The role of emergent features in metaphor comprehension

Abstract

Theories of metaphor differ in their predictions about what types of features will be present in a metaphor's interpretation. To test competing views, subjects in Experiment 1 were asked to list attributes of words in isolation (WORD), or in metaphoric statements. In the latter, they listed attributes of individual topic or vehicle words (SENTENCE), attributes shared by topics and vehicles (SHARE) or attributes of the vehicles that described the topics (ATTRIBUTE). Subjects' lists of attributes in the SHARE and ATTRIBUTE Conditions contained more emergent features (unique to the SHARE or ATTRIBUTE list) than common features (present in both WORD and SHARE or ATTRIBUTE lists). Few common features were listed. Subjects' rankings indicated common and emergent features were similar in importance for the metaphor. In metaphor interpretations (ATTRIBUTE Conditions), common features were more important to the vehicle than the topic of the metaphor. The specific content of emergent features depended more upon the vehicle than the topic. SENTENCE Condition subjects, who were told to attend to a specific word rather than the whole metaphor, found it difficult to suppress changes in word representations produced by metaphor. Experiment 2 subjects produced interpretations of Experiment 1 metaphors and rated their confidence that others would interpret the metaphor as they had. They also rated ease of interpretation, degree of metaphoricity, metaphor goodness, familiarity, and semantic similarity of topic and vehicle. Confidence ratings and actual interpretation overlap were positively correlated, indicating subjects were fairly accurate in determining when their interpretations were similar to others'. Metaphors rated high in confidence and interpretation overlap were also rated high in goodness, familiarity, ease of interpretation, and semantic similarity, but low in metaphoricity. Metaphors with higher numbers of common features in Experiment 1 were rated as very literal. Metaphors with many emergent features in Experiment 1 had low interpretation overlap scores...