UBC Theses and Dissertations
Psychosemantic aspects of figurative language Wilkinson, Walter Keith
Unusual but meaningful combinations of ideas and their expression in language are common in everyday life but no theory has yet been able to explain adequately their nature nor their effects upon human thought and behavior. Four hypotheses to explain different aspects of figurative language were proposed and elaborated: a structural hypothesis, an imagery hypothesis, and two linguistic hypotheses, one based upon semantic feature violations and the other upon Fillmore's case grammar. The structural hypothesis specified a number of conceptual relations which underlie the operation of figures—similarity, contiguity, hierarchy, and, most generally, conceptual integration. The imagery hypothesis distinguished those figures comprised of constituents with high imagery ratings from those with low imagery ratings and implied that certain combinations of high and low imagery constituents would produce more effective figures than other combinations. The linguistic hypotheses generated expectations regarding the relative effects of [-human] selectional restriction violations and the effects of violating agentive, objective, and dative case requirements. To test the hypotheses 240 grade eight, nine, and ten students from a high-SES area of Greater Vancouver were given group cued-recall tests, and likeability, similarity and comprehension scales for several lists of figurative expressions representing aspects of the various hypotheses in two syntactic patterns—nominal-copula-nominal figures, such as "A thicket is a city", and nominal-verb-nominal figures such as "The daffodil cripples the shadow". Grade trends were slight, although students in higher grades understood figurative language in a more abstract way and liked it better than students in lower grades. The psychological relevance of the structural hypothesis was particularly well substantiated by the observations. For recall and likeability, but especially for comprehension, students discriminated amongst figures in such a way as to show that the conceptual relations of similarity, contiguity, hierarchy, and integration are functional aspects of thought operative during the retrieval and interpretation of figurative language. Moderate similarity between concepts in high imagery metaphors produced more likeable figures than either extreme similarity or dissimilarity, but for figures of mixed imagery and embedded conceptual relations highly similar concepts produced more likeable figures. High imagery proved to be related strongly to figure recall, moderately to figure likeability but only modestly to figure comprehension. Figures with high imagery nouns surpassed those with high imagery verbs on all measures, and order of high and low imagery constituents favored Paivio's "conceptual peg" hypothesis for recall. Figures involving human semantic feature violations were less well recalled than those involving other violations, figures involving dative case violations were less well recalled than those involving objective case violations, and two case violations produced higher likeability and better comprehension than a single case violation. These effects are more difficult to integrate theoretically than those observed under the imagery or structural hypotheses. Because the structural hypothesis is more consistently supported by the data than are the linguistic hypotheses, and because the imagery hypothesis can largely be subsumed within the structural one, the latter, it is argued, provides the most adequate explanation of the psychological effects of figurative language. Fundamental in the comprehension of figurative thought is conceptual integration, a cognitive process facilitated by high imagery-inducing qualities of figure components, clear structural relations between them, and presence of sufficient semantic anomaly to maintain an optimal level of cognitive arousal.
Item Citations and Data