Social Context Matters for Turn-Taking Dynamics: A Comparative Study of Autistic and Typically Developing Children

Abstract

Engaging in fluent conversation is a surprisingly complex task that requires interlocutors to promptly respond to each other in a way that is appropriate to the social context. In this study, we disentangled different dimensions of turn-taking by investigating how the dynamics of child-adult interactions changed according to the activity (task-oriented vs. freer conversation) and the familiarity of the interlocutor (familiar vs. unfamiliar). Twenty-eight autistic children (16 male; $M_{age}$ = 10.8 years) and 20 age-matched typically developing children (8 male; $M_{age}$ = 9.6 years) participated in seven task-orientated face-to-face conversations with their caregivers (336 total conversations) and seven more telephone conversations alternately with their caregivers (144 total conversations, 60 with the typical development group) and an experimenter (191 total conversations, 112 with the autism group). By modeling inter-turn response latencies in multi-level Bayesian location-scale models, we found that inter-turn response latencies were consistent across repeated measures within social contexts, but exhibited substantial differences across social contexts. Autistic children exhibited more overlaps, produced faster response latencies and shorter pauses than typically developing children-and these group differences were stronger when conversing with the unfamiliar experimenter. Unfamiliarity also made the relation between individual differences and latencies evident: only in conversations with the experimenter were higher sociocognitive skills and lower social awareness associated with faster responses. Information flow and shared tempo were also influenced by familiarity: children adapted their response latencies to the predictability and tempo of their interlocutor's turn, but only when interacting with their caregivers and not the experimenter. These results highlight the need to construe turn-taking as a multicomponential construct that is shaped by individual differences, interpersonal dynamics, and the affordances of the context.

Keywords: Autism spectrum disorder; Child development; Conversational dynamics; Interpersonal coordination; Response latency; Social cognition; Social context; Turn‐taking.

Fig. 1

Plots to demonstrate the distributional properties of child response latencies across different combinations of contexts and familiarity. The black dashed line marks a response latency of zero, the purple dashed line marks the mean in the typical development group, and the green dashed line marks the mean of the autism group. As can be gleaned from the plots, response latencies involve both negative response latencies (overlapping speech), a large center of mass just above zero, as well as positive skew (long pauses between turns).

Fig. 2

Model estimates for child response latencies across different conversational contexts for autistic children (green) and typically developing children (purple). The points denote participant‐level posterior predictions from the model. The density plots and boxplots show aggregated posterior predictions for the three different social contexts: Matching With Parent, Convo With Parent (Conversations with Parents), and Convo With Experimenter (Conversations with Experimenter). The boxplots show the median (vertical line dividing the box), interquartile range (box edges), and most extreme data points (whiskers) for the aggregated posterior predictions. This figure shows the estimated $\mu$ parameter (central tendency) from our ex‐Gaussian model and specifically captures the location of the main distribution after accounting for systematic differences in extended pauses (right tails), which are separately modeled by the beta parameter. Note that these values differ from raw data averages as they account for the hierarchical data structure and separate the distributional components.

Fig. 3

Model estimates for change in child response latencies as a function of one standard deviation increase in each measure across the three social contexts: Matching Game, Convo With Parent (Conversations with Parents), and Convo With Experimenter (Conversations with Experimenter). The boxplots show the median (vertical line dividing the box), interquartile range (box edges), and most extreme datapoints (whiskers) for the aggregated posterior predictions. The top three measures (i.e., Social Cognition, Social Awareness, and Social Motivation) come from the SRS‐2, whereas Language Skills refer to CELF‐5 and Motor Skills refer to VAB‐3.

Fig. 4

Model estimates for how child response latencies change as a function of the predictability of the previous adult response latency (top) and the predictability of their own utterance (bottom) across the three social contexts: Matching Game, Convo With Parent (Conversations with Parents), and Convo With Experimenter (Conversations with Experimenter). The cosine similarity score here represents scaled values (i.e., one unit increase equals one standard deviation increase in cosine similarity). The faded lines are posterior predictions from the model for individual child participants in the study, whereas the thicker lines are average predictions.

Fig. 5

Model estimates for how child response latencies change with the latency of the previous adult response latency (top) and the previous child response latency (bottom) across the three social contexts: Matching Game, Convo With Parent (Conversations with Parents), and Convo With Experimenter (Conversations with Experimenter). The faded lines are posterior predictions from the model for individual child participants in the study, whereas the thicker lines are average predictions.