19th November SSF - Jalal Al-Tamimi

Talk Title: The role of the epilarynx in the production of guttural consonants in Levantine Arabic: An Ultrasound Tongue Imaging study

Abstract:

In this presentation, I will be evaluating the role of the epilarynx in the production of guttural consonants (i.e., uvular, pharyngealized, pharyngeal and glottal) in comparison with plain coronal consonants. We will look at the impact of larynx raising and tongue retraction, as a combined gesture, in explaining the gradience in epilaryngeal constriction. Following the “Laryngeal Articulator Model” (LAM, Esling, 2005; Esling, Moisik, Benner, & Crevier-Buchman, 2019), an epilaryngeal constriction leads to a raised larynx posture that induces retraction of the tongue due to constriction of the hyoglossus muscle, which pulls the tongue in a combined back and down gesture. Epilaryngeal consonants are expected to have this type of constriction, whereas pharyngeal/pharyngealized are expected to show partial “retraction” and uvular show nil “retraction”. The study looks at synchronised recordings of Ultrasound Tongue Imaging (UTI), Electroglottography (EGG) and acoustic data obtained from 10 Levantine Arabic speakers (5 males, 5 females) producing 21 consonants in Arabic in a /ʔVV1CVV2/ frame with three subsequent repetitions (VV = symmetric /iː aː uː/; C = plain /t d ð s z l/, velar /k ɡ x ɣ/, uvular /q/, pharyngealized /tˤ dˤ ðˤ sˤ zˤ lˤ/, pharyngeal /ħ ʕ/ and glottal /h ʔ/; n = 2034 items). Dynamic (multiple intervals within VCV) and static analyses are evaluated. We use GAMMs for UTI data, Convolutional Neural Networks (CNN) and random forests for acoustic data (bark-difference and voice quality metrics) and larynx height and constriction for EGG data. The results point to the use of tongue retraction as a gradient rather than categorical, in addition to changes associated with voice quality in gutturals. I discuss the implications of these results in the lights of LAM.