When semantics and phonology collide: Gender assignment in mixed Tsova-Tush–Georgian nominal constructions.

Aims and objectives/purpose/research questions: This paper investigates the gender assignment strategies employed when genderless Georgian nouns are inserted into gendered Tsova-Tush utterances. We explore the linguistic and extra-linguistic factors motivating the strategies, and compare how these code-switches behave in relation to loans. Design/methodology/approach: Taking a broadly usage-based approach, we collected three types of data: (a) naturalistic corpus data; (b) semi-naturalistic production data from a forced-switch director–matcher (DM) task; and (c) a three-response forced-choice acceptability judgement task (AJT). Data and analysis: The responses from the DM task (n = 12) and AJT acceptability (n = 12) were analysed using descriptive (Chi-square) and inferential (log-linear) statistics. The corpus data are described qualitatively. Findings/conclusions: Both the gender of the Tsova-Tush translation equivalent (TE) and the Georgian phonology of the code-switched noun were significantly related to the response, with the TE being the stronger determinant of the two. Only marginal evidence for a default strategy was found. Production responses were found to be more consistent than comprehension responses, with more frequent lexemes displaying higher inter-participant consistency in production. Originality: Tsova-Tush, an endangered Nakh–Daghestanian language with five genders marked by prefixes, offers much-needed diversification within the code-switching literature concerning grammatical gender. This complexity also raises new questions regarding the notion of default in mixed nominal constructions. Significance/implications: Our findings support the prediction that first language speakers of a gendered language prefer a TE strategy, but contradict a relationship between default strategy and language dominance. Phonological criteria display a stronger role in gender assignment than previously found. Frequency and entrenchment of gender–noun pairings partially explain inter-speaker and inter-stimulus variation and consistency, providing a plausible pathway from code-switches to borrowings. Limitations: An unavoidable limitation is the sample size, reflecting the small speaker population. We strongly advocate for similar research in other language pairs in the Caucasus where gender systems feature prominently. [ABSTRACT FROM AUTHOR]