Czech voiced labiodental continuant discrimination from basic acoustic data

The Czech voiced labiodental continuant has a special position within the system of Czech consonants. Due to its past it sometimes behaves phonologically as a sonorant, while its canonical articulatory qualities place it among obstruents. It is quite difficult to decide how much of its sound is sonorant and how much is obstruent by mere auditory analysis. We investigated whether basic acoustic parameters extracted from fluent reading material can help to discriminate if the Czech [v] leans more towards the obstruent or the sonorant class of consonants. Dynamic profile data were processed by factor analysis and, together with duration and harmonicity measurements, used as an input for cluster analysis, discriminant analysis, and classification-tree analysis. The results show that acoustically, the Czech intervocalic [v] in connected speech is more in line with its sonorant past rather than its obstruent presence.