Re: [tied] Thoughts on the existence of *H1

From: Piotr Gasiorowski
Message: 9361
Date: 2001-09-11

----- Original Message -----
From: "Glen Gordon" <
To: <>
Sent: Saturday, September 08, 2001 11:21 PM
Subject: [tied] Thoughts on the existence of *H1

> I'm going to be contraversial
as usual and deny the existence of *H1 as a single distinct phoneme. Its major diappearance and yet continuation in Greek as *e isn't surprising if it were truely *[&] in words like *h1sus and *h1rudHros. Schwa usually
disappears very easily.

Don't worry, you aren't the only disbeliever :). Phonetically, *h1 is the most enigmatic laryngeal. You in fact imply that *h1 = [&] must be interpreted as the weak counterpart of *e word-initially, since we have alternations like *esti : *&sonti.

Only one strange question for Piotr... The 3ps of Late IE *g^nox- wouldn't be *g^no:u, by chance, would it? That's what my current MIE-to-LIE sound rules predict. Based on MIE *t:W xe "two" (*dwo:u), *kWetWxe "eight" (ok^to:u) and *k:elaxWe "sis's hub" (*g^lo:u-), it seems clear that any instance of MIE *-aXe became *-a:we and then *-a:u by approximately 5000 BCE before undergoing vowel shift to *-o:u in Late IE. Hence, my enquiry.

You mean the 3sg. present of what is usually reconstructed as *g^noh3- 'know'? Well, the present stem is normally reduplicated and either suffixed as in *g^i-g^noh3-sk^e- or
athematic (in *-ti). Same for the preterite (*g^noh3-t). *(g^e-)g^no:u looks like a possible sandhi variant of the 3sg. perfect after contraction (*-g^noh3-e > *-g^no-e > *-g^no:(u), using standard transcription), though there might be attestation problems.

> Anyways, in words like *dheH1-,
*H1 is clearly not *&. Rather, we might envision a true long vowel here and reconstruct *dhe:-.
I don't think this will always work. Think of combinations like *-eRh1C-. If here *h1 is some sort of weak vowel, why does the zero-grade counterpart behave as if *-R- had _first_ become syllabic and _then_ affected by the laryngeal? In other words, liquids, nasals and glides are vocalised more readily than *h1. Strange, if *h1 should have been a vowel in the first place.
For example, *-Crh1C- becomes Greek -re:-, Latin/Celtic -ra:-, Sanskrit -i:r-/-u:r-, Baltic/Slavic *-i:r- (producing a different intonation than an original *-CrC-), Germanic *-ur-. In all these it's clear that either a cluster-breaking vowel was epenthesised early _between_ *r and *h1 (and then was lengthened as the *h1 was lost: *-Crh1C- > -Crah1C-
> -Cra:C-), or the normal outcome of syllabic *r was affected by compensatory
lengthening (or even, as in Germanic, the laryngeal was simply dropped). Conspicuous by its absence is any evidence for *-Cr&C-. Note that the developments delineated above are branch-specific and that the Brugmannian notion of "long syllabic consonants" in PIE (still employed here and there as a notational convention) is untenable, so a protolanguage "contraction" like *-Cr&C- > *-Cr:C- (a little odd in itself from the phonetic point of view) is ruled out.
More arguments could be cited, but the one above looks particularly convincing. Unless you resort to some form of special pleading, it is difficult not to conclude that *h1 was some kind of consonant (inherently less "vocalic" than liquids and nasals). [h] or [?] (or both) would do, but [&] wouldn't.