English-style and Chinese-style Topic : A Uniform Semantic Analysis

ENGLISH-STYLE AND CHINESE-STYLE TOPIC: A UNIFORM SEMANTIC ANALYSIS* Hui Xu Department of Computational Linguistics University of the Saarland, 66041 Saarbriicken, Germany xu@mail.coli.uni-sb.de ABSTRACT This paper provides a uniform semantic framework for Topic-Comment Structures in Chinese. It is assumed that the topic element relates to an anaphor occuring in the comment. The comment which contains an unresolved anaphor is semantically underspecified. The semantic interpretation of topic-comment constructions results from the resolution of the anaphoric relation between topic and comment. In the case where there is no explicit anaphor in the comment, a bridging inference takes place. The proposed analysis is modelled in a DRT-based framework. The introduction of a DRS segmented for topic-comment articulation, called TC-DRS, serves to compute the topic-comment relation within the discourse semantic theory DRT compositionally. 1 INTRODUCTION Topic-comment structures (TCS) have attracted much research in a wide range of languages. By studying Chinese data, the difference between the topic "English-style", which is co-referential with some (overt or covert) element in the comment, and the topic "Chinese-style", which is related to the comment by a relation of aboutness, is quite striking.' Both types of topic are exemplified in (1) and (2) respectively. (1) [Zhe ben xiaoshuoji Zhangsan kanwan� le� el. this CL novel� Zhangsan read finish Perf (This novel, Zhangsan has finished.) (2) Neike shu, yezi� da. that tree leaves big (That tree, the leaves are big.) In (1) there is an empty trace, interpreted as a zero anaphor that is bound by the topic. By contrast, there is no such syntactic position inside the comment associated with the topic in (2). Instead, topic and comment are related with each other pragmatically. Therefore, the topic in (2) is considered by some scholars to be "base-generated" in front of the sentence, and to have an "aboutness"-relation to the remaining sentence (e.g. [5]). However, a definition of "aboutness" is still not given. A pure syntactic definition of topic as an extraposition fails in the face of examples *The author is grateful to Manfred Pinkal for numerous fruitful discussions and for his encouragement. A special debt of gratitude to Claire Gardent and Johan Bos for their constructive comments. 'This dichotomy originated with [3]. Chinese-style topic constructions are also widespread in other topic-prominent languages such as Japanese, Korean, Lahu and Lisu ([5]). such as (2) while the relational approach advocated in [10] leaves open how to characterize the relation between topic and comment in (2). Furthermore, the semantics of the Chinese-style topic construction remains unexplored. In this paper, we propose a uniform semantic framework for interpreting the TCS in Chinese. We assume that the topic constituents bind anaphors inside the comments, such as pro-forms and implicit anaphora. Because the comments contain unresolved anaphors, they are semantically incomplete or underspecified. The semantic interpretation of TCS is given by anaphora resolution. In the absence of a direct anaphoric relation as in (2), a bridging inference is triggered. The proposed interpretation will be modelled in a discourse semantic framework by extending Discourse Representation Theory (DRT) …

Content truncated to 500 words