--- In
cybalist@yahoogroups.com, Miguel Carrasquer Vidal
<miguelc@...> wrote:
>
> On Fri, 28 Mar 2008 08:51:01 -0000, "jouppe"
> <jouppe@...> wrote:
>
> >I can also give you a good regular one:
> >
> >Finno-Ugric /s/ Samoyed /t/
> >
> >Material:
> >*sålå (UEW sala) 'to steal; (fi.) stealth hidden, secret' || Samoj
t-
> >*sårkå (UEW sarka) 'crotch, branch (fi. > gusset > strip)' ||
Samoj t-
> >*suksï (UEW sukse) 'ski' || Samoj- t-
> >*kåxsï (UEW kuse/kose) 'spruce, fir' || Samoj Selkup qu:t Taigi
kat
> >etc.
> >
> >Diachroically */s/ is reconstructed for Proto-Uralic.
>
> Yes, but I wonder whether that can be correct, given that
> the Ugric reflex of PU */s/ is PUgr */þ/. I can buy one
> unusual development (s > t in Samoyed), but two (s > þ in
> Ugric) seems a bit too much.
>
> =======================
> Miguel Carrasquer Vidal
> miguelc@...
>
I sympathise with the gist of your doubt although I am hesitant as to
the concept of "Proto-Ugric", which is loosing ground as a "valid
node" ("Proto-Samoyed" is not under question though). If we would
accept the proposed split of earlier reconstructed Ugric into Proto-
Vogul and Proto-Ostyak +separate Proto-Hungarian, the more valid your
doubt would become, as it would presuppose not two but three
independent unusual sound changes.
As far as I understand (here I do not remember Janhunens or
Sammalahtis underlying argumentation) a main problem of a
fricative "thorn" is the economy of distinguishing features in the
system.
There is already a fricative d in the system and the opposition
between the thorn and the fricative d would be the only point in the
system where voicing is the only and minimal distinguishing feature.
This problem may perhaps be solved by assuming a palatalized /t'/ for
this set of correspondences?? (my speculation)
Yet in this case there would only be one sibilant left, the
palatalized one: but nowhere else in the system does a palatalized
consonant exist without a non-palatalized (less marked?) counterpart.
May violate universals for phoneme systems?
I see no alternative solution, but may be there is one. I'm not sure
if one could then look towards the affricate area next, if one is not
ready to reconstruct */s/. There is already one affricate */c/ = [ts]
reconstructed so it is a bit crowded in this area already.
Jouppe