The recent cognitive developmental literature on rapid word learning, u sing either preferential looking or reaching methodologies, suggests that 12- to 18-month-old infants can learn novel label-object listener relations after as few as five exposures to a novel word-object relation. But methodological limitations of such studies render their conclusions equivocal; several studies fail to control, for example, for infants' prior learning of a behavioural preference for labelled objects, and several fail to report the extent to which individual performances are consistent with group findings. Further, none of these studies investigated the effectiveness of contingent positive reinforcement as an alternative means of establishing novel listener behaviours in preverbal infants.
Experiments 1 and 2 were conducted to reassess the effectiveness of exposure in establishing novel label-object listener relations under carefully controlled conditions, using both looking and reaching measures of comprehension. Contrary to previous findings, in Experiment 1, 13- and 17-month-olds did not learn to comprehend novel object names after up to 12 exposures to each of two novel objects accompanied by its verbal label. Only after between 18 and 24 such exposures in Experiment 2 was there evidence of listener learning in young infants on the looking measure.
Experiment 3 compared the effectiveness of positive reinforcement and exposure in infants' listener learning. Here, 13- and 17-month-olds, on the looking measure, showed evidence of learning novel label-object listener relations after up to 10 reinforced training trials, but not after up to 10 exposure training trials. Results with the reaching measure were equivocal in all three experiments.
These findings provide initial support for the effectiveness of positive reinforcement, and question the role of exposure, as a means by which 13- and 17-month-old infants can be taught to comprehend novel object-labels after very few training trials. Further investigation of these findings is merited, particularly within a longitudinal, singlesubject research paradigm.