Compensatory strategies for voicing of initial and medial plosives and fricatives in whispered speech in Dutch

In this study, the compensatory strategies speakers adopt for conveying the voiced-voiceless distinction of consonants in whispered speech were investigated. Of 8 native Dutch speaking subjects each, 26 Dutch minimal word pairs containing voiced and voiceless obstruents, were recorded. Acoustic analysis with the software Praat revealed that both in whisper and in normal speech (1) preceding vowels are longer for voiced obstruents, (2) duration of the silent interval and (3) of the burst are longer in voiceless obstruents and (4) the burst intensity in voiceless obstruents is greater. These recordings were subjected to a pool of Dutch listeners (N=16), judging for each of 1056 items if it was the voiced or the voiceless member of the minimal pair. This revealed no marked sensitivity of these contrasts in whispered speech. These results supports the Redundant Cue Hypothesis (RCH), stating that voicing perception in whisper depends mainly on the secondary voicing cues that remain in whisper.