Re: [tied] Quattuor (was: Latin barba ...)

On 2006-06-24 12:03, Miguel Carrasquer wrote:

> Perhaps. The Slavic forms (c^etyre, c^Ityre), like the
> Germanic form (Goth. fidwor) and Skt. catvá:ri, in any case
> do not continue *kWétwor-, with accent on the first element
> and short /o/ in the second, but are reflexes of collective
> *kWtwó:rh2, with originally zero-grade in the first element,
> and stress and long vowel in the second. Latin quattuor
> retains the zero grade (as "schwa secundum"). Elsewhere,
> *kWtwó:r may have been normalized to *kWetwó:r under the
> influence of *kWétwor-es (as it also acquired -e < *-es in
> Slavic under that same influence).

*kWtwó:r, with its o-vocalism, must be derived via a contrastive accent
shift from *kWétwo:r(h2), corresponding to animate *kWétwores. It's a
formation parallel to *d(H)g^Hó:m etc. We also have *-dk^omt- (in
decadic numerals), in my opinion the _substantivised_ counterpart of
*dék^m(t). I would expect the following forms in PIE:

a. *kWétwores (animate)
b. *kWétwo:r (neuter)
c. *kW(&)twó:r 'a group of four'
d. *kWtwr.- ~ *kWtru- (in compounds)

Contamination between c. and the cardinal numerals with initial accent
could of course have produced mixed forms.

Piotr