Lexical Contributions to Inflectional Variability in L2 Predictive Processing

Grammatical gender is one of the most difficult aspects to acquire in a latelearned second language (L2), especially for learners whose first language does not have grammatical gender, e.g. English (Franceschina, 2005). There are two aspects of gender that present challenges to learners: First, learners need to learn the target gender of individual nouns, i.e. they need to assign a noun (e.g. Spanish casa – ‘house’) to a particular gender category (e.g. feminine). In this regard, L2 learners often assign a non-target gender to an item or they waver between different gender classes for a specific item, i.e. gender assignment is indeterminate. In short, L2 learners display lexical variability in gender assignment. Second, learners need to acquire morphosyntactic agreement properties of grammatical gender. Gender marking is (also) realized on lexical items that are grammatically dependent on a head noun, e.g. determiners and adjectives (e.g. la casa roja – ‘theFEM houseFEM redFEM’; Corbett, 1991). Adult L2 learners frequently omit gender agreement marking in production or they are not sensitive to gender violations in comprehension. In short, L2 learners display syntactic variability in gender agreement. In this study, we explore whether lexical and syntactic variability in L2 gender are causally related. We focus on real-time processing, as inflectional variability is more pronounced in on-line production and comprehension than in untimed judgment tasks (e.g. Grüter, Lew-Williams & Fernald, 2012). In principle, there are three different ways in which lexical and syntactic variability may be related in (predictive) L2 gender processing. First, they may constitute different problems, with syntactic variability representing non-target syntactic computation and lexical variability denoting non-target item-based lexical learning. This view is traditionally adopted in studies on syntactic variability in gender processing (e.g. Sabourin & Stowe, 2008). Methodologically, this means that studies only analyze the trials for which L2 learners had assigned the target genders. In essence, this approach then abstracts away from lexical variability in testing for syntactic variability with L2 gender.