Abstract
Children learn language in face-to-face interactions with their caregivers where they are continuously exposed to multimodal input in which verbal content is combined with non-verbal signals (e.g., caregivers looking and pointing at an object while naming it). However, little is known about how caregivers’ use of the full range of multimodal behaviours is linked to children’s word learning. Using the ECOLANG Corpus, we examined how caregivers’ multimodal behaviours predict children’s word learning during a semi-naturalistic parent-child interaction. Thirty-seven caregivers (age: 29-48, 36 female; British and American English speakers) interacted with their child (age: 3-4, 19 female) talking about 12 unknown objects for 30 minutes. Caregivers’ verbal (e.g., label repetitions, speaking rate) and non-verbal (e.g., points, gaze, gestures) behaviours were annotated and used as predictors in regression analyses of children’s learning of the novel object labels along with child vocabulary scores. Results showed that, in the verbal domain, caregivers’ use of high pitch and more lexically diverse speech were associated with better word learning, while caregivers’ use of representational gestures was a significant non-verbal predictor. In particular, using more representational gestures when labelling unknown objects predicted better word learning specifically for children with lower vocabulary scores, suggesting that non-verbal cues such as manual gestures can provide additional support for children with less advanced linguistic abilities. This work presents the first comprehensive investigation of the combined role of verbal and non-verbal caregivers’ behaviours and provides robust evidence that multimodal caregiver communication can shape preschool-aged children’s lexical development.