| This dissertation offers an application of the theory of cognitive grammar to the semantic theory of metaphor in which concepts from one domain are mapped onto and understood in terms of conceptual structure from another domain of experience. The central contributions of the dissertation to this theory are in three areas: (i) the productivity of metaphor, (ii) the licensing of metaphor, (iii) the relation among domains, image schemas and construal. These results contribute new specific detail to cognitive semantic theory.;An introduction to cognitive grammar and the cognitive semantic theory of metaphor are presented in Chapter 1, with other relevant models of concept representation, lexical semantics and conceptual development.;Chapter 2 applies a cognitive theory of morphology to the cognitive theory of metaphor. Metaphors vary in their generality (schematicity) and productivity, and their behavior can be described with the same model used for morphological productivity. The results demonstrate that morphological and semantic structures are organized according to the same principles.;Chapter 3 considers the problem of how the semantic structure of metaphors licenses domain mappings. The Invariance Hypothesis (IH) is a constraint on possible metaphor mappings, limiting what components of the source and target domains are preserved in a mapping. These components, called "image schemas", are idealized knowledge structures lacking the rich detail of sensory images. The analysis of metaphors involving container image-schemas demonstrates that different aspects of containers are mapped onto different target domains. Therefore, the IH is a necessary but not a sufficient condition on possible metaphor mappings.;Basic to any cognitive semantic theory are the notions of cognitive domain, image schema and construal. Chapter 4 addresses the pressing need to more precisely clarify these notions which are important for understanding word meaning, grammatical semantics, and metaphor. A large set of basic domains are analyzed (e.g., SPACE, TIME, SCALE, PITCH, TEMPERATURE, TASTE) by extending the existing theory of domains. From this it is argued that possibly all image-schematic structures can be reduced to either domains (or dimensions of domains) or construals. Chapter 5 summarizes these results and briefly discusses their implications for knowledge representation and language processing. |
