Learning gestures as object labels through cross-situational learning in 3- to 4-year-old children

This study examined whether 3-to-4-year-old toddlers can learn unfamiliar sign language-like gestures (pseudosigns) as linguistic labels by tracking their co-occurrence with spoken words across ambiguous learning trials (cross-situational learning). Forty-four children participated in a brief learning phase involving 6 spoken word-gesture pairs. Learning was assessed via eye tracking using a preferential looking paradigm in two test blocks: one evaluating word-to-gesture associations ( association block ) and another evaluating gesture-to-referent mapping ( mapping block ). A control block confirmed that children could successfully map familiar spoken words to referents, validating task comprehension and engagement. In the main analysis, no evidence of learning emerged at the group level: children did not show systematic preferences for the target, and looking behaviour did not correlate with standardized measures of cognitive or language skills. Exploratory analyses indicated age-related variability in looking behaviour across blocks. Although no reliable learning effects emerged in the full sample or age groups separately, reducing inter-individual variability based on task engagement ( control block ) revealed significant looking preferences in the older group, potentially reflecting increased symbolic flexibility at this age. Taken together, these findings suggest that any learning effects, if present, were small and heterogeneous, and methodological factors – including sample size, task complexity, and limited exposure – may limit their detection. While preliminary, these findings help delineate methodological considerations for extending cross-situational learning paradigms beyond the spoken modality and inform future research on novel approaches to investigate multimodal language acquisition in early development.