Franck, J. 1 , Lassotta, R. 1 , Omaki, A. 1 & Rizzi, L. 2, 1
1 University of Geneva
2 University of Siena
Children show sophisticated abilities to assign Agent and Patient thematic roles to arguments before age 2 (e.g., Hirsh-Pasek & Golinkoff, 1996). However, it remains unclear whether the successful comprehension is guided by the abstract phrase structure of the target language, or by children’s item-based, verb-driven knowledge (Tomasello, 2003). The present study focuses on word order, and provides novel evidence for abstract, grammatical representations at 19 months.
Previous research on the acquisition of word order is amenable to alternative interpretations. Gertner et al. (2006) tested how 21-month-old English-speaking infants interpret grammatical NP1-Verb-NP2 sentences containing pseudo-verbs. Infants demonstrated a preference for videos illustrating Subject-Verb-Object interpretations over those illustrating Object-Verb-Subject interpretations. They concluded that infants already represent the Subject-Verb-Object structure abstractly, independently of verbs. However, this could also be explained by a general preference for Subject-Object order characterizing most languages (Tomlin, 1986).
The present eye-tracking study, conducted with French-speaking 19-month-olds (N=17), aims to tease apart the two explanations. Our experiment presented both grammatical (NP1-Verb-NP2) and ungrammatical (NP1-NP2-Verb) sentences with pseudo-verbs, together with two causative action videos in which the same action is performed by two puppets but the Agent-Patient relation is reversed. The inclusion of an ungrammatical condition critically allows contrastive predictions for the two explanations. If comprehension is guided by abstract word order knowledge, infants should show a preference for the video illustrating the correct Subject-Verb-Object interpretation in the grammatical condition, but not in the ungrammatical condition. In contrast, if the universal Subject-Object preference guides interpretation, a preference for that video should emerge in grammatical and ungrammatical conditions alike.
In the grammatical condition, infants looked significantly longer at the video illustrating the Subject-Verb-Object interpretation, whereas no preference was observed in the ungrammatical condition. These results suggest that abstract phrase structure representations guide sentence comprehension in children as early as 19 months.