A fragment of Korean phrase structure grammar
Date
1988
Authors
Kim, Yong-Bum
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This dissertation attempts to provide a phrase structural account of the Korean language within a restricted syntactic framework which dispenses with multi-stratal representations of sentences. It is shown that the major portions of the Korean syntax can be accounted for by a slight extension of the currently prevailing theory of Generalized Phrase Structure Grammar. The findings obtained in this thesis are two-pronged. One is related to the way in which some Korean language facts are analyzed: this aspect of our results led us to conclude that an adequately restricted descriptive model serves as a heuristic in the analysis of languages. The other involves the formal aspect of the theory which necessitated a slight relaxation of restrictions imposed on the grammar.
It is shown in Chapter 2 that what is traditionally known as a subject marker in Korean is also a focus marker in the sense that it encodes exclusiveness of given information; it is also claimed that two types of multiple nominative constructions are identifiable. Chapter 3 is devoted to justifying the existence of two separate structures of similar appearance: the missing object construction and the stative sentential subject construction. Chapter 4 analyzes the Korean honorific system, especially the subject honorific system as a control-agreement phenomenon. Chapter 5 provides an unified account of two linguistic phenomena known as scrambling and topicalization. Focusing is included in our account m addition to topicalization and scrambling. These three phenomena are captured by a single generation mechanism, while differences in subjacency and the use of resumptive pronouns are explained through parochial statements.
If the framework employed here is claimed to be a universal linguistic model, some of the restrictions imposed on the theory should be relaxed. Based on the analysis of Korean, the following changes are suggested: metarules should be able to apply to their own output and in non-Lexical environments; extraction from fillers should be allowed: the unbounded dependency feature (or SLASH) should be able to take a set of categories as its value. We believe that implementation of these suggestions into the grammar will increase the generative capacity of our model beyond that of the context-free grammar.