On 2007-09-24 22:02, fournet.arnaud wrote:
> Interesting question !
>
> My own guess :
> it is known that Attic had [ü] for [u]
> so that
> we can posit that :
> ghwer > gh-ü-er > palatalized gyh-ü-er > tyh-ü-er
> ü is lost > tyh-er- > palatalization is lost > ther
But the same happens to inherited labiovelars: *kWe > te, etc., where
the labialisation doesn't alternate with *u. It seems that the labial
component of the consonant was palatalised before front vowels and *kW
became *kY (with an off-glide phonetically similar to what French has in
<cuir>), further palatalisation producing *ts' > t. It's interesting to
note that in Albanian, too, *kW and *g(H)W have undergone palatalisation
before front vowels, merging with the reflexes of PIE *k^w and *g^(H)w
and ending up Modern Albanian s, z, whereas "plain" *k and *g(H) have
remained stops. The impression one gets is that some kind of common
areal tendency affected both archaic Greek and Proto-Albanian.
Piotr