In Favor of [Fortis]: Evidence from Setswana and Sebirwa

This paper argues that in favor of adding the feature [fortis/lenis] to the set of generally accepted phonological features, as originally proposed by Trubetzkoy (1939).  Phonetic data from the consonant systems the Southern Bantu languages Setswana and Sebirwa is presented.  Measurements of consonant closure and voicing duration in Setswana show that post-nasal “strengthening” and the distribution of ejective stops in this language can best be accounted for phonologically with the feature [fortis] rather than [+/- voice] or [constricted glottis].  In contrast, the phonetic data show Sebirwa to be a “true voice” language, demonstrating that the patterns of Setswana are not phonetically inevitable.