Poster C35, Friday, August 17, 10:30 am – 12:15 pm, Room 2000AB

Neural processes of sarcasm interpretation: the role of tone of voice

Maël Mauchand1, Jonathan Caballero1, Xiaoming Jiang1, Marc Pell1;1McGill University

The indirect nature of sarcasm renders it challenging to interpret: the actual speaker’s intent can only be retrieved when the incongruence between the content and pragmatic cues, such as context or tone of voice, is recognized. The cognitive processes underlying the interpretation of irony and sarcasm, in particular, the effects of contextual incongruence on brain activity have recently been examined through event-related potential (ERP) techniques. The role of the tone of voice (prosody) in different stages of processing sarcastic speech, however, remains to be understood. This study aims to investigate this role by assessing differences in the processing of sarcastic and literal speech in the absence of context, when the tone of voice is the only cue to disambiguate the speaker's intent. Literal and sarcastic stimuli were created by recording verbal compliments (e.g., You are such an awesome driver) with different tones of voice to suggest a literal interpretation (perceived as friendly in a validation phase), or a sarcastic one (a veiled criticism, perceived as unfriendly in a validation phase). Later, these stimuli were presented to 21 subjects while their brain activity was recorded through EEGs during a friendliness rating task. ERPs were computed for each type of stimulus at the utterance onset, to compare the isolated effects of prosody, and at the onset of the critical word (i.e., awesome), to investigate the point at which tone of voice allowed listeners to confirm the literal intent of the compliment or its sarcastic interpretation and suggested criticism. At sentence onset, early differentiation of sarcastic versus literal utterances appeared at the N1/P2 complex over frontal-central electrodes. Compared to literal speech, sarcasm yielded a reduced amplitude at the N1, known to index discrimination of acoustic features, and a greatly reduced amplitude at the P2, a processing stage for marking the motivational salience of sounds. A later, long-lasting differentiation was characterized by higher positivity for sarcastic utterances around 600-1000ms, resembling a late positive component (LPC) over right central-parietal electrodes, implying differential and sustained cognitive analysis of the sarcastic prosody before semantic information is fully available. At the critical word onset, a negative shift for sarcastic vs. literal utterances at 600-800ms over frontal-central electrodes was detected, in a period suggested to index pragmatic (re)interpretation processes. Results show that, even in the absence of context, sarcasm can be differentiated from literal speech at the neurocognitive level. Discrimination of sarcastic intentions starts very early based on perceptual differences in the tone of voice (N1) and differences in how strategic attention reinforces and is allocated to prosodic distinctions underlying literal and sarcastic utterances as they emerge (P2-LPC). Neural processes that encode differences in the prosodic form of utterances, once integrated with the key word, then impact on late pragmatic interpretation processes when the incongruent nature of content and prosody in sarcastic messages has been made clear. Our preliminary results argue that the tone of voice plays a significant role at multiple processing stages during sarcasm perception, integration and interpretation.

Topic Area: Meaning: Prosody, Social and Emotional Processes