A REINTERPRETATION OF LOWER-VOCAL-TRACT ARTICULATIONS IN CAUCASIAN LANGUAGES1 Alexandre Arkhipova,b*, Michael Danielc**, Oleg Belyaevb,d,e, George Morozc** and John H. Eslingf aHamburg University, Germany; bLomonosov Moscow State University, Russia; cLinguistic Convergence La- boratory, HSE University, Moscow, Russia; dInstitute of Linguistics, Russian Academy of Sciences, Moscow, Russia; ePushkin State Russian Language Institute, Moscow, Russia; fUniversity of Victoria, Canada [email protected], [email protected], [email protected], [email protected], [email protected] ABSTRACT the tongue root closed against the rear pharyngeal wall. Catford’s term, ‘epiglottopharyngeal’ [3] im- The articulatory interpretation and transcription of plies stricture deeper than the lingual pharynx. Later lower-vocal-tract sounds in Caucasian languages has laryngoscopic research identified the sphincteric ac- long been a source of confusion. The Russian litera- tion of the aryepiglottic folds in airway closure [28]. ture, notably by Kibrik and Kodzasov, provides in- Roach observed that some glottalized consonants are sight into the auditory distinctions in Archi, Agul, and ‘made with closure not only of the true vocal folds but Dargwa, among others. However, the model of vocal also of the false vocal folds and the aryepiglottic tract articulatory function and the experimental tech- folds’ [25]. Gauffin specified that protective airway niques available in their time lacked precision in iden- closure by sphincteric laryngeal tightening constricts tifying the laryngeal and pharyngeal strictures and ‘larynx tube opening’ [14]. These lines of evidence movements responsible for laryngeal/pharyngeal point to full aryepiglottic closure as the maximum sounds. We reinterpret the pharyngeal, epiglottal, and stricture in swallowing or gagging, and as a speech pharyngealized contrasts in Caucasian languages us- sound. Catford termed this full closure a ‘pharyngeal- ing a model of laryngeal articulation that specifies de- ized glottal stop’ or ‘strong glottal stop’ in languages grees of laryngeal constriction, vibratory effects, and of the Caucasus, a ‘pharyngeal stop’ in Chechen [5], larynx height parameters. We focus on Archi, Agul, and a ‘ventricular (plus glottal) stop’ [4]. and Dargwa languages Mehweb and Shiri. Phonetic categories in the earlier inventories are assigned re- 1.2. A revised lower-vocal-tract model mapped articulatory definitions. The Russian re- searchers’ observations of degrees of epiglottis low- Phonetically we term this full engagement of the lar- ering are seen as an indication of the laryngeal artic- yngeal sphincter an epiglottal stop. Experimental in- ulatory mechanism constricting, with larynx raising, vestigations of the lower vocal tract over the past 20 to form constriction. years have resulted in a model that defines the articu- latory functioning of the laryngeal constrictor mech- Keywords: Caucasian, laryngeal, pharyngeal, epi- anism, with degrees of closure of its various parts glottal, pharyngealization. from open to fully closed [8, 9]. The model defines a range of manners of articulation: fricative, approxi- 1. LOWER-VOCAL-TRACT PHONETICS mant, tap, trill; glottal stop, ventricular stop, epiglot- tal stop; concomitant phonatory vibrations; and mod- 1.1. General lower-vocal-tract research background erating larynx height adjustments to alter resonance quality. Rather than specifying different places in the Early work at UCL [17] drew a relationship between pharynx where articulations are made, this scheme increasing constriction and ventricular activity, and identifies a single lower-vocal-tract stricture point, Sapir & Swadesh described Ahousaht Nuuchahnulth compressing and eventually shutting for stricture and pharyngeals as ‘laryngealized glottals’ [26]. Jacobsen unfolding for opening. We use this mechanism in re- [16], citing Sapir, called Nuuchahnulth /ʕ/ a ‘pharyn- interpreting the sound-symbol associations in the gealized glottal stop’. Gaprindashvili [13] described Caucasian phonetic inventories. a ‘pharyngealized glottal stop’, and Kodzasov [21] described a ‘strong glottal stop’ in Nakh-Daghesta- 2. CAUCASIAN PHONETIC DESCRIPTIONS nian languages. All of these descriptions imply a component added to glottal stop somewhere above 2.1. Early Russian research on Caucasian phonetics the glottis. Hockett [15] recognized that ‘a complete closure can be made in the lower pharyngeal region’, Russian research has identified lower-vocal-tract and characterized /ʕ/ as a ‘pharyngeal catch’ in Ara- consonantal articulations, syllabic properties, and bic, in parallel to ‘glottal catch’ [ʔ], but he assumed secondary vocalic qualities that characterize the lan- was viewed as ambiguous, since different authors