Language learning from inconsistent input: Bilingual and monolingual toddlers compared

Abstract

This study examines novel language learning from inconsistent input in monolingual and bilingual toddlers. We predicted an advantage for the bilingual toddlers on the basis of the structural sensitivity hypothesis. Monolingual and bilingual 24-month-olds performed two novel language learning experiments. The first contained consistent input, and the second occasionally contained inconsistent input (i.e., “errors”). Neither group showed learning of the novel pattern in the consistent experiment. The bilingual toddlers, but not the monolinguals, showed learning in the inconsistent experiment, which suggests they are better at detecting regularities from inconsistent input than monolinguals. Highlights: Language learning experiments consider consistent input, whereas inconsistent input is likely to occur in real life. Monolingual and bilingual toddlers' performance on a language learning task of non-adjacent dependencies was assessed, using inconsistent input. The group of bilingual toddlers was able to find the language pattern despite the inconsistencies, whereas the group of monolingual toddlers was not.