Judgments of Discourse Quality and Impression Formation of Adults with Autism Spectrum Disorder

Background: Studies of spoken discourse in Autism Spectrum Disorder (ASD) essentially rely on transcripts, and suggest less coherence relative to neuro-typicals (NT) (e.g. Baixauli, Colomer, Rosello, & Miranda, 2016; Stirling, Douglas, Leekam, & Carey, 2014). Individuals with ASD are also perceived as socially more awkward than their typical peers (Bone, Black, Ramakrishna, Grossman, & Narayanan, 2015; Grossman, 2015). Crucially, negative first impressions of adolescents with ASD have also been associated with reduced intentions to take up or maintain social interactions with these individuals (Sasson et al., 2017). However, up to date, no study has sought to directly relate the linguistic features identified in transcripts with impression formation of the speaker. Furthermore, studies have focused essentially on the impressions of typically developing individuals.

Objectives: The aim of the current study is to relate discourse features identified in transcripts with their perception by naïve listeners with and without a diagnosis of ASD as well as their contribution to impression formation of the speaker. Our research questions are threefold:1) Do naïve listeners perceive discourse features identified in transcript analyses? 2) If so, do they influence rater’s impression of the speaker? 3) Are there group differences in the perception of discourse features?

Methods: To measure participants’ impression of discourse abilities and the speaker, a rating scale was designed. Seven items targeted discourse quality (i.e. relevance, referential cohesion, coherence, level of detail, naturalness, fluency and perseverance) and four items targeted subjective impressions of the speaker (i.e. ease of understanding speaker, ease of being understood by speaker, likelihood of becoming friends, and typicality). Rating stimuli were 12 audio recordings of a dyadic conversation on the topic of relationships. Six audios involved an experimenter and a NT speaker (3 females, 3 males) and six audios involved the same experiment and an ASD speaker (3 females, 3 males). Raters were 16 participants with ASD (7 males), mean age = 28.98 (SD = 7.81), mean IQ = 113.01 (SD = 16.79). Testing is still on-going for 4 participants with ASD and the NT participants (matched on sex, age and IQ).

Results: There was a significant main effect of speaker’s diagnostic (χ²(1) = 44.926, p < 0.001), with ASD speakers receiving overall lower ratings (M=4.57, SD=1.78) than NT speakers (M=5.10, SD=1.55). There was also a significant interaction between speaker’s diagnostic and scale items (χ²(10) = 48.889, p <0.001). ASD speakers received significantly lower ratings on discourse relevance, referential cohesion, coherence, level of detail and fluency, as well as speaker’s perceived typicality.

Conclusions: These preliminary results suggest that ASD participants rated ASD speakers’ discourse, as well as ASD speakers themselves, as being less typical than their NT peers. Interestingly, there was no significant effect of speakers’ diagnosis on perceived ease to understand the speakers and being understood by them as well as likelihood to become friends. It remains to be determined whether NT raters will perceive discourse abilities and the speaker similarly to ASD raters, viz. whether speaker’s diagnostic will have an effect on the same scale items.