Intentional inference during infants’ observational word learning

Abstract

Word learning theories propose a dynamic development mechanism in which infants recruit statistical and perceptual cues at an earlier stage and social cues at a later stage, but it still remains unclear what promotes transition between the two stages. Recent studies have shown that the ostensive cue promotes infants’ social learning, de-emphasizing, and even denying, the contribution of their development of teleological cognition. The present study was operationalized to explore the influence of the bare teleological inference by teaching infants words in a monadic observational context where they were the only participants, and where no ostensive cues were provided by other agents. We utilized Woodward's (1998) experimental scheme and presented 12- and 17-month-old infants with a hand or a rod repeatedly approaching and grasping or touching a pair of objects, paired with two novel word labels respectively. In the test phase, where the position of objects was changed, two word–object associations were either switched or not. No detection of the changes was found for the non-intentional rod in both age groups. In contrast, with the intentional hand agent, the older group showed detection of the referential relation between word labels and objects, but the younger group simply showed sensitivity to the position. These findings provide an important insight into issues such as how infants used their teleological inference in word learning, and how multiple cues are integrated during the transition of infant word learning from statistic and perceptual learning to social and interactive learning.