Re: [tied] PIE Stop System

From: Jens Elmegaard Rasmussen
Message: 25833
Date: 2003-09-15

On Sun, 14 Sep 2003, Mate Kapovic wrote:
> So you do find it possible that at a period there was only one phonemic
> vowel in PIE?

Yes. The result would be as in Sanskrit where /a/ has precisely that

> Many linguists do not want to accept that possibility
> because
> of the typological reasons (although there were some analysis of the
> supposed one vowel phoneme-languages in Caucasus) in spite of the fact
> that
> there are a lot of facts which point in that direction - most stems
> having
> *e vocalism and those which do not have it seeming younger, three velar
> series (*K, *K', *Kw) which would point to the palatalisation and
> labialisation of velars being the way of preserving the distinctions of
> prior neuter-front-back vowel which have all collapsed  to *e, maybe some
> loans into IE like Semitic *sab?atum > *septm etc.

What you find in specific terms is *never* likely to be exactly as it is.
I have the highest respect for Illic-Svityc's analysis of the three velar
series as reflecting earlier vowel oppositions that are retained in
distant relatives of IE. I wis, however,h I were more competent to
evaluate it.

> But why these words? How do you then explain the Vollstufe/Nullstufe
> alternation in these words easily explained wtih *eh2/h2? Doesn't Lith.
> nósis (with acute, from *neh2s-) and Slavic nos7 (from *nh2s-) just
> demand
> to be reconstructed with a laryngeal (also Slavic sol7 < *sh2l- and Lat.
> sa:l < *seh2l-)?

If nothing plays a trick on us (as much of it of course may well do, but
IF), this is not possible:

The zero-grade of a presumed *neH2s- should be *nH2s- which would give
either /nis-/ or /a:s-/ in Sanskrit; the weak form of the stem /na:s-/ is
in fact /nas-/.

In parallel fashion, *seH2l- should have the weak alternant *sH2l-,
which is incompatible with Skt. salila- 'ocean'.

Therefore, if the material is trustworthy, we must posit *na:s-, *sa:l-
with long vowels, which are retained in strong paradigmatic forms and
shortened to *nas-, *sal- in weak forms.

Note that I am not insisting upon these analyses; I can also think of many
ways in which the development can have taken a funny turn. But even if
they have not, and the forms are reliable, we can accomodate the evidence
in a neat account.