Abstract / Summary
Prosodic modifications in caregiver speech are related to infant attention, word learning, and long-term language development, yet how moment-to-moment caregiver prosody relates to infant attentional dynamics during naturalistic interaction remains unclear. Here, we used unsupervised learning to categorize 12,311 caregiver utterances produced during free play across the second year of life (12-24 months) into four prosodic contour types: fall, hill, rise, and valley. Infant visual attention was measured using head-mounted eye-tracking. Proportions of contour types were stable from 12 to 24 months. Despite this stability, specific contours were temporally coordinated with the onset and offset of infant sustained visual attention (SVA). Rise contours coincided with the initiation of SVA in older infants, beginning approximately one second before onset, and valley contours coincided with the period following SVA offset in both age groups. At the dyad level, two patterns emerged. Caregivers who produced higher levels of vocalizations that coincided with the onset or offset of SVA had infants with more SVA episodes and total SVA time overall. Further, caregivers who exhibited a specific prosodic pattern in the two seconds following SVA disengagement, a fall contour in the first second and a valley contour in the following second, had infants with longer average SVA episodes. These results dissociate two facets of dyadic temporal coordination: a broad association between caregiver prosody and the quantity of SVA episodes, and a specific post-disengagement prosodic pattern linked to the duration of SVA.