Consonants voicing in Surgut Khanty based on electroglottography data

Abstract

This paper presents an experimental phonetic study of consonant voicing in Surgut Khanty conducted with one na-tive speaker of the Trom-Agan sub-dialect using a Rose Medical EGG-D200 laryngograph and icSpeech software. The vocal cord activity was assessed non-invasively with sensors on the speaker’s throat. The recordings were seg-mented and annotated via Praat software and then statistically processed with the Emu-SDMS corpus manager and R programming language. A voicing rate was calculated as the relative duration of the sound segment defining the fundamental frequency. Over 770 sound segments were processed. The obstruent consonants /p/, /t/, /c͡ç/, /k/, /s/, /ɬ/, and /ʎ̥/ were pronounced voicelessly, partially voiced in intervocalic position but never having sonants’ and vowels’ typical values. The obstruent consonants had the voicing coefficients of 0–0,44, with the mean values for different phonemes of 0,02–0,06, while vowels feature the voicing coefficients of 0,44–0,89 and the mean values of 0,53–0,82. The sonor consonants /m/, /n/, /r/, /j/, /ɣ/, /ɲ/, /ŋ/, and /w/ were pronounced as voiced in the initial and medial positions. In the final position, the sonor consonants were realized as devoiced or voiceless depending on the syntagmatic conditions, with the voicing coefficients of 0–0,83 and average values of 0,4–0,63. The paper provides oscillograms and glottograms for the sounds investigated. A control evaluation was conducted using the acoustic data. The comparison of the consonant voicing data from the audio recording and glottography revealed only the latter method to accurately detect a boundary of the voiced segment.