Poster E39, Saturday, August 18, 3:00 – 4:45 pm, Room 2000AB

Integrating speaker and meaning in individuals with and without Autism Spectrum Disorder: evidence from eye-tracking and ERPs

Mahsa Mirza Hossein Barzy1, Jo S Black1, David M Williams1, Heather J Ferguson1;1University of Kent

Evidence suggests that while comprehending language, people build mental models that include knowledge about the speaker. For instance, using event related potentials (ERPs), a semantic anomaly type response (i.e. N400 effect) was observed when typically developing (TD) adults listened to sentences in which the content of the message mismatched the voice of the speaker (van Berkum, et al., 2008). It is claimed that individuals with Autism Spectrum disorders (ASD) have difficulties integrating information from the context to build pragmatic mental models while comprehending language (Happe´, 1996). Hence, in this study we examined whether adults with ASD exhibit comparable integration and anticipation processes for speaker and meaning as TD adults. Experiment 1 employed the visual world paradigm, and tracked the timecourse of anticipatory biases to meaning based on a speaker’s voice. Forty-eight participants (N=24 in each group) listened to sentences, in which the voice of speaker was either consistent or inconsistent with the intended message (e.g. “On my last birthday, I got an expensive electric shaver/car” in a child or an adult’s voice), and concurrently viewed visual scenes that depicted these consistent/inconsistent objects alongside distractor objects. Participants’ eye movements were recorded and they were asked to select the picture that best matched the audio description. Behavioural results showed that all participants were slower to select the correct object when it was inconsistent with the speaker’s voice than when it was consistent, though this difference was greater in the TD group than the ASD group. Eye-tracking results revealed a visual bias towards the object that was consistent with voice of the speaker from at least 600ms before the disambiguating word onset (i.e. “shaver/car”), showing that participants rapidly integrated the speaker’s voice and used this to anticipate the content of forthcoming language. In Experiment 2 we used recorded ERPs to explore how consistency between the voice of speaker and the message influences integration processes. Forty-eight participants (N=24 in each group) listened to sentences of the same type as in Experiment 1 (e.g. “I tried to refresh my lipstick in front of the mirror” in a man or a woman’s voice). EEG activity was recorded from 32 electrodes, time-locked to the onset of the disambiguating target word, which was either consistent or inconsistent with the speaker’s voice. A third sentence condition included a semantic anomaly (e.g. “I tried to refresh my seashell in front of the mirror”), and thus provided a baseline of anomaly detection effects on the N400 for comparison with speaker consistency effects. Results revealed an enhanced N400 for inconsistent sentences relative to consistent sentences, which was comparable to the N400 elicited by anomalous sentences. Further analyses examine group differences in the timecourse of these effects in both experiments. Overall, these results show that contrary to previous suggestions of pragmatic dysfunction, people with ASD are sensitive to integration between speaker and meaning.

Topic Area: Meaning: Discourse and Pragmatics