Re: Laryngeals Indo-Uralic

> > > I don't even have a theory of how those two are
> > > related. I stick to loans. In this case from a substrate to
> > > both, presumably one that knew ice.
> >
> > > Torsten
> >
> > Hm, but if the word is a substrate loan into PIE and PU both, how
> > do you rule out the possibility of this being of common
> > inheritance after all?
>
> Several small things:
>
> 1) My proposal covers both the 'ice' and the 'icle', but they are not relatable within PIE with known rules, which they would have been if they were inherited.

Nothing prevents one of these being a loan (from some IE, Uralic or related source) and one inherited. The two elements seem to be of different age anyway: "ice" (zero initial?) would fall under the IE *H <> Uralic *j correspondence, while "ickle" (< †gicel) has likewise *j- and might thus be newer (within IE).

Speaking of derivation, how much of your proposed derivation *iNgs > *eis would be during the IE evolution and how much within the donor language? I'm not aware of any regular law of compensatory loss of *Ng. If anything, that looks like we should get *ks.

> 2) The limited and northern geographical distribution of the PIE words cognates (except for the Iranian word, but who knows what nomads pick up).

It's only natural that words meaning "ice" might be lost in more southern descendants.

Hm, but could it be Iranian palatalized this word to get *-s-, and they loaned it to Germanic? Except the voicing doesn't quite fit.

> 3) The derivation with a genitive partitive -s as in *gl-a-s (and,
> I suspect, *gr-a-s) points to Aestian or whatever preceded it.

Where does Aestian come into this?

> > Etherman brought up other examples of a correspondence
> > of PU *ä to IE *ei not long ago on the Nostratica list.
>
> That is not a counter-argument, loans of one and the same path also
> show regular substitutions.
>
> Torsten

OTOH regular patterns should be explained as regular loaning only if inheritance can be ruled out. (The correspondence is also non-trivial so the point of divergence needs to be pre-PU or pre-PIE anyway.)

John Vertical