Toward a Psycholinguistic Model of Irony Comprehension

This dissertation examines how listeners reach pragmatic interpretations of irony in real-time. Over four experiments I addressed limitations of prior work by using fine-grained measures of time course, providing strong contexts to support ironic interpretations, and accounting for factors known to be important for other linguistic phenomena (e.g., frequency). Experiment 1 used a visual world eye-tracking paradigm to understand how comprehenders use context and frequency information to interpret irony. While there was an overall delay for ironic utterances compared to literal ones, the speed of interpretation was modulated by frequency. Participants interpreted frequent ironic criticisms (e.g., “fabulous chef” about a bad chef) more quickly than infrequent ironic compliments (e.g., “terrible chef” about a good chef). In Experiment 2A, I tested whether comprehending irony (i.e., drawing a pragmatic inference) differs from merely computing the opposite of an utterance. The results showed that frequency of interpretation (criticisms vs. compliments) did not influence processing speed or overall interpretations for opposites. Thus, processing irony involves more than simply evaluating the truth-value condition of an utterance (e.g., pragmatic inferences about the speaker’s intentions). This was corroborated by Experiment 2B, which showed that understanding irony involves drawing conclusions about speakers in a way that understanding opposites does not. Opposite speakers were considered weirder and more confusing than ironic speakers. Given the delay in reaching ironic interpretations (Exp. 1), Experiments 3 and 4 examined the cognitive mechanics that contribute to inhibiting a literal interpretation of an utterance and/or promoting an ironic one. Experiment 3 tested whether comprehending irony engages cognitive control to resolve among competing representations (literal vs. ironic). Results showed that hearing an ironic utterance engaged cognitive control, which then facilitated performance on a subsequent high-conflict Stroop trial. Thus, comprehenders experience conflict between the literal and ironic interpretations. In Experiment 4, however, irony interpretation was not facilitated by prior cognitive control engagement. This may reflect experimental limitations or late-arriving conflict. I end by presenting a model wherein access to the literal and ironic interpretations generates conflict that is resolved by cognitive control. In addition, frequency modulates cue strength and generates delays for infrequent ironic compliments