«A laryngoscopic study of glottal and epiglottal/pharyngeal stop and continuant articulations in Amis—an Austronesian language of Taiwan Jerold A. ...»
A laryngoscopic study of glottal and epiglottal/pharyngeal stop and continuant
articulations in Amis—an Austronesian language of Taiwan
Jerold A. Edmondson
University of Texas Arlington
Department of Linguistics and TESOL
Arlington, TX 76019-0559
John H. Esling and Jimmy G. Harris
Department of Linguistics
University of Victoria
Victoria BC V8W 3P4
Department of Indigenous Languages and Communication
Dong Hwa University
Hualien, Taiwan, Republic of China
1. Introduction. There are several themes running through this paper. The first is that sounds can be produced by glottal articulation, by epiglottal articulation (laryngeal constriction), and by epiglottal articulation with lingual pharyngealization (laryngeal constriction with substantial pharyngeal tongue retraction). A second claim emerges from the first, that glottal and epiglottal/pharyngeal locations of stricture represent a natural class of sounds that can enter into phonological rules. Finally, sounds with epiglottal and pharyngeal points of stricture are significantly represented in the Austronesian language family, just as they are in Semitic, Caucasian, and in the Salish and Wakashan languages of the northwest coast of North America. On the first theme,
Esling argued in his paper “The IPA categories “pharyngeal” and “epiglottal”:
laryngoscopic observations of pharyngeal articulations and larynx height” that there was not good evidence for “two distinct places of articulation in the pharynx”, i.e. epiglottal and pharyngeal. In Arabic, for example, some have described the “voiced resonant” sound of the lower throat as a fricative, others as an approximant, and still others as a stop (with the description of the location of stop stricture being ambiguous). Esling has argued that instead of place, distinctions in pharyngeal sounds are a function of manner of articulation or a function of larynx height. He makes this argument on phonetic grounds. John McCarthy in his paper “The phonetics and phonology of the Semitic pharyngeals” makes similar arguments on phonological grounds. Traditionally, even the reconstructed proto-phoneme set includes only the glottals /ʔ h/ and the “pharyngeals” /ʕ ħ/. The “pharyngeals” of Arabic and Oriental Hebrew had been described by Laufer and Condax (1981) as phonetically epiglottal, with the epiglottis as the active articulator.
McCarthy (nd:4) also says, “The main gesture in the production of the pharyngeals is an approximation of the posterior wall of the laryngopharynx and the tongue root from the epiglottis down to the larynx. Both the posterior pharyngeal wall and the tongue root are moved inward from their rest positions. Concomitantly, the larynx itself and adjoining structures are raised considerably. The constriction is significantly narrower for ħ than for
ʕ.” Continuing he speculates:
The voiceless pharyngeal ħ is some kind of fricative or approximant (or perhaps even a glide). The realizations of ʕ vary dialectally or even individually between a stop (presumably epiglottal) and an approximant or fricative. Al-Ani (1970) found that he and three Iraqi informants produced ʕ as a stop (cf. El-Halees 1985). …But, according to Catford (1983: 347), there is a Caucasian language ("the Burkikhan dialect of the Dagestanian language Agul") which does have distinct stop and continuant ʕ phonemes. Another possibility of phonemic contrast is that between plain and glottalized ʕ, found in Columbian Salish (Kinkade 1967). Further properties of ʕ and ħ involve the larynx. I have already noted that the larynx is considerably raised during the production of the pharyngeals (Ghazeli 1977), and ʕ is often accompanied by creaky voice.
This phenomenon is probably not unique to Arabic; Hayward and Hayward (1989), citing Sasse (1979) and Hayward (1989), note that ʕ is frequently "glottalized" in Ethiopian (Semitic and Cushitic) languages.
Keeping phonological and phonetic descriptions distinct, we will call ʕ a “voiced resonant” and ħ a “pharyngeal fricative,” and we recognize that different dialects of Arabic have different phonetic realizations of the “voiced resonant”. Our examinations of Arabic suggest that the “voiced resonant” may occur as a voiced approximant or as an epiglottal-pharyngeal stop. Esling has shown in his studies (1996, 1999) that in all of the so-called pharyngeal and epiglottal articulations, the epiglottis is the passive articulator and the aryepiglottic folds are the active articulator. Cf. the evidence of Zeroual (1999,
2000) and of Zeroual and Crevier-Buchman (2002) compared with the video images below.
Figure 1: Arabic /ʔi/ ‘yes (Palestinian)’ vs. /ʕi/ ‘understand’
In the Handbook of the IPA, Thelwall and Sa’adeddin (p. 53) describe /ʕ/ as a “Retracted Tongue Root glottal stop” and say, “Nowhere have we observed a pharyngeal fricative.” We agree that “fricative” is not a good description and instead would suppose that it is an approximant, possibly with a bit of friction due to slight trilling of the aryepiglottic folds.
Our Palestinian pictures showed little of the aryepiglottic folds, but auditorily it did make the impression of being trilled. Zeroual has shown that the Moroccan varieties of these sounds are a pure approximant (with the trilled enhancement in prosodically emphatic situations) and the point of articulation is aryepiglotto-epiglottal (2002). Single slides such as those in Figure 1 are not capable of showing the manner of articulation; that must be determined auditorily.
