Repnik, K. .
The University of Edinburgh
During word acquisition, children might use certain language-specific learning constraints, such as disambiguation. Hereby, word learning is said to be guided by the assumption that new words tend to refer to new referents. Children assume that each object only has one label. This allows to build up a vocabulary without overlapping meanings.
Associative network models describe word learning with domain-general processes that solve referential ambiguity whilst also considering ecological and environmental factors. They describe retention of words as gradual with word links being strengthened via statistical learning. Unlike traditional accounts, word learning is not regarded as sequential, but as a combination fast and slow mapping processes that operate on distinct but interactive timescales, which might impact learning simultaneously. Here, disambiguation is the consequence of an appropriately structured network rather than the reason for it. Interestingly, bilingual children make less use of disambiguation which leads to the question of whether they have a lesser structured network than monolinguals, or if their different structure leads to less use of disambiguation.
This poster will highlight results from on-going eye-tracking research with the aim to disentangle the concept of disambiguation by looking at retention and use of disambiguation in mono- and bilingual infants.