Abstract
Three experiments tested memory for sentences containing either metaphors (“Deserts are ovens”) or similes (“Deserts are like ovens”). Experiments 1 and 2 used forced‐choice recognition memory requiring a choice between the metaphor or simile form of the same sentence. Experiment 3 used cued recall, presenting the sentence stem (“Deserts are __”) and asking participants to fill in the rest. Experiment 1 presented a list of sentences, whereas Experiments 2 and 3 presented each critical metaphor or simile sentence at the end of a brief story that set up a meaningful context for it. As predicted by an extension of Glucksberg's (1991, 1998) attributive categorization model to memory, metaphors were consistently remembered better than similes, and concrete sentences were remembered better than abstract ones, in both recognition and recall. Contrary to predictions, the addition of the meaningful discourse context only slightly improved memory for the more difficult sentences, although it greatly improved correct rejection of nonpresented sentences. Results were interpreted in terms of discourse goals and textbase versus situation model levels of representation.