Does Conventionality Matter in Metaphor Understanding?

Metaphorical utterances differ in conventionality. Most models of metaphorical language processing do not address conventionality, with the exception of the Graded Salience Hypothesis (GSH), (Giora 1997, 1999, 2003, Giora and Fein 1999). The GSH predicts that sentences with conventional metaphorical meanings will be processed faster in metaphorical contexts than those with unconventional meanings. The current study tests whether or not conventional metaphorical linguistic expressions are processed faster than unconventional equivalents in metaphorical contexts.

            There are three major accounts of metaphorical language processing. The first, the three-stage model (Searle 1979), implies that it takes longer to process metaphorical than literal sentences. The second is the direct access view (Gibbs 1980, 1982, 2002), which states that metaphorical sentences are comprehended faster within appropriate contexts. The last is the GSH. Focusing on the salience,such as familiarity, frequency, conventionality, and contextual information of a sentence, this theory states that the more salient the sentence, the faster the processing time. Investigating effects of conventionality on metaphor processing allows the current study to distinguish among these theories.

            This study consists of two experiments. Based on thirteen well-known Conceptual Metaphors (Lakoff and Johnson 1980), twenty pairs of conventional and unconventional expressions were created, each having a possible metaphorical and literal meaning. For example, for the Conceptual Metaphor love is a journey, paired examples of a conventioanl and an unconventional expression would be It was a long, bumpy road and There were a lot of potholes, respectively. Both can be interpreted metaphorically in the context of a love relationship, but can also be interpreted literally in the context of a journey or travel. We prepared two three-sentence paragraphs as contexts in order to prompt either literal or metaphorical interpretations of the target sentence for each pair of conventional/unconventional expressions.

            Experiment 1 is a rating task. We ask twenty subjects to rate how well the last sentence, either a conventional or an unconventional expression, of a four-sentence paragraph fits with the preceding context using a seven-point scale. This experiment examines to what extent people are aware of sentence conventionality, and also serves as a norming study for experiment 2.

            Experiment 2 is a self-paced reading task, where the processing time of conventional and unconventional expressions in both metaphorical and literal contexts (the same four-sentence paragraphs in which the last sentence is the target sentence) are measured. Forty subjects are asked to read these paragraphs on a computer screen, pressing a key immediately after reading each sentence. The RT of the target sentence is recorded and analyzed, with the literal context serving as a control.

            Preliminary results from this ongoing study indicate that there is an interaction effect where conventional utterances are processed faster in metaphorical than in literal contexts (mean RT:1376ms and 1455ms respectively), relative to unconventional equivalents, which are processed slower in metaphorical than in literal contexts (mean RT:1439 ms and 1345ms respectively). This suggests that sentences with conventional metaphorical meanings are processed faster than those with unconventional metaphorical meanings in metaphorical contexts. Conventionality matters in metaphor understanding.