V2 all the way down

Overgeneralisation of Embedded V2 (EV2) has been reported for monolingual children, particularly of Scandinavian languages (e.g., Ringstad & Kush 2021, Westergaard & Bentzen 2007). Its existence in bilingual contexts remains highly understudied (see Müller 1994, et seq., for an exception). In this paper, we contribute to existing literature by presenting novel corpus data on 5 German–Italian bilinguals. All children, bar one, show systematic overgeneralisation of V2 in embedded contexts. We argue against existing accounts of comparable bilingual data based on parameter mis-setting and transfer (Müller 1994, 1996, 2003). We propose instead an analysis in terms of CP-complexification, drawing on Bhatt & Yoon’s (1992) two-way structural distinction among complementisers and on non-cartographic implementations of V2+ systems (e.g., Walkden 2017). The present data has implications for the ‘stability’ of V2, the crosslinguistic typology of (E)V2, and for contact-induced word-order change under child bilingualism.