Semantic-based synthesis of Chinese idioms (chengyu). (Volumes I and II).

摘要

This dissertation explores the theoretical and operational aspects of a computational system which synthesizes Chinese sentences with emphasis on idioms (chengyu). The syntactic module, semantic module, and sentence separation module which form an integrated framework in automatic processing of synthesis are proposed and utilized. It attempts to achieve three purposes. (1) To create a model of computer-assisted language generation and teaching for Chinese idioms, and to provide a theory of Chinese idioms as a phenomenon in natural language; (2) To illustrate a scheme for semantic coding of synthesis based on Case Grammar Theory (Cook 1989); and (3) To demonstrate that MT in which Acceptable Quality Translation can be achieved, can be applied to language teaching and learning. Chinese idioms are defined. The collection of 680 idioms are analyzed and categorized in terms of internal constituent construction, function, and case semantics. Well-formed syntactic structures are developed in terms of 'kernel analysis'. The program is written in Turbo Prolog. Word co-occurrence patterns are used at the phrase level. Sentence Boundary Markers (SBM) and Context Free Markers (CFM) are used at sentence level. The program contains an Add In Module, a Delete From Module, a Display Module, and a Synthesis Module.