Context, conflict, and the time-course of interpreting irony

Irony requires listeners to infer that speakers mean the opposite of what they say (e.g., “What a fabulous chef he is” to imply that he cooks poorly), but it remains unclear what interpretative algorithms enable listeners to arrive at stable yet flexible meanings during comprehension. Across four experiments, listeners heard various speakers describe two characters while their eye-movements were measured to these referents. Afterwards, listeners provided judgments about the speaker’s intent. Eye-movements revealed frequency effects for literal but not ironic meanings, systematic reference restriction for ironic but not “opposite” speakers, and late-emerging conflicts between literal and ironic meanings. Judgments revealed intuitions about the pragmatic function of irony and relations to truth conditions. These findings demonstrate that ironic interpretations are not directly retrieved from conventionalized representations in the lexicon. Instead, they involve real-time reasoning about utterance semantics and speaker intentions, and that these dual processes yield distinct signatures during comprehension.