On Mon, 28 Jul 2003 03:26:04 +0200, Miguel Carrasquer <
mcv@...> wrote:
>Classical Armenian has a category of nouns and adjectives (the so-called
>wo- and ea-stems), characterized by a nom. sg. in -i and a rather bizarre
>set of endings.
>
>Plain o- and a:-stems are declined as follows:
>
>o-stems
>
>NA -0 < *-os, *-om
>L -0, -i < *-o: (?), *-oi
>GDAb -oy < *-ósyo
>I -ow < *-ó-bhi
>
>N -k` < *-o:S
>AL -s < *-o(:)ns
>GDAb -oc` < *-ó-sk^i
>I -owk` < *-óbhiS
>
>a:-stems
>
>NA -0 < *-a:, *-a:m
>GDL -i < *-a:i
>Ab -ê < *-a: + *eti
>I -aw < *-á:-bhi
>
>N -k` < *-a:S
>AL -s < *-a:ns
>GDAb -ac` < *-á:-sk^i
>I -awk` < *-á:bhiS
I had forgotten that Birgit Olsen does not derive o-stem locatives in -i
(besides -0) and the a:-stem GDL in -i directly from PIE forms ending in a
diphthong. The assumption is that PIE *-oi, *-o:i, *-a:i (as well as *-ei,
*-e:i and *-eu, *-e:u / *-ou, *-o:u of the i- and u-stems) gave Armenian
-0. The a:-stem GDL in -i must then come from -iya:s of the de:vi: class
or -iyos from the vrki:s class, while i-stem GDL -i and u-stem GDL -u are
from the generalized secondary type Gen. -iyos, -uwos, Dat. -iyei, -uwei,
all with -iy- and -uw- by Sievers law, and with analogical locatives.
I'm sure Armenologists haven't just overlooked the possibility of a direct
development -oi, -o:i, -a:i, -ei, -e:i > -i, or -eu, -e:u, -ou, -o:u > -u,
so my question is: why has his possibility been discarded?
In its favour is the fact that the endings of the vocalic classes can be
explained straightforwardly from the most common PIE endings and declension
types:
o-stems:
L -i < *-oi (D. *-i replaced by -oy, L. -0 from the Ins. in *-o:)
a:-stems:
DL -i < *-a:i (G. -a:s > -0, so analogical -i)
i-stems:
GDL -i < *-eis/*-eiei/*-e:i
[expected pl. *-ik` or *-ek` analogically replaced by -k` afer Acc.pl.
*-ins > -s]
u-stems:
GDL -u < *-eus/*euei/*-e:u or *-ous/*-ouei/*-o:u
[expected pl. *-uk` analogically replaced by -k` afer Acc.pl. *-uns > -s]
Another argument is perhaps 1sg. aorist -i if from a perfect ending *-h2ai,
on the Latin (-i:) and Slavic (-e^) model, with 1sg. middle aorist -ay <
*-áh2ai.
Against it I can think of:
- The accentuation (e.g. Loc. *h1ék^woi > *és^i, but Armenian es^í).
However, words with unstressed residual -i and -u may heve been adapted to
the normal end-stressed pattern.
- The Dsg. of the C-stems, which ends in -0, not *-i (but perhaps dative
and locative had merged in Pre-Armenian?).
Any other arguments that I'm unaware of?
=======================
Miguel Carrasquer Vidal
mcv@...