ap- guages of the Caucasus [18, 19, 20]. Kibrik and Ko- plied it variably to zone A, Ƃ or B. This is a telling dzasov’s schema (Table 1) compares glottal state with observation, suggesting that the pharyngealization observed epiglottal lowering (and lingual retraction). mechanism was not yet well understood but that its They list 20 languages having a 4-way laryn- auditory presence pervaded all lower-vocal-tract geal/pharyngeal opposition: /ʔ h ʕ ʜ/. Budukh adds zones that existed at the time. /ʡ/; Burshag and Burkikhan Agul add /ʡ Ř x̌/; Richa Voiced [ɦ] in row 1 and [ɦ̶] in row 2 were identi- Agul adds /Ř x̌/. Each phonemic category is given an fied as possible voiced variants of ‘aspiration’ (of [h] c 2 expanded, qualified phonetic description for each lan- and [h̶], respectively). The voiced [ ] in row 2 was guage. described as a glide, while all three elements in row 3 Experimental methods to image the lower vocal were classified as ‘obstruents’. [ʕ ʜ] were identified tract were not available to assist in the identification as ‘spirants’ (i.e. fricatives). It is safe to say, based on of categories established in early Russian research. Kodzasov’s and Catford’s writings, that rows 2 and 3 The schema in Table 1 presumed the epiglottis to be were viewed as fundamentally glottal-laryngeal, with an indicator, if not an articulator. Its increasing low- observed increases in ‘strength’ of lingual and epi- ering below ‘upper’ (uvular) indicated tightening in glottal retraction/lowering corresponding to audito- the pharynx, together with tongue backing for ‘mid- rily ‘deeper’ sounds. Row 3 values were held to differ dle’ (pharyngeal sounds). The assumption was for in degree from row 2 values, potentially varying allo- ‘lower’ (epiglottal) sounds to be accompanied by gen- phonically but not contrasting phonemically in any eral tension in the pharyngeal walls, perhaps with one language. In Burshag and Richa Agul (also generalized laryngeal lowering. Inkhokvari and Chirag Dargwa), a laryngeal quality accompanying vowels and sonorants in certain lexical Table 1: Kibrik and Kodzasov’s diagram and items was called ‘compressed voice’, described audi- table (translated) for describing Caucasian ‘lar- torily as ‘hoarse-gravelly’ («сиплый») and ‘tense’ yngeal sounds’: row 1: ‘glottal laryngeals’, [18, 20]. rows 2-3: ‘emphatic laryngeals’ [18, p. 312]. In Burkikhan Agul [18, p. 340], ‘emphatic laryn- geals’ [ʕ ʜ] are realized in all positions within a word Articulation zones of the pharynx as ‘obstruent spirants’; [Ř x̌] originate from pharyn- A = upper [zone] (uvular sounds) Ƃ = middle [zone] (pharyngeal sounds) gealized uvulars, ‘middle’ [Ř] effectively alternating B = lower [zone] (epiglottal sounds) ➝ with ‘lower’ [ʕ]; and [x̌] is produced with more noise ➝ rows 2 and 3 in table below than [Ř] [20]. Voiceless [x̌] therefore fills the slot of / /, equivalent to IPA [ħ], and in contrast to /ʜ/. Vocal folds Closed Adducted Abducted h̶ Burkikhan Agul in the UCLA archive [27] includes Epiglottis sounds transcribed as /ʕ ħ ʜ ʡ/: [ʕ] as a voiced phar- Neutral position ʔ h yngeal fricative (or approximant, since [ħ] is called a c Moderately lowered ʡ h̶ voiceless pharyngeal fricative); [ʜ] as an epiglottal fricative; and [ʡ] as an epiglottal stop; although not all Strongly lowered ʔ˭ ʕ ʜ distinctions have been attested as phonemic in recent For all languages, [ʔ h] were defined as ‘plain Russian fieldwork. Generally, in Northeast Caucasian (glottal) laryngeals’. [ʔ] could be ‘strongly occluded’ languages, distinctions are not so much a phonemic or ‘weakly occluded’. ‘Pharyngeal’ and ‘epiglottal’ issue as whether there is pharyngealization on the root places of articulation were identified for many lan- and whether the word is native or borrowed. In a phar- guages, and uvulars were viewed as having more or yngealized root, the more extreme phonetic variants less deep localization in the upper pharynx, some- from Table 1 are realized. These may include [ʡ] for the stop and [ʜ] for the fricative; although no [ʢ] sym- times replaced by or varying with pharyngeals. [Ř x̌], specific to the three Agul dialects, were defined as bol was included in Table 1 in parallel to [ʜ] and in pharyngeal spirants (fricatives), produced by narrow- contradistinction to [ʕ ħ]. ing the pharyngeal passage at the level of the tongue 2.2. Recent Russian research on Caucasian phonetics root, presciently described in 1986 as ‘a sphincteric compression of the pharynx’ as well as ‘a strong re- Archi [1] has glottals /ʔ h/ in plain roots (/baʔbos/ ‘to traction of the tongue body’ [20]. ‘Epiglottals’ (tradi- kiss’, /barhas/ ‘to babysit’) and in pharyngealized tionally ‘emphatic laryngeals’) were seen as produced roots (/ʔiˤʔuˤbos/ ‘to crow’, /haˤht’i/ ‘breath’), where