Despite the number of studies and overviews of Chicano English (ChE) that have appeared in recent years, many ChE features await full investigation. Final (z) devoicing is one such feature. This study, based on multivariate analysis of 1,827 tokens extracted from sociolinguistic interviews with adolescents and young adults in a public housing project in south Texas, shows that (z) devoicing is highly systematic and subject to multiple linguistic and social constraints. Specifically, devoicing is conditioned by features of the preceding and following segments. Results also show that devoicing is constrained by the morphological status of the variable, with devoicing more likely for inflectional (z) than when it is part of a monomorpheme. In addition, although the prevalence of devoicing in ChE is often attributed to Spanish interference, results show that devoicing is not affected by the speaker’s first language. Finally, speakers who express a desire to leave the community are less likely to produce the devoiced variant than those who express no such desire, and women are more likely to devoice than men.
