Articulatory phonetics and vowels

4/1/2018

Some time ago I grumbled that the Wikipedia article on Articulatory Phonetics had plenty of information about consonants but only a couple of sketchy notes about vowels. I've made a start on improving the situation by editing in a summary of the main articulatory variables affecting vowels, but it's pretty sketchy, like its predecessor. I hope that at least the framework is more appropriate. I have made no attempt to provide a detailed treatment of each of the variables - they are already amply covered in the huge network of overlapping phonetics articles in WP.

I've had to read the material in the corresponding WP article on Vowel as I went along. Something that has puzzled me is the statement there, in the section "Front, raised and retracted", that The conception of the tongue moving in two directions, high–low and front–back, is not supported by articulatory evidence and does not clarify how articulation affects vowel quality. Vowels may instead be characterized by the three directions of movement of the tongue from its neutral position: front, raised, and retracted. There is a reference to a paper by John Esling at the end of this section, so I suppose this comes from him (I haven't been able to get hold of the article yet). I'm afraid I can't see why the traditional open/close and front/back dimensions are inapplicable, as long as one doesn't treat them as absolute determinants of vowel quality.

8 Comments

Sidney Wood link

4/4/2018 06:11:05 pm

"The conception of the tongue moving in two directions, high–low and front–back, is not supported by articulatory evidence". We owe the notion of backness (a continuum of vowel constriction locations from front to back) to A. M. Bell (Visible Speech 1867:15-16), an inspired guess but no more, he had no way of knowing where the tongue was. And he was wrong. He tied it in with the single-resonance theory (already flawed by Helmholtz 1863 who reported two resonances for front vowels). Bell never mentioned it again and we owe its success to the efforts of Henry Sweet and the Neo-Grammarians and after that there were too many reputations and careers bound up in it. Bell's idea was flawed again in 1911 by Léonce Roudet (La classification des voyelles de M. Sweet; Revue de Phonétique 1:347-356) who pointed out that the vowel resonance depended on the volume of the front cavity, set both by backness and height (he was in fact arguing that height and backness would be mutually compensating and not the independent parameters Bell intended). The Bell model has never been confirmed by X-ray studies. Jones had made X-rayed profiles of his cardinal vowels in 1917 and immediately ran into trouble that took 15 years to work around. He finally withheld all the images except the four corner vowels. The full story is here: https://swphonetics.com/2018/03/04/150th-anniversary-of-the-bell-vowel-model-5/ and on other pages. Good luck.

Peter

4/8/2018 06:29:06 pm

Thanks for the clarification and useful references. I appreciate that it would be impossible to demonstrate a clear relationship between tongue frontness/backness and perceived vowel quality or measured formant structure. But is it incorrect to state (as a matter of physiology, not of phonetics) that the tongue may be moved forward and backwards in the mouth in such a way that the highest point of the tongue may be observed to be nearer to the front or to the back of the oral cavity?

Sidney Wood link

4/16/2018 02:41:02 pm

Sorry about the delay, response coming shortly

Sidney Wood link

4/16/2018 05:07:40 pm

I'd like to expand my first post, then take your comments.

The Bell model is dead but won't lie down. A reason I suggested was all the reputations and careers invested in it. For example, I went through about 20 recent phonetics manuals intended for students, and all but a couple never offered even a hint there might be any problem with the Bell model (or Jones' version of it). A second reason is that phonetics and phonology have mainly needed a system for classification, not physiology, and the Bell model offered 3x3x3x3x2x2 (324) classes, more than enough for any language. It didn't matter the Bell tongue positions were unconfirmed conjecture. Students have been given wrong tongue position advice for 150 years, they passed their exams anyway by listening and learning by rote.
The Bell model is wrong on at least two counts: (i) The single resonance theory, (ii) the location of a constriction somewhere along the palate ("backness") to tune resonance. The theory had already been flawed by Helmholtz, so the explanation (ii) should've fallen with it. Bell never mentioned it again and reverted to the ancient throat-tongue-lip model that had described tongue articulation in terms of palatal, velar and pharyngeal constrictions since Panini to the end of the 19th century. Bell's son Graham introduced the double-resonance theory in 1879, giving renewed life to A M Bell's constriction as the divide between buccal and pharyngeal cavities. Both theories assumed a single cavity uniquely for each resonance. This was clearly wrong, once more resonances had been identified for which there were no more unique cavities available (see Fant 1961 ICPhS). For good measure Fant showed in 1980 (Phonetica) that F2 of [i] was more sensitive in the pharynx, contrary to the dual cavity theory. But all vowels were distinguished, which was good enough for phonology.
Bell's account of tongue behaviour was imaginative guesswork. The discovery of X-rays offered the first opportunity to observe tongue articulation properly. Vowel profiles have been studied since around 1900, revealing a bewildering account of tongue heights: [ɪ] is lower than [e:], 66% of [ɔ] and 33% of [o] are the same height or lower than [a ɑ]. The same X-Ray studies consistently support the three constriction locations of the ancient throat-tongue-lip model and reveal a fourth in the upper pharynx. so there are palatal [i e] etc at the hard palate, [u ɯ] etc at the velum, [o ɔ ɤ ʌ] etc in the upper pharynx and [æ a ɑ] etc in the lower pharynx. No wonder Jones had trouble with the cardinal vowel profiles he X-rayed in 1917, finally accepting the corner vowels but rejecting the others in favour of the equidistant heights he wanted to believe in. Neither Bell or Jones had any idea on how tongue height might affect sound. It was simply assumed it would cause timbre progressions like [i] to [a].
Perturbation theory, for phoneticians, has seen the single and double resonance theories come and go. Current theory, based on the work of Chiba & Kajiyama (1948) and Fant (1960), sees a standing wave from the lips to the glottis for each resonance (nasality complicates matters as it introduces a branch pipe). For oral vowels, the resonance frequencies are modified by widening or narrowing the vocal tract at the nodes and antinodes of the respective standing waves. Local widening or narrowing is mainly a task for the tongue body, but the tongue tip and blade will also work independently in the alveolar region and the tongue root in the lower pharynx. A modification might affect more than one resonance frequency, while any resonance frequency can be altered at more than one location. Only four constriction locations are relevant for monophthongs (see above). For diphthongs, the tongue starts and ends at one or other of the four locations, then sweeps rapidly between them. K N Stevens work on quantal theory is relevant here too.

