世新大學圖書館

Source: Dissertation Abstracts International, Volume: 62-05, Section: A, page: 1816.

Adviser: Sally Rice.

Thesis (Ph.D.)--University of Alberta (Canada), 2000.

This dissertation addresses issues in lexical categorization. The main research question is how the meanings of a complex lexical item can best be represented. More specifically, I ask how many meanings are to be postulated, and whether and how these different meanings are related to each other. The focus of the study is on the Japanese particle, <italic>ni </italic>, whose senses are wide-ranging and span many different grammatical categories. Traditional models of lexical meaning or linguistic categorization simply cannot handle such diversity of behavior within a single lexical category in any coherent way. This study comprises a unified and empirically grounded analysis of <italic>ni</italic>'s diverse behavior.

Based on a detailed semantic analysis of an extensive array of synchronic data, I propose a network model for the semantic structure of <italic>ni</italic>, taking into account the highly polysemous nature of <italic>ni</italic>. By polysemous, I mean that one linguistic form is associated with multiple meanings. In fact, <italic>ni</italic> seems to be highly heterosemous (i.e., one form is associated with multiple meanings and grammatical functions) as a lexical item. The proposed domain-based model accommodates the diverse senses of <italic> ni</italic> by integrating them in terms of image schemas and metaphorical extensions. I argue that although <italic>ni</italic> exhibits extensive semantic diversity in its equally extensive syntactic distribution, such variation is far from random or idiosyncratic. Even seemingly contradictory senses exhibit some similarities when examined closely.

The model is then subjected to assessment and evaluation by various empirical and experimental data. Data from a grammaticalization study of <italic>ni </italic> and similar particles in typologically diverse languages indicate how the semantic distribution that <italic>ni</italic> exhibits synchronically may be associated with the grammaticalization pathways that it has undergone through its semantic development. Data from a text count study, a child acquisition study, and a series of off-line psycholinguistic experiments, also support the main characteristics of the proposed model. It is claimed that a network model for this lexeme can best handle its complex syntactic behavior and semantic functions, although the specific configuration of any representational model is very much task- and context-dependent.

School code: 0351.