Psychological Studies in Metaphor Processing: Extensions to the Placement of Terms in Semantic Space

Psychological-processing models force the theorist to posit a model of the representation of knowledge in permanent memory and of the type of mental operations that one can apply to this knowledge. This paper describes one such model in detail. Experimental tests of the model are described to illustrate the range of phenomena being studied, as reflected in the psychological literature, and those domains of theory that are most in need of further development. The psychological investigation of metaphor can be conceptualized as encompassing three separable (but overlapping) areas of cognitive research. The first can be labeled the "recognition problem"; the question addressed here is how we recognize an utterance as figurative (and not as literally true or a sentential anomaly). In other words, why do we treat such a sentence as "all cars are lemons" differently than such a sentence as "all cats are beliefs"? The second area can be labeled the "context problem." The focus here is not on the metaphor itself, but on the extended context in I would like to acknowledge the Natural Sciences and Engineering Research Council of Canada (Grant A7040) for their support of this research. Thanks are also due to Dr. Mary Walsh for her helpful comments on an earlier draft of this manuscript. Poetics Today 13:4 (Winter 1992). Copyright ? 1992 by The Porter Institute for Poetics and Semiotics. CCC 0333-5372/92/$2.50. This content downloaded from 207.46.13.129 on Mon, 18 Apr 2016 06:36:18 UTC All use subject to http://about.jstor.org/terms 608 Poetics Today 13:4 which the metaphor is embedded. The assumption of researchers working in this tradition is that pragmatic principles (activated by context) determine whether or not a given utterance is figurative and, even if figurative, determine the nature of the non-literal interpretation. Thus a statement such as all men are animals (1) will be interpreted literally if embedded in the context of a biology class, metaphorically if presented by one student to her roommate after an unhappy relationship, and, perhaps, as ironic if said by the same student about a man whom she thinks too effeminate. How the sentence is used becomes the prime focus of study. The third area can be labeled the "computation problem." Even if we recognize that a figurative meaning is intended by a speaker, we must still compute the intended meaning. That is, even if we recognized that the intent of the statement relating cars to lemons or men to animals is to comment on a characteristic of cars or men, how do we represent such concepts as "car" and "lemon" so that the nature of the relationship can be specified? It is this area of research that has dominated the interest of cognitive scientists and that will be my focus here. Following a brief discussion of representational characteristics, I will examine how the adoption of a given representational scheme has psychologically testable implications. In concluding, I will discuss the issues raised for computational approaches by the recognition and context problems. The Representation of Concepts and Metaphor Computation To give you a feel for the distinction between recognition and computation, let me give you as an example a comment made by my daughter when she was five years old. She had asked me to buy her a toy that I thought looked better than it was. When I told her that "the grass is greener on the other side," she hesitated, looked at me, and said, "I know you are using an expression, but what does it mean?" My daughter, while unique in many ways, is apparently not unique in distinguishing between the intended non-literalness of a statement and the actual meaning one intends to convey: Ellen Winner (1988) reports that, in general, children can often recognize that non-literal meaning is intended before they can interpret it. How, then, do we compute the intended meaning? The traditional approach to this question taken by cognitive scientists has been to assume what Allen Newell (1980) has called the "physical symbol system," namely, that the mind consists of symbols that can stand for objects and events in the world and that, when combined in This content downloaded from 207.46.13.129 on Mon, 18 Apr 2016 06:36:18 UTC All use subject to http://about.jstor.org/terms Katz * Metaphor Processing 609 order, can produce new symbol systems.' Consequently, to understand metaphor comprehension (and any other intelligent activity) one has to have a model of the representation of knowledge in permanent memory and of the type of mental operations that one can apply to this knowledge. Early models of this sort assumed the representation of each concept to consist of a set of features and the problem of metaphor to be one of identifying the features shared by different concepts (see Malgady and Johnson 1980). Consider, for instance, the sentence