On the gradience in perceptibility of word-final voicing contrast in Russian

Abstract

Russian is one of the languages in which the underlying voiced obstruents devoice in word-final position, resulting in voicing neutralization (e.g., /rok/ [rok] “fate” vs. /rog/ [rok] “horn”). However, recent studies have shown that word-finally devoiced (i.e., underlying voiced) obstruent and the underlying voiceless one are acoustically different (Chen 1970, Dmitrieva et al. 2010, among others). That is, word-final devoicing shows a case of incomplete neutralization. Also, those acoustic differences are in some degree perceptible for listeners (Matsui 2011, Kharlamov 2012). This paper presents a perceptual analysis of incompletely neutralized obstruents in Russian. Pseudo-nouns produced by the native speakers of Russian were presented to the listeners as auditory stimuli. Sixteen native listeners identified what they heard in a forced-choice identification task. The most striking result to be reported in this paper is that the listeners' sensitivity to the underlying voiced and voiceless stimuli is different between obstruent types: the underlying voicing contrast in stops is harder to perceive than that in fricatives for listeners. Other than obstruent type, the effects of the stimuli presentation type and of the magnitude of the acoustic difference will also be discussed.