From: Patrick Ryan
Message: 61157
Date: 2008-11-01
On 2008-10-31 16:08, Arnaud Fournet wrote:
> When unstressed, all vowels disappear in PIE stage,
> so kuH2on- when stressed on kuH2on-""i
> becomes kH2n-""i > kan-""i
> U disappears like all other vowels do.
> No big deal.
Well, it didn't disappear in Greek, Sanskrit, Avestan, Lithuanian,
Latvian or Irish. They all show clear reflexes of *k^un-, not *k^&2n-.
The PIE zero-grade of *-uh2V- is *-uh2-, not *-h2-.
> What about Latin can-is ?
> You still have not explained how Latin fits here.
It doesn't. I'm under no obligation to include <canis>. It doesn't match
the other forms, so it's probably something else, e.g. a derivative of
the root *kan- 'young, little' (used of humans and animals), cf. Skt.
káni:yas- 'younger'/kanis. t.Há- 'the youngest', kani:- 'girl', Gk.
kainós and Lat. recens 'new, recent', MIr. cano 'wolf cub', etc.
> You little cheap cheater...
Mind your step, Arnaud.
> You can't choose which data you want to explain
> and which data you want to discard
> because of your preconceived dogmas.
I _can_ choose what I want to include, based on common sense. If I don't
include OCS pIsU, Eng. dog or Sp. perro, is that dogmatic of me? You
include <canis>, reconstruct the root as *kuh2on-, and what do you gain?
One problem remains in Latin (no matter how loudly you shout, this *u
should not be allowed to disappear; PIE *u does not disappear in the
zero grade) and plenty of problems arise in several other branches: in
Greek and Tocharian, *uh2 would have given *wa:, in Indo-Iranian and
Celtic, long *u:, in Baltic, an acuted long *u:, etc. Your
reconstruction solves one problem (Lat. /a/ in <canis>) but screws up
everything else. Please don't ask me to take it seriously.
> it ruins the premice
> this word is PIE stage
> I suppose you understand that.
> It makes a big difference.
It can be reconstructed confidently for non-Anatolian IE, so it has at
least the same status as a verb like *bHer-e/o-. Whether there was a
cognate of *h1ek^wos in Anatolian or not depends on what one makes of
the Luwian data. Melchert (1987) has a rather solid case for reading the
HLuw. 'horse' word as <azu(wa)->, distinct from what we find in the
Hittite horse-training manual (i.e., from the actual Mitanni-Indic
loan). And if <azu(wa)-> is native, why not Lyc. esbe?
> The problem is the forms that don't match the others.
Like what?
Piotr