Are subjective ratings of metaphors a red herring? The big two dimensions of metaphoric sentences.

What makes some metaphors easier to understand than others? Several psycholinguistic dimensions have been identified as candidate answers to this question, including appeals to familiarity and aptness. One way to operationalize these dimensions is to collect ratings of them from naive participants. In this article, we question the construct validity of this approach. Do ratings of aptness actually reflect the aptness of the metaphors? Are ratings of aptness measuring something different from ratings of familiarity? With two experiments and an analysis of existing datasets, we argue that ratings of metaphoric sentences are confounded by how easily people are able to understand the sentences (processing fluency). In the experiments, a context manipulation was designed to affect how fluently people would process the metaphors. Experiment 1 confirmed that the manipulation affected how quickly people understood the sentences in a response time task. Experiment 2 revealed that the same manipulation influenced ratings of such dimensions as familiarity and aptness. Finally, factor analyses-on the ratings data from Experiment 2 and from several existing datasets-revealed two underlying sources of variance in sentence-level ratings of metaphors (the "big two" dimensions of metaphoric sentences): processing fluency and figurativeness. We discuss the implications of these findings for theories of figurative-language processing by emphasizing more careful treatment of subjective ratings of metaphoric sentences, and by suggesting the use of alternative methods to manipulate and measure such dimensions as familiarity and aptness.