Two experiments investigated the nature of the emotional differences between figurative language and literal counterparts. The semantic differential method was used with principal component analysis as a data-driven implicit method for distinguishing emotional variables. The first experiment found that metaphoric stories were reliably different in emotionality than their literal counterparts along three different data-defined dimensions. The second experiment extended the conclusions to the evaluation of individual words used figuratively (including simile and metaphor). In both studies, principal component analysis revealed three distinct underlying sources of variance implicit in the ratings of experimental items including the dimensions of dynamism and depth, as well as an evaluation scale in each case. Notably, all three implicit scales, though orthogonal to each other, were found to correlate with explicit judgments of emotional valence of the stories in Experiment 1. Data-derived implicit measures are an effective way of discriminating among affective dimensions in figurative linguistic stimuli.
