The good, the bad, and the gradient: The role of losers in code-switching

Although formal analyses of code-switching have enjoyed some success in determining which structures and interfaces are more fertile environments for switches than others, research exposing recalcitrant counter-examples to proposed constraints and axioms responsible for governing code-switching is abound. We advance the claim here that sub-optimal representations, i.e., losers, stand to reveal important information regarding the interaction of grammatical principles and processing strategies of bilingual speakers and that any comprehensive analysis of code-switching phenomena should include them. These losers are the result of gradient activation in both input and output forms. We demonstrate how the formalism Gradient Symbolic Computation (GSC; Smolensky et al., 2014) can account for both of these observed facets of bilingual grammars in a unified manner. Building upon the work of Goldrick et al. (2016a,b), we provide an analysis of mixed determiner phrases (DPs) as an example of the fundamental components of a GSC-analysis.