AN INVESTIGATION OF THE SIMILARITY AND CONTRAST MODELS OF METAPHORICAL AND CATEGORICAL SEMANTIC PROCESSING

RAYMOND HENRY HANSON, Purdue University

Abstract

The similarity (Ortony, 1979a) and contrast (Tversky, 1977) models are examined as formalizations of the semantic processing underlying the comprehension of metaphorical and categorical statements. Following an introduction noting the theoretical claims, applicability, and potential of the models, six experiments provide an extensive test of the assumptions and predictions of the models. The first experiment introduces a systematic approach to the generation of the stimulus metaphors and category statements to be used in subsequent tests of the models. The second experiment tests Ortony's assumption that categorical comparisons do not involve salience imbalances and that metaphorical comparisons do. The results reject a strong assumption but support a weaker assumption that is consistent with Ortony's original claim. The third experiment confirms Ortony's prediction that greater salience imbalance leads to higher degrees of metaphoricity, but only when degree of metaphoricity is distinguished from goodness of metaphor and ease of interpretation. Additional ratings of goodness of metaphor revealed that greater salience imbalance actually produced lower goodness ratings. Two procedures in the fourth experiment test the prediction of both models for directional asymmetries in categorical and metaphorical comparisons. Asymmetries were not observed for similarity ratings of separate forward and reverse statements, but were observed for ratings obtained by presenting forward and reverse statements as endpoints on a similarity scale. Degree of imbalance also affected the degree of asymmetry as the models predict. The remaining experiments tested a number of hypotheses regarding the processing mechanisms proposed in the models to account for asymmetries. The results supported a revised model similar to Ortony's in the processing of metaphors, but revealed no additional support for the Tversky model in the processing of either categorical or metaphorical comparisons. The discussion focuses upon the revisions of the models that are required by the present results, the possible extension of the models to semantic flexibility and configural effects, and the potential for future research to refine and develop the contribution of the salience framework to models of semantic processing.

Degree

Ph.D.

Subject Area

Psychology|Experiments

COinS