SY_21.3 - The emergence of lexical networks: Insights from TRACE simulations

Mayor, J.

Basque Center on Cognition, Brain and Language, Donostia, Spain

Recent evidence reveals that two-year-olds display effects of phonological priming (Mani and Plunkett, 2008), similarly to adults. When infants are primed with a picture whose name shares the same onset as the target (e.g., “bed” and “boot”) their looking time to the target is reduced, as opposed to when the prime is unrelated (e.g., “cat” and “boot”). This inhibition effect to be modulated by the neighbourhood size of the words tested, suggesting a lexical basis for the reported effects. In line with the successful comparison of TRACE with adults' looking behaviour described in Allopenna et al. (1998), we model phonological priming experiments in infancy. TRACE's mental lexicon was created by compiling typical lexicons for 24-month-olds from the British CDI, using token frequencies from the CHILDES database. In a first set of simulations, Mani and Plunkett's (2008) study is modelled successfully; priming effects interact with cohort sizes. In the presence of lexical competition, priming can be either facilitatory or inhibitory, depending on neighbouring size. In contrast, when lexical competition is removed, only facilitation can be observed. We argue that only inhibitory effects in phonological priming can be taken as evidence for lexical competition whereas facilitation effects can be driven by sub-lexical priming effects both with and without lexical competition. A second set of simulation captures White and Morgan's (2008) findings that 19 month-olds displayed a graded sensitivity to the severity of word mispronunciations. Crucially, TRACE only displays a graded sensitivity to mispronunciation severity, in absence of controlled cohort sizes, when lexical competition is suppressed.
Together, these simulations suggest that lexical competition is absent at 19 months of age whereas it is present from 24 months of age. Further re-analyses and simulations will aim at identifying more precisely the age at which words start competing for recognition.