| Three experiments examine the relative comprehensibility and memorability of conventional and nonconventional language. In Experiment 1, innovative denominal verbs (Nonconventional) or conventional verb phrases (Conventional) were embedded in identical contexts. Subjects needed the same amount of time overall to read the two versions of the items, but more time on a per-word basis for the Nonconventional than for the Conventional versions. Subjects' ratings of inferences accompanying the sentences indicated that their comprehension levels were equivalent for the two versions. However, in a surprise recognition task, they recognized the Nonconventional versions more often than the Conventional versions. The contrast between conventional and nonconventional language in Experiment 2 took the form of literal (Conventional) or figurative (Nonconventional) readings of the same phrase, as induced by different contexts. Reading times were longer for the Nonconventional than the Conventional versions, but again the comprehension levels were equivalent. The recognition task showed an apparent memorability advantage for the Conventional reading, but when imageability ratings were partialled out, the memory levels did not differ significantly.;In Experiment 3, one set of subjects ("describers") described pictures under instructions simply to describe them or to describe them to be comprehensible, memorable, or both. A second set of subjects ("finders") used these descriptions to choose the correct pictures, and later recalled the descriptions in a surprise memory task. More abstract pictures tended to elicit more nonconventional and metaphorical descriptions, whereas more representational pictures tended to elicit longer, more concrete descriptions. The finders were more accurate when their descriptions were more concrete or more conventional, and they were less confident of their choices for shorter or less conventional descriptions. Recall was poorest for longer descriptions in terms of both gist and proportional recall measures.;The results from the three experiments are interpreted in terms of two models of the relationship between comprehensibility and memorability. A cost-benefit analysis of communicative effectiveness is posited as a larger model accounting for the comprehension and memory results and for why people often use nonconventional language instead of available conventional alternatives. |
