Towards a usage-based model of early code-switching:evidence from three language pairs

Abstract Usage-based studies trace children’s early language back to slot-and-frame patterns which dominate spontaneous language use. We apply the Traceback method to data from three bilingual children with English as one of their languages and Polish, German, or Finnish as the other to examine what these children’s code-switching has in common and how it differs in light of the genealogical distance between the languages used. Their bilingual constructions are derived from individual corpora of naturalistic interactions of each child respectively and traced back to monolingual language produced previously to establish whether they are unprocessed chunks or partially schematic units. Based on this, we propose a model of switching which helps us to distinguish between the qualitative aspects of bilingual use in these two types of combinations. Our results show that all three children filter out some mixing occurring in chunks before these give basis to longer units. Whatever bilingual combinations remain frozen in those units can be explained by phonological overlap of the children’s two languages, which is highest in the acquisition of English-German due to their genealogical proximity