Metaphor Interpretation as Selective Inferencing

Abstract

Metaphor pervades natural language discourse. T hi s paper describes a computational approach to t he interpretation of metaphors. It is based on a n a tural language processing system that uses the dis c ourse problems posed by a text to select the r e l e vant inferences. The problem of interpreting me taphors can then be translated into the problem of selecting the relevant inferences to draw from th e metaphorical expression. Thus, a metaphor is f r e quently given a cor=ect interpretation as a byproduct of the other things a natural language s ystem has to do. Two examples of metaphors are g iv e n -- a spatial metaphor schema from computer sc i e nce, and a novel metaphor --and it is shown ho w the interpretation problem for each can be translated into a selective inferencing problem and so lved by the ordinary operations of the system. This framework sheds light on the analogical processes that underlie metaphor and begins to explain ~he power of metaphor.