Page 175 - Encyclopedia of Philosophy of Language

interested in the role of metaphor in scientific language and theory, and cognitive psychologists interested in building models of mind which account for the role of metaphors in cognition. Mary Hesse (1966) develops the understanding of metaphors as 'sys- tematic analogies' with a strong affinity to scientific models, arguing that they are not only heuristically valuable but are irreducible in explanation and pre- diction. Dedre Gentner (1982) proposes evaluative procedures by which to determine what makes some metaphors better than others in serving cognitive ends. The Experientialist Theory of Metaphor of Lak- offand Johnson (1981) stresses the systematic coher- ence of metaphor and its role in grounding the human conceptual system in lived experience. The Semantic Field Theory of Metaphor (Kittay 1987) employs the linguistic tools of semantic field theory to develop the work of Black and Goodman by showing how metaphor transfers the semantic structures and relations from the semantic domain of the source to that of the topic, thus inducing a new structure in the topic field.
In reaction to the predominance of interactionist- inspired theories, Donald Davidson (1978) and others reconsider the view that metaphors are implicit com- parisons. Although these theories have been quite influential amongst philosophers, they have not attracted scholars interested in developing formal approaches to natural language sufficient to the chal- lenges and opportunities provided by computational technologies. Because of the availability and promise of these technologies, metaphor research has become an area of concern for computer scientists and arti- ficial intelligence researchers attempting to enable computers to understand natural language. These researchers have developed theories using the formal tools of contemporary linguistics, such as com- ponential analysis, fuzzy logic, model-theoretic sem- antics, and semantic networks. A variety of computational approaches both to the generation and comprehension of metaphor are now available (Indur- khya 1986; Chandler 1991).
2. Figurative Language, Literal Truth, and Metaphor
An adequate theory for the identification of meta- phors must first distinguish figurative from literal lan- guage, then distinguish metaphors from other figures. Figurative utterances somehow breach the norms of literal language and yet are understood as meaningful utterances. Rules governing literal language include syntactic, semantic, and pragmatic constraints. Figu- rative utterances generally obey syntactic rules, some- times flout semantic rules, and most often violate pragmatic constraints. The flouting of these rules results in sentences which are either obviously not true or are clearly inappropriate if understood literally. Since metaphor, although occasionally identified with figurative language in general, is more precisely a par-
ticular trope, it is to be contrasted with other tropes such as irony, hyperbole, metonymy, and synecdoche. Metaphors are characteristically identifiable by the form of the semantic and pragmatic violation.
2.1 Semantic Deviance
An utterance may fail to be literally true either because it is empirically false or because it cannot be assigned any literal truth-value. An utterance cannot be assigned any literal truth-value if the terms to which its predicates are applied fail to satisfy 'selectional restriction' conditions—rules restricting the cat- egories to which the predicate can be literally applied. For instance, the predicate 'drinks' requires something animate in the thematic role, 'agent,' and something liquid in the role, 'patient.' Sentences employing syn- ecdoche, metonymy, and often metaphor fail to be literally true because they contain predications viola- ting selectional restrictions. While the violation of selection restriction is often a mark of these types of figuration, synecdoche and metonymy violate these conditions differently than does metaphor. Consider the synecdoche 'A hundred feet marched up the hill.' Although one can say that it is feet that march, it is more accurate to say that one marches with feet. That is, the selection restriction rules on 'march' call for the restriction [human] on the agent role, and for [means of mobility]on the thematic role of instrument. 'Feet' should occupy the thematic role of instrument; instead it occupies the thematic role of agent. While it is not literally correct to speak of feet marching, however, 'feet' are not conceptually unrelated to 'march.' This sort of deviance is also characteristic of metonymy, but different from what occurs in the case of meta- phor. In the metonymic sentence, The White House issued a statement affirming the right of gays and lesbians to serve in the military,' the predicate 'issues a statement' has [human] as a selection restriction on the agent role. 'The White House,' however, is where the statement is issued and so fills the thematic role of location.
In the case of metaphors, however, the violation of selection restrictions results from the importation of a term from a distant semantic domain, rather than the term's deviant occupation of a particular thematic role. For instance, Socrates explains the distress of his student, Theaetetus, by declaring that the young man is 'giving birth to an idea.' The difficulty in giving 'Theaetetus is giving birth,' a literal reading is because the procreative domain encompassing the activity of giving birth is not appropriately inclusive of human males. It is not because 'Theaetetus' occupies an inap- propriate thematic role, agent; Theaetetus is the agent under consideration.
2.2 PragmaticConstraints
Philosopher H. P. Grice provides a 'logic of con- versation' consisting of maxims obeyed by speakers
Metaphor
153