Taking your questions:
Highest point of the tongue. The answer has to be yes. The difficulty with respect to vowel articulation is that the location of the highest point of the tongue is not relevant for tuning the resonances of the vocal tract (dealt with once and for all by Fant 1960, 1961). Similarly, the position of the wrist is not relevant for a note being played on a flute. Worse still the highest point of the tongue is not defined, but that’s not the problem, defining it wouldn’t help. Nod your head forward and tip it back, intuition should be reliable enough to tell you the location of the highest point moves about along the surface of the tongue. Where is it when you lie down on your back and speak? Somewhere near your tongue tip? Jones probably meant something like “the point of the tongue surface most remote from the feet when the legs and body are straight”. For earlier X-ray studies of steady vowels, the head was oriented to the Frankfort plane and held stationary in a ceph

Sidney Wood link

4/16/2018 05:17:10 pm

My post was too long and the end was excluded. Here it is:

....., but we can speak while moving the head freely. The head positions of the subjects of my X-Ray motion films were not restrained. The tongue heights of each subject were all over the place, but they all kept precisely to the four locations, whatever the language. These are the same locations as for dorsal consonants. So the tongue body is moved directly between these locations as required.
Frontness, backness, timbre and formant structure. Vowel positions in the diagram are codified by e.g. Bell’s original description or the IPA. So every student “knows” that [i] is high front. If they do their homework properly they always get timbre, frontness and backness right. Regarding formant structure, Essner (1947) and Joos (1948) both reported that F1 and F2 frequencies are similar to corresponding judgments of height and backness. But that’s not the same as a causal relationship. No-one knows why. Essner explained it in terms of the dual-resonance theory (already nearing the end of its life). Joos explained it as mental acoustic analysis expressed unconsciously as articulation. The motor theory might come closer.

Peter

4/17/2018 10:01:20 am

It's really valuable to have all this explained so clearly. I wish I could think of a way of getting at least the general points into Wikipedia to correct the pervasive view of fonrt/back vowels.

Sidney Wood link

4/17/2018 02:50:47 pm

You're welcome. I've never contributed to Wikipedia because of their principle "Anyone with Internet access can write and make changes to Wikipedia articles", together with anonymity. I don't want to spend a lot of time writing something and then have to defend it against anonymous comments or even see it edited.

You were puzzled by "Vowels may instead be characterized by the three directions of movement of the tongue from its neutral position: front, raised, and retracted" in the Wikipedia 'Vowel' page. So am I. At first sight it looks like the three vowel constrictions of the ancient throat-tongue-lip model - palatal, velar and pharyngeal. But that's guessing. I've requested a copy of Esling's referenced article but had no response yet.

Sidney Wood link

4/22/2018 12:24:44 pm

I realise that wasn't very helpful. A new article is probably the way to go rather than attempt to modify an existing article. I'll see what I can manage during the summer.

Articulatory phonetics and vowels

Leave a Reply.

A blog that discusses problems in Wikipedia's coverage of Phonetics

Peter Roach

Archives

Categories

Articulatory phonetics and vowels

Leave a Reply.

A blog that discusses problems in Wikipedia's coverage of PhoneticsPeter Roach

Archives

Categories

A blog that discusses problems in Wikipedia's coverage of Phonetics

Peter Roach