The perception of breathy voice in Gujarati

Abstract

Breathy voice, also referred to as murmur, is characterized by an increased open quotient and increased airflow. Acoustical correlates of breathiness include an increased spectral slope, most notably as measured by H1-H2 and H1-A3, as well as an increase in noise as measured by harmonics-to-noise ratio and cepstral peak prominence. Gujarati is relatively unique in that it contains phonemic breathiness in both consonants and vowels, creating meaningfully different Cɦ V and CV̤ sequences. Acoustic analysis of this contrast indicates that while the acoustic correlates of breathiness are consistent across these two sequences, there are differences in both the timing and magnitude of the cues of breathiness (Esposito and Khan 2012). Perceptual studies have indicated that speakers of languages with phonemic breathiness are capable of reliably differentiating breathy and modal sounds, but no study to date has addressed the perceptual salience of the timing and degree differences associated with breathiness in consonants versus vowels. The present study addresses the potential confusability of Cɦ V and CV̤ sequences in Gujarati through a series of perception tasks. Results indicate that there is an asymmetrical relationship between Cɦ V and CV̤ sequences that may be attributable to differences in the magnitude and timing of acoustic cues.