> The *th and *dh are very useful since we can have *ath>aw > u,o and
> > ay> e,i.
> That would only require a single vowel as a start.

So useful, I fear, that little confidence can be placed in any
reconstruction that invokes them. You speak as though you were
playing scrabble!

Well, I am reading the book by Bomhard and Kerns and in the beginning there
is a
long discussion of the sound changes of IE. Right now I am on the section by
Szmerenyi, and it looks so far like a perfect case of "overfitting".

Mine is the reverse. It is minimalism.

Some of this internal reconstruction would certainly benefit from checks
from outside IE. The two confident internal reconstructors on the list,
Jens and Miguel, do take an interest in Nostratic links.

The internal reconstructions do hang together. Your model seem to have
many waves of borrowing.

> PT inital-n disappeared. e.g,. KBal nalmaz=almas (diamond) and nakut
> yakut).
> Both yakut and almas are said to be Arabic and almas is said to be
> related to
> diamond, adamant etc. Here they show up with initial-n, and it is
not a
> reconstruction. Such anomolies can either be buried away or
attempts can be
> made to handle them.

I'm not familiar with the abbreviations for Turkic languages. What
forms are you citing for which form?

Karachay-Balkar (Siunchev and Tenishev dictionary).

Thanks. I mangled the question. In which languages do we have nalmaz,
almas, nakut and yakut, and what do nakut and yakut mean?

> > Looking at Torsten's list of <plosive>Vr roots at
> > , I can't help
> > thinking that there's some sort of sound symbolism going on
here. In
> > English we've got 'whir' and 'whirl', and in PIE there's a root
> > with several extensions, all related to turning.

> I do not see onamatopeia.

Do you not see onomatopoeia in 'whir' and 'whirl'?

No. I dont' make sounds when I turn. Do you mean spin something fast and
force it to make swishing sounds?