From: Piotr Gasiorowski
Message: 2487
Date: 2000-05-22
----- Original Message -----From: Mark OdegardSent: Monday, May 22, 2000 2:15 AMSubject: [TIED] Beekes' PIE Consonants & Glottalized Consonants.The transcription used by Beekes actually implies ejectives [t’] rather than preglottalised stops (such as [ʔt] – a sound often heard in British English cat).
I’m going to devote a separate posting to phonation types and airstream mechanism; it will be posted to phoNet, with a carbon copy for interested Cybalist members. Here I’ll only say that ejectives cannot be voiced. The stop burst accompanying the release of an ejective is indeed very strong, but when you say that nasalisation ‘is difficult to avoid’ it makes me wonder whether you managed to produce a true ejective or some other type of sound involving a glottal closure. You’ll find a recipe for ejectives in the forthcoming phoNet posting.
The problem with reconstructing the PIE stop system is that the traditional reconstruction, which seems to optimise the naturalness of the assumed sound changes in the daughter branches, is itself objectionable from the point of view of linguistic typology.
The familiar triad *t, *d, *dH is virtually without typological precedent if *dH is interpreted as aspirated. In a decent stop system, a voiced aspirated series should not occur if voiceless aspirated stops are missing. The traditional reconstruction also fails to account for the non-occurrence of PIE roots with two plain voiced stops (ged-) and the absence or rarity of PIE *b (pace Szemerényi's attempts to prove its ‘robust attestation’).
The assumption that *d = [t’] and *dH = [th] gives us a plausible system with ample typological precedent. Multiple-ejective morphemes (k’et’-) are prohibited in many languages; the bilabial gap is also accounted for, as bilabial ejectives are universally rarer than coronal or velar ones, and ejecive systems often lack them altogether. However, the lack of comparative justification for these phonetic values is a serious weakness of all versions of ‘glottalic theory’, whatever their formal elegance. Any system with voiced *d and *dH is certainly more consistent with the known sound correspondences.
There have been several different attempts to get round the difficulties caused by *dH. Szemerényi postulated a four-way system, as in Indic (t : th : d : dɦ
). There are two difficulties with this solution: (1) the voiceless aspirated series is at best very weakly attested; (2) such systems, with th and dɦ grouped together as ‘aspirated stops’ seem to be something of an areal peculiarity, being found almost exclusively in various (often unrelated) languages of the Indian subcontinent, far from any plausible PIE homeland.Gamkrelidze & Ivanov reconstruct an elaborate system with variably aspirated *t(h) and *d(ɦ)
plus an ejective series, *t’. Despite all their prestidigitation involving positional allophones, they run into trouble trying to account convincingly for the reflexes of variable aspiration in the daughter branches and for Grassmannian phenomena. The generally voiced reflexes of alleged [t’] are as embarrassing in this model as they are in other ‘glottalic’ reconstructions.Hopper makes *dH a ‘murmured’ (breathy voiced) stop, which seems to make good sense. As Ladefoged & Maddieson note, breathy voiced stops may pattern as aspirated stops if voiceless aspirates are found in the same system (cf. Hindi) – aspiration is then interpreted as a period after the release of a stricture before the start of regular voicing. This doesn’t mean, however, that all breathy voiced stops function as aspirates, so if PIE *dH is taken to mean breathy voiced (not aspirated, though reinterpreted as such subsequently in some branches due to the local emergence of *tH), the *t : *d : *dH series can pass typological muster.
Of course Hopper reconstructs *t’ rather than *d. To have the best of both worlds, one might prefer a series of voiced stops with distributional limitations similar to that of ejectives. But what kind of voiced stops? Implosives, for example, wouldn’t do better than ejectives, since they can’t explain the labial gap: [ɓ] is not disfavoured in implosive inventories.
On the other hand, even in the traditionally reconstructed system the rarity of *b could be due to causes other than its airstream mechanism (e.g. an early merger of *b and *w, not without precedent). The coocurrence restrictions (absence of *ged- etc.) are more problematic.
One could consider *d = [d̰] (creaky voiced, a.k.a. ‘vocal fry’ in American terminology, i.e. strongly laryngealised stops). Creak phonation consists in the low-frequency, irregular vibration of an unusually heavy mass of strongly adducted, thickened and slackened vocal folds, which might easily result in its being limited to a single occurrence in a morpheme. Breathy voice, on the other hand, requires little voicing effort and would not be more restricted in its distribution than voicelessness or modal (regular) voice, hence the free occurrence of *dH.
Here’s my alternative interpretation (don’t take it too seriously):
*t = [t] (voiceless)
*d = [d̰] (voiced, laryngealised, low airflow rate, low-frequency vibration)
*dH = [dɦ] (breathy voice, audible friction noise, high airflow rate)
This posting is getting too long, so I’m not going to expatiate on vowels and laryngeals right now, but I’ll return to them later.
Piotr
Mark writes:
Beekes is of the opinion that the PIE consonants usually rendered [b], [d], [ǵ], [g], and [gʷ] are actually glottalized (the actual transcription done here is reproduced from Beekes' book).
In Section 11.4.8 (p. 132), he states he agrees in part with Gamkredlidze and Ivanov as well as Hopper. On reading this section, I am a little confused, mostly by the transcription, but also by the text.
My assumption is that Beekes is saying all of the above-mentioned consonants were actually preglottalized. In trying out this combination of glottal+[b] I get the idea that the two 'articulatory gestures' may actually have been essentially simultaneous. The result is odd to say the least. You get a rather strong gust of air once you open your lips, and for myself, nasalization is difficult to avoid. Yes, if he's correct, his comment about fortis and lenis is right on.
I don't quite know how to formulate my question. Szymeryni has different ideas than Beekes, especially about the value of the laryngeals. Then there is Beekes' ex cathedra statement about the non-existence of PIE [a]. I don't know what to think. Am I supposed to take all this seriously? Sz. comments someplace that this would make Germanic and Armenian *extravagantly* archaic and apparently turns the Grimm-Verner laws inside out with respect to the rest of IE.
This is difficult stuff.