The Effect of TUNE-in Treatment on Naturalistic Conversation in Adults with ASD: Speaking Rate Is a Temporal Marker of Rapport

Background: Social communication impairment is a core feature of autism spectrum disorder (ASD; APA, 2013) but contributors to natural conversation problems in adults with ASD remain poorly understood. For example, individuals with ASD are more likely to use unusual words (Volden & Lord, 1991), and engage less in shared conversational topics (Adamson, Bakeman, Deckner, & Nelson, 2012) than typical individuals, potentially leading to unnatural conversational flow. Beyond word-level differences, sometimes the sound of speech is off; individuals with ASD may talk too slowly, produce unusual pitch characteristics, or pause differently than typical peers (Parish-Morris et al., 2016a,b). These differences, while subtle, may contribute to the feeling of an odd conversational interaction. In this study, we test whether one subtle temporal feature that has been associated with ASD – slower than average speaking rate – improved after a 17-week cognitive-behavioral intervention.

Objectives: Determine whether Training to Understand and Navigate Emotions and Interactions (TUNE-In) treatment improves speaking rate during natural conversation in adults with ASD.

Methods: Twenty-one adults with ASD (age M=28y, IQ M=104) underwent Training to Understand and Navigate Emotions and Interactions (TUNE-In), which consists of components to improve social motivation, anxiety, social cognition, and social skills (Pallathra et al., 2017). At pre- and post-intervention visits, participants were given the Contextual Assessment of Social Skills (CASS, Ratto et al., 2011). The CASS includes two 3-minute conversations with undergraduate students (confederates) acting interested or bored. Here we report results from the “interested” condition only. Eleven different confederates contributed to the “interested” condition, and participants never spoke to the same confederate twice. Videotaped conversations were scored for overall rapport (e.g., combined effect of positive affect, social affect, involvement, vocal expressiveness, etc), by independent evaluators blinded to intervention status. The primary dependent variable of interest, speaking rate, was calculated by dividing the sum of all speech segments (not including inter-turn or intra-turn pauses) by the total number of words produced by each speaker. Speaking rate indexes how quickly a person says words, and has been found to be low in participants with ASD relative to typical controls (Parish-Morris et al., 2016a,b).

Results: Two separate linear mixed-effects regression models with participant ID and confederate ID as random effects and intervention stage (pre/post) as fixed effects revealed significant increases in speaking rates from pre- to post-intervention in participants with ASD, but not in confederates (Table). Participants who spoke more quickly post-intervention were rated as having significantly better rapport with confederates by independent evaluators (Spearman r = .65, p = .002; Figure).

Conclusions: Natural conversation is a significant challenge for individuals with ASD, but the specific contributors to awkward conversations are subtle and may be hard to specify. In this study, we found that one temporal feature of conversation, speaking rate, increased after TUNE-In and was associated with improved conversational rapport in adults with ASD. Speaking rate may thus be a feature that indexes the “goodness” of natural conversations, and may hold promise as a yardstick for evaluating the effectiveness of social skills interventions.