This chapter introduces the separationist approach to bilingualism and shows that its foundations are shaky. More generally, Creole continua argue that thinking of languages as discrete entities is a mistake. Code-switching involves integrating different types of linguistic material and establishing dependencies among them. The chapter also discusses methodological issues and argues that acceptability judgments of deep bilinguals are a good method to obtain data.

Ths chapter introduces MacSwann’s (1999) model. It is a minimalist framework within separationist assumptions to the extent that it is claimed that the bilingual has two lexicons and two PFs. The chapter also presents two empirical challenges to this model: mixed selection and noun classes. Mixed selection refers to the empirical fact that an item from “one lexicon” may select for an item in “the other” lexicon. Under the label “noun classes,” I show that an English noun can be inserted into a Swahili discourse, in the process acquiring a noun class. Both these well-known features of code-switching are empirical problems for any theory that posits separate lexicons. The chapter ends with a brief discussion of Multiple Grammar Theory highlighting the points of contact and divergence with the integrationist approach.