Regel, S. , Gunter, T. C. & Friederici, A. D.
Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
The comprehension of figurative language, especially irony, raises some interesting questions onto how implied meanings are processed. For instance, the sentence ‘That’s great’ can be interpreted either literally (i.e. meaning something positive), or ironically (i.e. conveying the speaker’s disappointment about an annoying event) depending on the context of this utterance. In order to comprehend an ironic sentence different, and often opposite, meanings beyond the literal sentence meaning need to be processed. In two experiments using event-related brain potentials (ERPs) we investigated the neurocognitive processes involved in the processing of ironic and literal sentences. In Experiment 1, participants listened to short stories consisting of three context sentences followed by a target sentence like ‚That’s really bland’, which achieved either an ironic or literal meaning by the foregoing contextual information. ERPs at the target sentence’s final word revealed a large late positivity (i.e. P600 component) but no N400 component for irony compared to equivalent literal sentences. In Experiment 2, this ERP pattern was replicated for the visual modality. Moreover, the observed P600 component appeared to be most robustly associated with irony processing since it was neither affected by task demands (comprehension task vs. passive reading), nor by probability of stimulus occurrence. Thus, the current findings suggest that the increasing P600 amplitude for irony is related to the processing of implied meanings, and might be a reflection of pragmatic interpretation processes. Comprehending figurative language does not necessarily evoke a semantic integration difficulty (absence of an N400 component), but rather seems to involve late inferential processes for understanding ironic meanings (presence of a P600 component). It appears that in the case of irony, the processing of implied meanings does not require a rejection of literal sentence meanings but rather a computation of appropriate sentence meanings.