of grammar.What follows is an attempt to provide an account of conversion in terms of grammatical comparison between English and Chinese language and to propose a semantic approach to conversion based on the predicate-argument structure. Apart from characterizing the morphological, syntactic and semantic properties of both English and Chinese syntactic categories, this thesis involves investigating the linking between the semantic level and the syntactic level of a sentence as well as determining the semantic solution to conversion in the following structures:NPs as Experiencer or Stimulus, subject-predicate structure, the transferred epithet and the motion events.