Cross-situational learning is a basic mechanism that enables people to infer the correct referent for a novel word by tracking multiple hypotheses simultaneously across exposures. Previous research has shown that adults are capable of exploiting cross-situational information, but recently this gradual statistical learning mechanism has been put under debate by researchers who argue that people learn via a fast mapping procedure. We compared the performance of adult participants on a word learning task in which information was manipulated cross-situationally with the performance of simulated learning strategies. Experimental evidence indicates that adults use cross-situational learning, which appears to be a robust mechanism that facilitates word learning even under cognitively demanding circumstances.

In this paper, we explore the cultural differences in theproduction of speech and speech+gesture combinations byinfants at the age of 17-18 months in Mozambique and theNetherlands. We found that Dutch infants produce morespeech and gestures compared to Mozambican infants. Infantsin both communities make most use of content words. Theresults further show that Dutch infants make more use ofproximal pointing than Mozambicans, whereas Mozambicansmake more use of the offering gesture. Finally, the amount ofsemantically coherent speech+gesture combinations of theMozambican infants is higher than of the Dutch infants

This study introduces a neural network that models thesocial interactions from a video corpus. The corpusconsists of recordings of naturalistic observations ofsocial interactions among children and theirenvironment. The videos are annotated multimodallyincluding features like gestures. We explore how thisvideo corpus can be utilized for modelling by trainingour model on a portion of the annotated data extractedfrom the corpus, and then by using the model to predictnovel interaction sequences. We evaluate our model bycomparing its automatically generated sequences to anunseen portion of the corpus data. The initial resultsshow strong similarities between the generatedinteractions and those observed in the corpus.