A linguistic model for literary language?

Abstract

A genuine dichotomy exists between language and literary colleagues of the same department. One reason for the schism, visible for example in literary critic vs. linguists, may be because the one has but limited relevance for the other (section 1). Formal models of natural language use have grown out of the union of artificial intelligence and computational linguistics; these ‘representational grammars’ (RG) overcome some of the inherent limitations for literary language of more orthodox Chomskian or Bloomfieldian models (section 2). As a very simple literary case, Aristotle's examples of metaphors (section 3) are parsed in RG, leading to some tentative generalizations about the nature of metaphor and to a graphical display of the elements that make a metaphor what it is (section 4). A particular example of a metaphor from the Oresteia is examined, leading to the recognition that such qualities as ‘novelty’ and ‘fit’ are necessary for the distinction of metaphorical from non-metaphorical; these can be formally specified in RG terms (section 5). One may conclude (section 6) that RG models are closer to modeling some literarily relevant aspects of language than are the more orthodox linguistic models, for RG models are not forced to ignore particular or even peculiar, indiosyncratic uses of language.