A Resolution for Polysemy: the case of Mandarin verb ZOU

A Resolution for Polysemy: the case of Mandarin verb ZOU (走) Yaling Hsu Meichun Liu National Chiao Tung University yealings@ms38.hinet.net mliu@mail.nctu.edu.tw Abstract In this paper, we propose a procedural schema as a model of cognitive processing of word senses, which can be viewed as a derivational resolution of polysemy. Previous researches, such as Frame -Based Lexicon by Fillmore [4] and Lexical Semantics by Cruse [2], are all concerned with word senses, but what is still missing is a holistic resolution of polysemy. Therefore, in this paper, we focus on the cognitive process from word form to word senses, based on corpus-based procedural resolution. In this way, we hope to provide an overall discussion and a computerizable way of solving multiplicity of semantic usages of a single word form. A case study of the Mandarin verb ZOU (走) is presented and used as an illustration. 1 Introduction Since ‘ two or more semantic elements may be expressed in a single monomorphemic lexical item’ (Bybee [1]), to understand the meaning of a word in a particular utterance, we need to resort to ‘ cognitive structures, knowledge of which is presupposed for the concepts encoded by the words’ (Fillmore [4]). According to Fillmore [4], we know word senses are not related to each other directly, but only by way of their links to common background frames and indications of the participant roles associated with such frames (i.e., Frame elements). However, when we turn to semantic multiplicity of a single word form in Mandarin, such as ZOU (走), the highlighted core elements of frames may not be enough to help us distinguish the different meanings of the single monomorphemic lexical item. The problem can be the spelled out as follows: a) Different senses of a single lexical item may have similar participant roles in the general terms and similar patterns of expressing these elements. Therefore, if we only depend on the information of core frame elements, how could we tell the different senses and in what way we can tell the non-prototypical senses from the prototypical one? Since the process of sense selection is under the force of many interacting factors (Bybee [1]), a reliable source of clues is collocational patterns that reveal lexical as well as grammatical associations of words. To fully utilize corpus data, we will look at Colloconstruction (a term adopted from Stefanowitsch and Gries’ idea with some modification), i.e., clause-internal, morpho-syntactic patterning characteristic of each sense, to further distinguish semantic polysemy. b) With the postulation of Colloconstruction, we may still encounter ambiguous cases where two different senses may share similar frame elements and similar Colloconstrucions. Thus, next in our cognitive resolution, we propose ‘ Contextual Dependence’ as another disambiguation factor which depends on discourse-level patterning across sentences, and we will have a detailed discussion in the following sections. 2 Cognitive resolution The resolution model proposed here intends to simulate human cognitive process of detecting word senses. As Cruse [2] describes, a lexical form may well be associated with an unlimited …

Content truncated to 500 words