Spontaneous Imitation and Affective Modulation in Minimally Speaking Autistic Preschoolers

Background Autistic children present distinct imitation patterns compared to neurotypical children, with differences in frequency, accuracy, and the degree to which imitation is modulated by the demonstrator’s behavior. However, little is known about minimally speaking autistic children's imitative abilities. Objectives To investigate imitation abilities in minimally speaking autistic preschoolers compared to speaking autistic and neurotypical peers. Methods Participants were 79 preschoolers (age M = 34.7 months, SD = 11.0), including 32 speaking autistic (ASD-S), 31 minimally speaking autistic (ASD-MS), and 16 neurotypical (NT) children. Minimally speaking status was defined as an expressive language age-equivalent ≤ 14 months based on MSEL or VABS-3 scores. Participants watched video-recorded stimuli of an unfamiliar demonstrator performing object-directed actions (dots and scribbles) under two conditions: (1) Playful, characterized by positive affect; (2) Neutral, characterized by a neutral facial expression. Eye movements were recorded using a Tobii Pro eye-tracking system, focusing on two Areas of Interest (AOIs: face and action), and extracting total fixation duration (TFD) and time to first fixation (TFF). After each video, children were provided with the same materials and allowed to play freely, without explicit imitation prompts. Spontaneous imitation was coded in terms of frequency (presence/absence) and accuracy (0–3 scale). Repeated-measures GLMs were conducted to examine effects of condition and group on visual attention and imitation accuracy, while logistic regressions assessed group differences in imitation frequency. Partial correlations examined associations between visual attention, imitation, and communication, controlling for group. Results A binary logistic regression on imitation frequency showed that ASD-MS children were significantly less likely to imitate compared to ASD-S children across both conditions (Neutral: OR ≈ 5.0, p = .006; Playful: OR ≈ 5.0, p = .006), and less likely than NT children, although this difference did not reach significance. A repeated-measures GLM on imitation accuracy revealed a main effect of condition (F(1,71) = 4.45, p = .04, η²ₚ = .06), indicating higher accuracy in the playful condition, and a main effect of group (F(2,71) = 4.49, p = .01, η²ₚ = .11), with ASD-MS children performing worse than ASD-S (p = .03). No condition × group interaction emerged. Eye-tracking analyses showed a main effect of condition on attention to the face (F(1,94) = 12.08, p < .001, η²ₚ = .11), with greater looking in the playful condition, but no interaction with group. For attention to the action, ANCOVA results (controlling for age) revealed a main effect of condition (F(1,93) = 4.85, p = .030, η²ₚ = .05), with greater attention in the neutral condition, and a main effect of group (F(2,93) = 6.05, p = .003, η²ₚ = .12), with ASD-MS showing significantly reduced attention compared to both ASD-S (p = .005) and NT (p = .001). Partial correlation analyses indicated that longer attention to the demonstration was associated with higher expressive language scores across conditions (p-values ranging from .003 to .045). Conclusions Spontaneous imitation appears to be selectively impaired in minimally speaking autistic children, who imitate less frequently and less accurately and show reduced attention to the demonstration. However, affective modulation is preserved, as all groups show improved performance in the playful condition. These findings suggest that interventions for minimally speaking autistic children should leverage affective engagement and target attentional mechanisms to enhance imitation and learning.