Re: [tied] PIE laringeals

From: Piotr Gasiorowski
Message: 44306
Date: 2006-04-20

On 2006-04-19 22:23, David Mandic wrote:

> What theories are there about the phonetic value of the PIE laryngeals?
> I've read somewhere they might have been pronounced as x', x and xw
> respectively (to fit into the pattern set by k', k and kw etc.).
> However, it doesn't seem plausible to me - if xw turned e into o, why
> didn't kw or gw?

This objection has often been raised and is certainly valid.

> According to me, h1 could have been a glottal stop,
> since it didn't affect a flanking e (excluding the lengthening), and
> disappeared early, even in Anatolian.

It's probably the majority view at present that *h1 was some kind of
glottal sound. When not vocalised, it seems to have had an aspirating
effect on a following (sic) stop in PIE, though the details of the
process are not entirely clear yet, which would favour a glottal
approximant/fricative [h] over a glottal stop [?]. However, it's hard to
rule out the possibility that the reconstruction *h1 covers two
different but hard-to-distinguish PIE phonemes, */h/ and */?/.

> H2 on the other hand might have
> been a sort of pharyngeal.

Yes, a back fricative, at any rate. Its main allophone was voiceless,
but the voicelessness doesn't seem to have been distinctive, as *h2 did
not participate in voice assimilation processes. Like *h1, it could
exert pre-aspirating influence on PIE stops. Its Indo-Iranian reflex
aspirated a _preceding_ stop (which indicates an /h/-like pronunciation
at that stage).

> As for h3, I've got no idea. Are there any
> sounds which display similar effects on vowels in other languages of the
> world?

You mean retraction and/or rounding? There is no shortage of phonetic
environments that might produce such coarticulatory effects. The
influence of *h3 on _consonants_ is perhaps more enlightening, since
there is some decent evidence of voicing assimilation produced by *h3,
as in reduplicated *pí-[b]h3-e/o- and in some "Hoffmann compounds" with
final *-h3on-, such as *h2ap-h3on- > Celt. *abon-. It would be
interesting to see if *s > *z before *h3. Unfortunately, PIE basically
excluded -es-stems as first members of compounds, so we would need a
root noun with final *-s before the Hoffmann element as a test case, and
I can't think of a good example at the moment.

> And what about the reflexes of the laryngeals in Slavic and
> Baltic - they also yielded prosodic (tonal) features.

Yes, but these features don't distinguish the laryngeals, so they are of
little use in reconstructing the phonetic details you're interested in.

Piotr