On Thu, 08 Apr 2004 22:01:13 +0400, Vadim Ponaryadov
<
ponaryad@...> wrote:
>Firstly, the idea that the original feminine ending was *-ih2, where *-i- was deleted in certain circumstancies, etc., is really not mine, but Miguel's. I only wrote that, if reconstructed so, similarities in Mongolic can be found. (Really, I thought about this for a long time before, and it seemed to me that the parallel should be false, becouse the correspondention IE -h2 ~ Mong. -i hardly can exist.) But, of course, before speaking about any external connections, we must be sure that the IE reconstruction itself is correct. So I asked Miguel about the reasons to reconstruct *-ih2, and they really exist, although this is not the only possible reconstruction, and the explanation you are writing about is perhaps better. But at the same time I still don't see why Miguel's explanation must be necessarily wrong.
Of course: it's correct :-)
My reconstruction explains the Sanskrit a:-stem paradigm,
which made no sense according to the traditional view.
The usually reconstructed forms are:
nom. *-eh2 > *-a:
acc. *-eh2-m > *-a:m
voc. *-e(h2) > *-a
gen. *-eh2-os > *-a:s
dat. *-eh2-ei > *-a:i
loc. *-eh2-i > *-a:i
ins. *-eh2-eh1 > *-a:
But Sanskrit, unexpectedly, shows:
nom. -a:
acc. -a:m
voc -e:
gen. -a:ya:s
dat. -a:ya:i
(loc. -a:ya:m)
ins. -aya:,
which is completely regular in my scheme, from:
nom. -é-(i)h2 > -eh2 > -a:
acc. -é-(i)h2-m > -eh2m > -a:m
voc '-e-ih2 > -oi(h2) > -ai
gen. -e-íh2-as > -oyéh2os > -a:ya:s
dat. -e-íh2-ai > -oyéh2ei > -a:ya:i
ins. -e-ih2-ét > -oyh2éh1 > -aya:
Note that this also explains the Slavic ins.sg. -ojoN <
*-oyh2áh1 + -m, and the Armenian feminine gen/dat/loc/ins
-oj^ < *-oyya: < *-oyh2ah1 (e.g. kin -> knoj^ "woman", mi ->
mioj^ "one", ea-stems -woj^).
The irregular forms now seem to become -a:s, -a:i, -a: for
expected -oya:s, -oya:i, -oyya:. While it's possible that
-y- was irregularly lost to shorten these long endings, it
seems more likely that the forms -a:s, -a:i, -a: descend
directly from a non-mobile IE paradigm, e.g. for
Balto-Slavic:
gen. *-é-(i)h2-os > *-áh2os > *-ã:s
dat. *-é-(i)h2-ei > *-áh2ai > *-ã:i
ins. *-é-(i)h2-eh1 > *-áh2ah1 > *-á:
*-e-ih2-éh1 > *-oyh2áh1 > *-oyyá:
(the instrumental has a strong tendency to attract the
stress).
The existence of the two types is easily explainable: the
non-feminine thematic stems have columnar accent, the
feminine *ih2-stems (being ordinary C-stems) have mobile
accent. The a:-stems are a combination of the two: thematic
vowel + *-ih2. The accent could (and did) go according to
either model.
=======================
Miguel Carrasquer Vidal
mcv@...