Abstract

One of the most extraordinary aspects of human development is how children acquire their language(s) by listening to spontaneous speech. Perhaps more remarkably, they do so even though speech is often highly disfluent. To better understand how language acquisition unfolds, this dissertation explored the effects of speech disfluencies on real-time word comprehension and on word learning for listeners from different language backgrounds and levels of expertise: monolingual children, bilingual children, and bilingual adults.

Manuscript 1 reports two word comprehension studies, looking at the ability of children and adults to use disfluencies to predict whether a speaker will name a novel or a familiar object. This ability was investigated by presenting sentences with disfluencies in listeners’ native and non-native language(s). Study 1 tested 32-month-old monolingual and bilingual children, and Study 2 tested bilingual adults. Results from Studies 1 and 2 indicate that listeners looked more at the novel than the familiar object upon hearing a disfluency, irrespective of participants’ language experience, and whether the disfluency was in participants’ native language(s). Importantly, the results suggest that listeners might attend to a speaker’s uncertainty more than the particular realization of the disfluency.

Manuscript 2 investigates the impact of speech disfluencies on novel word learning in monolingual and bilingual 32-month-old children. We considered two contrasting possibilities: (1) Disfluencies will facilitate novel word learning, since listeners direct looks to novel objects upon hearing a disfluency, versus (2) disfluencies will hinder novel word learning, since they signal a speaker’s uncertainty about an object’s label. The results indicate that disfluencies may hinder novel word learning: Children did not learn the novel words following disfluencies, nor the novel words following fluent speech. Though somewhat inconclusive as children did not learn words in either case, these results suggest that children’s word learning may be hindered when a speaker is disfluent.

Together, the results from the two manuscripts in this dissertation suggest that speech disfluencies are a double-edge sword: they can be helpful for making predictions during real-time comprehension, but could hinder word learning. These findings have important implications for understanding the role of speech disfluencies in language acquisition.