Evidence for Relational Selectivity in the Interpretation of Analogy and Metaphor

Analogies and metaphors are pervasive in language and thought . They range from scientific analogies, such as "electricity is like flowing water," to expressive comparisons such as "the moon, the massy pearl of night," to whole systems of extended meanings, such as "rising and falling GNP" (Lakoff & Johnson, 1980 ; Kittay & Lehrer, 1981 ; Nagy, 1974). In this article we ask how such comparisons are interpreted : that is, how the meaning of an analogy or metaphor is derived from the meanings of its terms . We compare four accounts of the interpretation process : Tourangeau and Sternberg's (1981) multidimensional space account, Ortony's (1979) salience imbalance theory, Holyoak's (1985) pragmatic account, and Gentner's (1980, 1983) structure-mapping theory . We argue, both on theoretical and empirical grounds, that despite its representational complexity, only the structure-mapping account is adequate to describe the phenomenona . Before laying out the theories let us be clear about the issues on which we are focusing and the kinds of comparisons we want to explain . Our concern is with the interpretation of metaphor and analogy, that is, with the way in which the meaning of a metaphor or analogy is derived from the prior representations of its constituent terms . We are not aiming at this stage to describe the real-time processing steps by which the inter-