Comparing the acoustics of voiced and voiceless fricatives in Deg Xinag

Abstract

Few studies have looked at the acoustic properties of fricative voicing and place in Native American languages despite their relatively rich fricative inventories of rarely studied fricative places. Deg Xinag, an endangered Athabaskan language spoken in Alaska, provides us with a rare opportunity to investigate fricative place and voicing within a single language: it has eight places of articulation for voiceless fricatives, six of which have voiced counterparts, including some rarely studied place contrasts (e.g., palato-alveolar versus retroflex, uvular versus glottal, lateral versus alveolar). In this study, pre- and post-vocalic fricatives were digitally recorded in the field from eight speakers (two males, six females) using a head-mounted mic to control for distance from the source. The segmental context was also controlled for, the neighboring vowel being [a] in all cases. Each speaker produced four repetitions of each word. Each fricative was analyzed qualitatively using impressionistic transcription and spectrographic investigation, and quantitatively using a set of widely employed measures: (a) widely employed spectral measures (center of gravity, skew, kurtosis, standard deviation, lowest spectral peak), peak and rms intensity of frication, overall duration and duration of voicing. [Work supported by NSF.]