In this paper I argue for a syntactic analysis to gender assignment in codeswitched speech. To sustain this claim, I examine gender assignment in Spanish Det(erminer)–English noun switches (i.e., el bishop ‘the.m’) in 76 sociolinguistic interviews of approximately one hour each from a bilingual community in Southern Arizona, U.S. (The CESA Corpus, Carvalho 2012). Based on the findings from this dataset, I demonstrate that the distribution of gender assignment in codeswitched speech poses a serious challenge to current models of the bilingual architecture rooted in the distinct-lexicons perspective (MacSwan 2000 et seq.). Rather, I show that biological gender (interpretable gender) plays a crucial role in the assignment mechanism and the representation of gender features in the bilingual architecture. Taking gender assignment as a case study, I outline a single-lexicon approach to the bilingual grammar compatible with a Late Insertion view of the morphosyntactic model (Halle & Marantz 1993). In particular, I highlight the crucial relevance of a theme position at the morphological module to guide the bilingual speaker to the insertion of phonological matrices (language exponents) when codeswitching.