Re: [tied] The centum-word.

From: Miguel Carrasquer Vidal
Message: 6891
Date: 2001-03-31

On Thu, 29 Mar 2001 21:52:00 , "Glen Gordon"
<glengordon01@...> wrote:

>Now, what about the lengthening? Lengthening can be caused by many, many
>factors from analogy to loss of laryngeal and so it drives me up the wall
>that this phenomenon is used as some kind of proof that *-d- must have
>existed in an entirely hypothetical decad stem **-dk^ont-. The *d just isn't
>there and we will have to eventually swallow this reality when centuries go
>by without ever finding this devilish leprachaun attested.

Well, insisting on attestation of a mere preconsonantal *d, following
so close on reconstructions that have practically no segments in
common with reality (*t:eu "1" or *kWelei "3", for instance) seems a
bit odd. PIE *d was, as is well known, a somewhat unstable segment,
with a tendency to disappear (*dwi- ~ *wi- "two", *dak^ru- ~ *ak^ru
"tear"), most notably in Tocharian, where the loss of *d is regular in
certain environments. The lengthening, as if by *h1, seen before
*-(d)k^omt(h2) is attestation enough, as far as I'm concerned. If you
need more proof, there's the lenition caused by *d (> *h1) in Latin
-ginta:. Case closed.

>As was mentioned by Piotr, formations like *trix-k^ontx, with plural *-x,
>combined with analogy, is a possible alternative to explaining the
>lengthening from 30 to 90.

I don't think Piotr said plural *-x (-*h2), because no such thing
exists. Piotr's point was rather that the only viable starting point
for an analogical process would have been a form like Greek tria:konta
(with *h2, rather than *h1), not any of the attested forms of
"twenty". To which one can add that *trih2-k^omth2 is not necessarily
the PIE Urform (*tri-h1k^omth2 works fine for anything but Greek and
Tocharian *<ta"rya:ka> (< *t(i)rih2-k^mt-)), and that it failed to
trigger any analogical processes even in Greek itself, where the other
decads have *-h1- (< *-d-).


=======================
Miguel Carrasquer Vidal
mcv@...