Hearing what you're thinking: acoustic cues of speaker intentions

How do we understand if an utterance is sarcastic or not? While we generally seem pretty good at spotting a sarcastic comment, this ability is lacking in many mental disorders (e.g. schizophrenia, autism, depression, etc.). What are the grounds for this impairment: Is this due to high level deficits in theory of mind, or at lower level perceptual and attentional processes not picking up on cues for sarcasm? Understanding the way we cue for and understand sarcasm might thus be very important in granting us a handle on social impairments in mental disorders.
In this project we analyze the relative importance of verbal and visual cues in recognizing sarcastic utterances in English and Danish corpora, with a particular focus on acoustic/prosodic properties. Machine learning techniques are employed on staged sarcasm and the resulting models are applied to spontaneously produced sarcastic utterance, with the aim of understanding which cues are crucial to individuate sarcasm, and could therefore be impaired in mental disorders.
This project has been supported by seed funding from the Interacting Minds Centre (https://sites.google.com/a/interacting-minds.net/seeds/home/2012/weed)