Predictive Sentence Parsing in Monolingual and Bilingual Learners.

Abstract

There is now substantial empirical support for incremental models of speech processing where language is processed in real-time as it unfolds. Developmental studies indicate that very young children process speech incrementally, demonstrating the ability to "listen ahead." The predictive use of linguistic information to anticipate upcoming words has primarily been studied in monolingual children from widely represented linguistic communities (e.g., English speakers). In this study, bilingual Mandarin-English-learning and monolingual English-learning Singaporean children were tested on their use of verb information to predict upcoming food and body part nouns, including in the presence of mispronunciations of the same nouns. When nouns were preceded by a constraining verb, children's target fixation increased. However, overall target fixation remained higher for correct pronunciations than for mispronunciations, suggesting that children were also sensitive to mispronunciations. Further, constraining verbs facilitated children's target fixation even in the presence of speech errors. The difference between target fixation on correct pronunciation compared to mispronunciation trials was smaller in bilinguals than monolinguals when verbs were present, and this effect appeared at a later time window across trials for monolinguals than it did for bilinguals, suggesting earlier resolution of conflicting linguistic cues in bilinguals. Findings contribute to growing knowledge about how language experience interacts with the temporal dynamics of incremental sentence processing in young learners.