Two-year-olds learn words for absent objects and actions

Two studies of word learning in 24-month-old children are reported, one involving an object word (Study 1) and one involving an action word (Study 2). In both studies, non-verbal scripts of playing with novel objects/actions in particular ways were established before the child was exposed to any language models. Following this pre-training, children heard an experimenter announce her intention to either find an object or perform an action. In the referent condition, children then saw the intended referent (object or action) immediately after hearing the language model. Children in the absent referent condition experienced the same non-verbal scripts and language models, but never saw the referent object or action after hearing the language model: at the appropriate juncture in the script they were told that the toy barn in which the target object had been previously located was 'locked', or that the toy character who had previously performed the target action was missing. Comparisons with two control conditions indicated that children were able to learn words for a novel object and a novel action in both the referent and absent referent conditions and, moreover, that learning was equivalent in these two conditions. These results show quite clearly that early lexical acquisition does not depend on temporal contiguity between word and referent-or indeed any perceptual pairing between word and referent at all-but rather it relies on children's active understandings of a speaker's referential intentions in particular discourse contexts.