LOCATION CHANGE – Campbell Hall 2122A and Zoom: https://ucla.zoom.us/s/91305505603

In the bilingualism literature, most agree that the bilingual lexicon contains elements from both languages that can be activated non-selectively during processing (Kroll et al., 2013). Popular models of the bilingual lexicon often take an emergentist approach, such as Dijkstra & Van Heuven’s (2002) BIA+ model, drawing direct connections between form and meaning. A strong interpretation of such models predicts language switch effects during processing due to a change in phonological form. Studies that manipulate language switching between non-composed words to understand the nature of language control often observe processing delays (e.g., Meuter & Allport, 1999) and neural activation differences (e.g., Abutalebi et al., 2008). However, manipulating language switching between composed words fails to elicit processing delays/activation differences (e.g., Phillips & Pylkkänen, 2021). I argue that syntactic/semantic composition eliminates language switching effects by increasing the predictability of input via morphemes. By taking a syntagmatic approach to bilingual input, I will present recent, ongoing, and planned work that (i) utilizes artificial language learning to develop a working model of how bilinguals develop morphemes from two languages, and (ii) tests this working model in bilingual-developing children.