Santolin, C. 1, 2 & Saffran , J. 1
1 Waisman Center, University of Wisconsin-Madison
2 De Vincenzi Foundation & University of Trento
When infants begin to acquire grammatical patterns of their native language, they learn that words are grouped into categories and organized according to hierarchical structures. They also learn that word classes comprise more than one lexical item (e.g., English determiners could be a or the), and that the presence of a member of a class predicts the presence of a member of another word class, a key statistical cue to discover linguistic phrase structure. Previous evidence showed that 12-month-olds learn phrase structure from artificial languages, leaving open the domain-specificity of this process. The current study investigates the extent to which infants can track phrase structure from non-linguistic materials. A group of infants was familiarized with an artificial grammar including predictive dependencies between sound classes (P-Grammar), a second group of infants was familiarized with a grammar that did not comprise statistical dependencies (NP-Grammar). At test, all infants heard grammatical vs. novel ungrammatical sentences. Preliminary analyses reveal a preference for ungrammatical over grammatical strings after exposure to P-Grammar but not to NP-Grammar, suggesting that predictive dependencies may facilitate learning of phrase structure also in non-linguistic input. This result represents a promising starting point towards the discovery of the beginnings of grammatical acquisition