An important question that lies at the interface between phonetics and phonology is "To what extent do phonological alternations necessarily depend on phonetic principles?" Central to this question are debates on "post-nasal devoicing" in Setswana and related languages (Coetzee et al. 2007, Gouskova et al. 2011, Hyman 2001, Sole et al. 2010, Zsiga et al. 2006). As the alternation is usually described, voiced stops become voiceless after nasals, resulting in neutralization with underlying voiceless stops. This goes against a supposedly universal tendency for post-nasal stops to become voiced (Pater 1999). Recent phonetic studies, however, have found a great deal of variability in the realization of voiced and voiceless stops in Setswana (Coetzee et al. 2007, Coetzee et. al. 2010, Gouskova et al. 2011), raising questions of whether post-nasal devoicing is a) categorical, b) productive, and c) really devoicing. In this paper, we present the results of a perception study involving 10 Setswana listeners, which show that post-nasal realization is categorical, but not productive. We also present the results of an acoustic experiment involving 27 Setswana speakers producing voiced and voiceless stops in phrase initial, intervocalic, and post-nasal position. Acoustic analysis of these results revealed that the change in post-nasal position is not one of devoicing, but showed that there is phonetic post-nasal voicing, consistent with universal tendencies. Consequently, we conclude that there is a categorical post-nasal alternation in Setswana, but it must be characterized as fortition rather than devoicing.
Selected Proceedings of the 43rd Annual Conference on African Linguistics: Linguistic Interfaces in African Languages
edited by Ọ
la Orie and Karen W. Sanders
Table of contents