Gervain, J. 1, 2 & Werker, J. F. 3
1 CNRS-Universite Paris Descartes, Paris, France
2 Basque Center on Cognition, Brain and Language (BCBL). Donostia. Spain
3 University of British Columbia, Vancouver, Canada
We ask how 7-month-old bilinguals learn the word orders of their languages when these are in conflict, i.e. Verb-Object (OV) order in one language (e.g. English), VO order in the other (e.g. Japanese).
Frequency and prosody correlate with word order. In OV languages, frequent words, i.e. grammatical functors, appear in phrase final positions, whereas in VO languages, they occupy initial positions. Infants can use frequency to parse word order (Gervain et al, 2008). Similarly, while OV languages show prominence-initial prosody, realized as increased pitch/intensity, VO languages have prominence-final phrases implemented as increased duration (Nespor et al. 2009).
We created an artificial grammar using these cues. It comprised an alternating sequence of frequent and infrequent syllables, mimicking functors and content words, respectively. This way, the basic unit could be parsed as frequent-initial as well as frequent-final. For half of the participants, the stream was synthesized using OV prosody, i.e. increased pitch on infrequent syllables, whereas for the other half, we used VO prosody, i.e. longer duration on infrequent syllables. To test whether OV-VO bilinguals use prosody to bootstrap word order, we conducted a headturn preference study. After familiarization with the stream (with OV or VO prosody), infants were tested on frequent-initial (VO) and frequent-final (OV) test items (with flat prosody). Infants in the VO condition looked significantly longer at the VO items, while infants in the OV condition looked significantly longer at the OV items. To test whether this prosodic bias requires language experience (Iversen et al. 2008, contra Hayes 1995), in a third condition, we tested monolingual English infants in the OV prosody condition, i.e. using prosody unfamiliar to infants. Infants showed no significant difference in their looking behavior to the two types of test items.
Our results suggest an important role for prosody in bilingual grammar learning.