Speaker variation and phonation type in Tsonga nasals

Tsonga has a phonological contrast between normally voiced and breathy voiced nasals. The latter are associated with large perturbations in F0 and appear to be subject to speaker variation involving the phonetic details of phonation type. The two questions of interest are, first, whether there are measurable properties of the nasals that will allow one to quantify the speaker variation in phonation type and, secondly, whether this variation is constrained by a principle of relative in variance or is random. The study suggests that there are reliable speaker differences based inter alia on spectral tilt and that most, but not all, subjects produce the phonological contrast between normally voiced and breathy voiced nasals in a relatively invariant way.