The phonetics of contrastive phonation in Gujarati

Abstract

The current study examines (near-)minimal pairs of breathy and modal phonation produced by ten native speakers of Gujarati in connected speech, across different vowel qualities and separated by nine equal timepoints of vowel duration. The results identify five spectral measures (i.e. H1–H2, H2–H4, H1–A1, H1–A2, H1–A3), four noise measures (i.e. cepstral peak prominence and three measures of harmonics-to-noise ratio), and one electroglottographic measure (i.e. CQ) as reliable indicators of breathy phonation, revealing a considerably larger inventory of cues to breathy phonation than what had previously been reported for the language. Furthermore, while the spectral measures are consistently distinct for breathy and modal vowels when averaging across timepoints, the efficacy of the four noise measures in distinguishing phonation categories is localized to the midpoint of the vowel's duration. This indicates that the magnitude of breathiness, especially in terms of aperiodicity, changes as a function of time. The current study supports that breathy voice in Gujarati is a dynamic, multidimensional feature, surfacing through multiple acoustic cues that are potentially relevant to the listener.