No abstract

No abstract

Setswana, spoken by about 4.5 million people throughout Botswana, is well-known in the literature for “post-nasal devoicing,” in which /b/ and /l/ become [p]‌ and [t] after nasals, contra the expected, phonetically-grounded pattern of post-nasal voicing. Sebirwa, in contrast, has at most 15,000 speakers concentrated in the far eastern corner of the country. Sebirwa is being overwhelmed by Setswana, and in a process of “massive Tswananization,” has borrowed some aspects of post-nasal devoicing. Our analysis, based on fieldwork in the village of Molalatau, shows that the Sebirwa pattern is doubly unexpected: only /b/ devoices, not /d/ and /g/. We attribute the asymmetry to frequency effects from Setswana, where, due to a skewed voicing inventory, the majority of lexical items that exhibit the alternation have underlying /b/. We discuss the implications of this type of borrowing, both for the typology of alternations, and for patterns of language loss.