Perceptual Consensus across Neurotypes: Autistic and Non-Autistic Raters Agree on What Makes a Good Narrating Voice

Background: The double empathy problem suggests that communication breakdowns between autistic and non-autistic people are a bidirectional issue, stemming from a mismatch in interaction styles and understanding (Milton, 2012). A key, yet underexplored, aspect of this theory is whether fundamental perceptions of speech differ across neurotypes. It remains unclear if autistic and non-autistic listeners agree on what makes speech engaging or preferable, and which specific acoustic features—such as vocal pitch, intensity, or fluency—drive these judgments. This study aims to directly test how autistic and non-autistic listeners perceive thin slices (Grossman et al, 2015) of autistic and non-autistic storytelling. Objectives: First, we tested the hypothesis that autistic and non-autistic raters would agree in their perceptual judgments of thin slices of autistic and non-autistic narrative speech (Q1). Second, we predicted reliable differences in ratings assigned to thin slices based on speaker neurotype (autistic vs. non-autistic) (Q2). Finally, we aimed to identify which specific acoustic features of the speech samples were the strongest predictors of more favorable ratings across rater groups (Q3). Methods: 30 speakers (15 autistic, 15 non-autistic) were recorded talking about a recent vacation or trip. For each speaker, we selected a 10-second clip, starting as close as possible to after the full minute of narration. Each thin-slice clip was rated by 150 raters (75 autistic via SPARK and 75 non-autistic via Prolific. Raters used ungraduated slider bars to indicate their agreement with four statements, such as “The person is enthusiastic about the topic” or “I would enjoy listening to this person read me a story.” From each clip, we extracted four acoustic measures: Pitch IQR, Percent Pause, Spaciousness, and Wiggliness (Wehrle et al, 2020). Each of our three research questions was addressed with a separate Bayesian linear model. Results: Autistic and Non-autistic raters were nearly unanimous in their ratings (Bayesian R2 = .99) (Q1), and there was no reliable difference in ratings of autistic and non-autistic speakers (mean diff effect size autistic raters = 5.2 [CI: -3.23 - 13.40], mean diff effect size non autistic raters = 5.89 [CI: -5.47 - 17.36] (Q2). When collapsed over rater and speaker group, only spaciousness reliably predicted preference (effect size 6.78 [CI: 2.24 - 11.33] (Q3) (Figure 1). Conclusions: Autistic and non-autistic raters agree in their ratings of thin-slices of narrative speech samples, suggesting shared perceptual preferences, specifically for spaciousness, i.e. wider ranges in pitch “jumps” during narratives that seem to be interpreted as indicating engagement and enthusiasm. This finding demonstrates that differences in perception or preference of prosodic features are unlikely to be the primary driver of communication breakdowns between neurotypes. References Grossman, R. B. (2015). Judgments of social awkwardness from brief exposure to children with and without high-functioning autism. Autism, 19(5), 580–587. https://doi.org/10.1177/1362361314536937 Milton, D. E. M. (2012). On the ontological status of autism: The ‘double empathy problem.’ Disability & Society, 27(6), 883–887. https://doi.org/10.1080/09687599.2012.710008 Wehrle, S., Cangemi, F., Hanekamp, H., Vogeley, K., & Grice, M. (2020). Assessing the Intonation Style of Speakers with Autism Spectrum Disorder. Speech Prosody 2020, 809–813. https://doi.org/10.21437/SpeechProsody.2020-165