Incorporating Memory Processes in the Study of Early Language Acquisition

Critical to the learning of any language is the learning of the words in that language. Therefore, an extensive amount of research in language development has examined how infants learn the words of their language so rapidly. In particular, research on statistical learning has suggested that sequential statistics may play a vital role in the discovery of candidate words, that become available to be mapped to meaning. One important limitation of this previous research is the lack of attention given to the memory processes involved in statistical word learning. Thus, the current set of experiments examine the availability of statistically defined words as object labels after a delay. To examine whether statistics found in speech supports infants’ memory for label-object associations, in Experiment 1, 22to 24-month-old infants were presented with 12 Italian sentences that contained 2 high transitional probability words (HTP) and 2 low transitional probability words (LTP). Ten-minute after familiarization, using a Looking-WhileListening procedure (Fernald et al., 2008), infants were trained and tested on 2 HTP and 2 LTP label-object associations. Results revealed that infants were able to learn HTP but not LTP words, suggesting that HTP words make better labels for objects after a minimal delay. Experiment 2 examined infants’ memory for meaning representations that are statistically defined or not. Stimuli and procedure were identical to that of Experiment 1, except that the 10minute delay was implemented after the referent training phase instead of after the familiarization phase. Infants in Experiment 2 were able to remember both HTP and LTP words when tested following a 10-min delay. Together, the findings suggest that statistical learning facilitates future word learning.