108 Thursday, March 15: Poster Abstracts Number agreement and attraction in late Italian-English bilinguals

some circumstances, their production is affected by syntactic properties of the language they are not currently speaking (non-current language), suggesting the current language is not always the most strongly activated language [1]. But under what circumstances does this occur, and what can this tell us about bilingual processing? We report an experiment that investigated whether co-activation of languages affects the computation of bilingual agreement in bilingual production. Moderate/high proficiency late Italian-English bilinguals produced English and Italian completions for singular complex subject NPs involving English and/or Italian head and local nouns (1a-h). We manipulated current language (i.e., language of completion; English vs. Italian), language task (1- vs. 2-language), and local noun number (singular vs. plural). Previous research in monolingual production suggests the local noun is normally “insulated ” from the agreement between the head noun and verb, but sometimes this insulation breaks down, yielding attraction errors (e.g., The report of the forest fires were unbelievable). We examined whether such errors also occur in bilingual production, and if so, the factors that affect them. Following earlier work, we expected more errors following plural local nouns, but that this effect would be stronger in two-language production (i.e., more errors in (1d) than (1b) and in (1h) than (1f)), because the current language would be more strongly activated than the non-current language, yielding a weaker influence of the head noun (and a correspondingly stronger influenc