The interaction of lexical and phrasal prosody in whispered speech

The production and perception of Dutch whispered boundary tones, i.e., phrasal prosody, was investigated as a function of characteristics of the tone-bearing word, i.e., lexical prosody. More specifically, the disyllabic tone-bearing word also carried a pitch accent, either on the same syllable as the boundary tone (clash condition), or on the directly adjacent syllable (no clash condition). In a statement/question classification task listeners showed moderate, but above-chance performance for both conditions in whisper, which, however, was much worse as well as slower than in normal speech. The syllabic rhymes of speakers' productions were investigated for acoustic correlates of boundary tones. Results showed mainly secondary cues to intonation, that is, cues that are present in whisper as in normal speech, but minimal compensatory cues, which would reflect speakers' efforts to enhance their whispered speech signal in some way. This suggests that multiple prosodic events in close proximity are challenging to perceive and produce in whispered speech. A moderate increase in classification performance was found when that acoustic cue was enhanced that whispering speakers seemed to employ in a compensatory way: changing the spectral tilt of the utterance-final syllable improved perception of especially the poorer speakers and of intonation on stressed syllables.