This thesis is an investigation into the semantics of abstract words and figurative language. Prompted in part by the claims of George Lakoff and Mark Johnson that abstract concepts are fleshed-out by systems of conceptual metaphors, this investigation finds that there is no semantic deficiency in the meaning of the abstract expressions game and play that necessitates augmentation with metaphorical meaning. The introduction of the thesis sets the scene for this investigation in the diverse literature on meaning, metaphor, and abstraction. It also describes the main tool of semantic analysis used in the investigation, the Natural Semantic Metalanguage developed by Anna Wierzbicka. The central chapters explicate in detail the literal and figurative meaning respectively of selected senses of game and play based on examples drawn from the Australian Corpus of English and WordBanks. The literal explications are applied to explicate the metaphors business is a game, life is a game, and doing business is playing. A tripartite schema for explications of metaphors is adapted and formalised from one developed by Cliff Goddard. This schema embodies the literal meanings of the topic and vehicle terms and a metaphoric dictum relating them in the specific context of use. Explications in this schema provides a description of, and insight into, the meaning of these and other metaphors. It is argued that this mode of presentation satisfies a range of linguistic and psycholinguistic constraints. These abstract concepts of game and play are shown to be more than mere skeletons in need of fleshing out by conceptual metaphors. Indeed, in and of themselves game and play act as vehicles (source domains) for metaphors on business, life and doing business. Out of these explications, analyses and discussions emerges a clear and coherent demonstration of the central role of the lexically encoded literal meanings of the topic (target) and vehicle (source) terms in understanding the contextualised meaning of a specific metaphor. While other scholars have tacitly or expressly acknowledged this priority, this thesis represents the first substantial demonstration of this priority using naturally occurring examples and a rigorous method of explication with a constrained metalanguage. It also suggests some accepted psycholinguistic approaches that can be used to test these hypotheses.

There are, broadly speaking, two traditions of semantics: the linguistic or conceptual tradition, which sees meaning as a cognitive phenomenon, and the logical or formal tradition, which sees meaning in terms of correspondences (truth-conditions) with an objective reality. This article adopts the linguistic or conceptual perspective. Many aspects of human cognition, such as basic perception, attention, and visual processing, are substantially shared with other primates. Language is primarily relevant to higher-order cognitive processes which are largely, if not entirely, species-unique. Importantly, human cognition not only includes reasoning and information processing about physical reality, it also includes so-called social cognition (Tomasello, 1999), that is, assessing and reasoning about intentions, mental states, and social situations, and it is in this arena that language has some of its clearest cognitive effects. (See Categorial Grammar and Formal Semantics; Social Cognition)