On Sat, 27 Oct 2001 03:18:32 +0200, Miguel Carrasquer Vidal
<
mcv@...> wrote:
>[*] *-nt- needs no further introduction. It's interesting to consider
>what happened to possible neuter stems ending *-nkW-, *-ng^-, or
>*-nk^-. A possible case of *-r(g^)/*-ng^ is the comparison of Arm.
>kol/r "branch" with Pol. gal/a,z' "id." (< **g(W)olr ~ **g(W)leng^-
>?).
I think I've found an -r(k) ~ -nk- stem [*]: it's the "bone" word.
The reconstruction is complicated not only by the initial consonant
(which gives h- in Hittite, zero everywhere else, but k- in Slavic and
in the Latin variant <costa> "rib(s)", I'll notate *Q), but also by
the fact that this word is not a typical r(C)/n(C)-stem, but an
i/n-stem. We have OIr. asna ~ esna and MWelsh eis(en), pointing to
*Qostn-. Latin oss- and costa, Albanian asht, TochB a:sta point to
*Qost- (n.pl. *Qost(e)h2). Greek osteon, Hittite hasta:i and Slavic
kostI point to *Qost(e)i-. Sanskrit has ásthi, G. asthnás, which
suggests *QostHi ~ *QostHn-. The key, I believe, is in Greek ostakós
"lobster" (*Qostn.k-os) and Armenian oskr "bone" (*Qostkr). Assuming
the i/n alternation derives from a palatalized */n^/ (as is my theory
about the i/n-stems), and that a *-k in Auslaut developed into *-h2,
we can derive all the forms from N. *Qostn^k, G. *Qostn^kos[**]. The
N. gives *Qostn^k > *Qostn^h2 and then by metathesis *Qosth2n^ >
*Qosth2i, except in Armenian where we have *Qostn^k > *Qostnk >
*Qostrk > *Qostkr > oskr [at least that's one possibility]. The
oblique gives *Qostnkós > *Qostnós with the usual analogical loss of
the consonant after the /n/.
[*] Rather, -r(h2) ~ -nk- stem.
[**] Maybe from something like **Qa:stink-, where **-ink/*-n^k- does
appear to be a suffix (in fact the well-known suffix *-(e)nk-). It is
quite possible, therefore, that some of the IE forms can be derived
from simple unsuffixed *Qost-.