Re: *a/*a: ablaut

fournet.arnaud wrote:
I think such a root as wreH1g is impossible.
=======
From: Piotr Gasiorowski
It's your problem. I don't know of any such restrictions.
==========

I give you an example of how
I handle some data.

Step 1
Greek laas "stone"
Possible Root *l_H2
next
PIE *leH1-p "stone"
I consider that H1 in this root is
*caused* by -p which is *unvoiced*.

Step 2
Because there is no voiced reflexes
of H2-p Eastern PIE,
we can assert that H2 was not glottalized.
It must be a weak voiced : &ayin or h (voiced velar)

Step 3
Arabic has laxfa "a flat and thin stone"
x is velar unvoiced.

Conclusion
The root is *l_h "stone"
h being a voiced velar spirant.
Derivative *l_h-p > l_x-p

Something like **H1g conflicts
in the worst possible way
with some of my beliefs.
PIE Phonotaxis did not allow that.

We can discuss this in a different string.

Arnaud
=============