The Neutralization of the Voice Quality Distinction in Dinka Songs: A Production and Perception Study

The purpose of this study is to determine if the phonemic voice quality distinction of Dinka is also conveyed in songs. Acoustic measurements (formant tracking and spectral regression) were applied for the first time to a song data set. The results clearly pointed towards a neutralization of the voice quality distinction in songs. A perception experiment was performed to back up the results of the production study. Listeners were first trained to distinguish between modal and breathy vowels in speech, and then tested on speech and song data. The experimental results corroborate the claim that no voice quality distinction is conveyed in singing. Given the absence of any significant acoustic cue shown by the acoustic analysis or detected by the listeners, the author concludes that context seems to be the only means to disambiguate homophones in songs.