The idea of common pharyngeal-epiglottal linguistic terrain is also reflected in the IPA phonetic symbols. Also in that book, in the new revised symbol tables of the International Phonetic Association there has been a significant increase in the number of recognized sounds in the lower throat to five, so now the IPA now has number 144, a voiceless pharyngeal fricative [ħ], number 145, a voiced pharyngeal fricative or approximant [ʕ], number 172, a voiceless epiglottal fricative [ʜ], number 173, an epiglottal plosive [ʡ], and number 174, a voiced epiglottal fricative [ʢ], cf. Handbook of the IPA (1999:163).
This paper will be looking at the case of Amis, an endangered language of Taiwan that has one fricative /h/ and two stops /ʔ ʡ/ in the lower throat or "laryngeal' region. The glottal stop /ʔ/ is not an underlying lexically determined segment, but added by a rule to “cover” syllables that begin or end with a vowel. In contrast to /ʔ/, which is introduced by rule, is a lexically specified second laryngeal stop, which we identify as epiglottal/pharyngeal. It may occur initially or medially but when it appears word-finally it requires fortition or strengthening. Indeed, generally, stops and continuants receive greater force of articulation when they appear in word-final position after all morphological units—of which there can be many—are assembled. Thus, there is a glottal stop introduced to cover and separate syllables, as in ina [ʔinɑʔ] ‘mother’, mama [mɑmɑʔ] ‘father’ and saan [sɑʔɑn] ‘to say’ and a second stop with epiglottal/pharyngeal stricture that is lexically specified, as 'oner [ʡonər] ‘snake, serpent’ and 'epah [ʡə̥ɑħ] ‘alcohol’.1 The glottal stricture in a word such a ina resembles what we have called in earlier work a moderate glottal stop, which has vocal fold adduction, followed by ventricular incursive (VI) partial Scovering of the vocal folds by the ventricular folds cf.
Esling and Harris (2003). The second phonologically contrastive sound has a different point of constriction that begins with a moderate glottal stop as just described but then the The writing system for representing Amis in this paper is a slightly modified version of the system used for the Bible translation and Fey’s dictionary 1985 and is very close to the underlying phonological representation of words. The apostrophe signifies an epiglottal stop and nothing is written for epenthesized glottal stops, c is /ts/, d is /ɬ/, and e is /ə/. There is still debate about whether distinctive graphs /o/ and /u/ need to be written.
aryepiglottic folds and arytenoid cartilages constrict upward onto the laminar undersurface of the epiglottis to form an epiglottal stop. And finally, in lexical items with word-final /ʡ/ a different point of constriction can be observed. After the formation of an epiglottal stop (which itself presupposes a moderate glottal stop), the epiglottal surface is drawn back over the laryngeal vestibule as a consequence of strong lingual retraction inherent in sphinctering so that the tongue and epiglottis are pressed against the pharyngeal wall. Phonologically, only /ʔ/ and /ʡ/ contrast, and they are implemented phonetically as [ʔ] in amis [ʔɑmisː] ‘Amis’, kaen [kɑʔən] ‘to eat’ or roma [romɑʔ] ‘other’, and are distinct from /ʡ/, which is implemented as [ʡ] initially and medially, and as [ʕ͡ʡħ] finally, e.g. in 'icep [ʡtsəpʰ] ‘betelnut’, po'ot [poʡɔ̰tʰ] ‘small knife’, and loma' [ɾumɑ̰ʕ͡ʡħ] ‘house’.
It is also possible for three progressive points of stricture to occur in fricatives that correspond to the stop sounds. The data to support the brief descriptions just presented will be based on direct observation using the Kay Elemetrics Rhino-LaryngealStroboscope RLS 9100 System with Olympus ENF-P3 flexible fiberoptic bundle.
Images were captured on digital video for later viewing and editing. Before turning to that, however, we present a brief introduction to the Amis language.
2. Amis. Amis is an Austronesian language spoken on the east side of Taiwan from Hualien in the north to Taitung in the south. It extends along the coast from Peipu village in Hualien County to Malan village in Taitung County and is also found in one inland rift valley paralleling the coast. The population is said to number 140,000, but less than 50% of these can speak the language today and most of those are over 50 years of age. They form over 38% of the aboriginal peoples living in Taiwan and Orchid Island.
Huang Tung-chiou is a speaker of the Hsiukuluan variety of Amis about 50 km south of Hualien from a village called 舞鶴 Wǔhè.
Figure 2: Map of the Amis area
3. The glottal stop, epiglottal stop, and epiglotto-pharyngeal stop. The first of the stops, the moderate glottal stop—is describable in terms of the glottic and ventricular planes alone. The gesture train in this sound involves adduction of the vocal folds as its first element and the partial covering and damping of the vocal folds by the ventricular folds (ventricular incursion or VI). In Figure 3 below we present slides taken from the videos for Amis ina ‘mother’ [ʔinɑʔ]. Frame 1 begins in the breath position. In Frame 5 adduction of the vocal folds has taken place, which is then quickly followed by VI in Frame 7 and by the glottal closure with ventricular damping and reinforcement in Frame
9. In Frame 13 the ventricular folds uncover, and voicing of the vocal folds begins. But in all these stills one can see the open or later the hawk eye-shaped aperture onto the glottal plane afforded by a somewhat closed epiglottal plane that never achieves complete sphincteric closure. In the accompanying spectrogram the frames have been located in time by inspection of the gesture train in the Wavesurfer acoustical analysis software to show the relationship between the articulatory events and the sound production events.
Sometimes there can be small errors of timing, as the capture of events takes place at only 30 frames/second, and afterward they had to be matched visually against a corresponding slide from the video.