以“修辞结构理论”为导向的回指优选解析——对中心优选理论的理论性修正  

An RST-oriented Optimizing Model of Anaphora Resolution:A Theoretical Amendment to Centering Optimality Theory

作　　者：王大方[1] 

机构地区：[1]中国人民大学外国语学院,北京100872

出　　处：《外语教学》2012年第5期13-17,22,共6页Foreign Language Education

基　　金：中国人民大学科学研究基金新教师启动金项目"基于语篇结构分析的英语第三人称代词优选解析"(项目编号:10XNF027)的成果

摘　　要：回指解析一直是语言研究中最为基本和重要的问题之一。中心优选理论将优选论和中心理论有机地结合在一起,在很大程度上简化并优化了以往的回指解析方案。遗憾的是这一理论仅从语篇的线性发展入手,总是到回指语所在语句的前一句去寻找它的先行语,但先行语的实际位置显然要自由得多。针对中心优选理论的缺陷,本文尝试利用修辞结构分析对优选框架下衍生器和评估器的工作机制进行调整,从而构建一个更为理想的回指优选解析模型。Anaphora resolution has long been one of the most fundamental and vital issues in the study of natural language. Centering Optimality Theory（ Beaver 2004） reformulates the central ideas of Centering Theory（ Grosz et al. 1983, 1995） within the framework of Optimality Theory （ Prince ＆ Smolensky 1993 ）, simplifying and refining other centering-based anaphora resolution algorithms to a great extent. Regrettably, processing discourse in a linear manner, COT always looks for the antecedent within the sentence preceding the anaphor, but the proper search scope of potential antecedents is obviously much larger. Based upon hints from rhetorical structure analysis, this paper endeavors to tackle the problem through adjusting the working mechanism of GEN and EVAL. In this way, a more desirable optimizing model for anaphora resolution is constructed.

关 键 词：回指解析 中心优选理论 修辞结构理论 

分 类 号：H030[语言文字—语言学]