I'm going to assume for the time being that my hypothesis
about the exceptions to Hirt's law involving (i/u)h(2/3) is
basically correct. There are a number of good positive
examples (*pih3-, *bhuh2-, *gWih3-, *pluh3-, *kuh2-,
*ruh2-), and very few negative ones (Latv. gru~ts, if
*gWruh2tos [= Lat. bru:tiu?] was oxytone in PIE). If you
think about it, it makes sense that we should find
*(i/u)h(2/3) refelected differently from *(i/u)h1 in
Balto-Slavic. As Jens says in his article "IH, UH and R.H
in Indo-European", there is no reason to think that Greek,
Armenian *and* Tocharian (where the phenomenon has been
attested) form a valid subgroup of Indo-European. If this is
so, then the isogloss must also occur outside these three
groups, and Balto-Slavic especially can be seen as a
"bridge" here between Greek/Armenian and Tocharian. And if
Proto-Balto-Slavic already had *i&2/3 and *u&2/3 (later > i:
and u:), then Hirt's law cannot have worked here, because
Hirt's law only works on consonantal laryngeals, not
vocalized ones.
Meanwhile, while reading through the literature, I've come
up with other questions.
The central hypothesis of the "Moscow Accentological School"
is that the Slavic accent paradigms correspond to the
Lithuanian accent paradigms as follows:
Lith. 1 (not mobile, acute root) = Slav. a
Lith. 2 (not mobile, not acute) = Slav. b
Lith. 3 (mobile, acute) \
= Slav. c
Lith. 4 (mobile, not acute) /
This works fine almost everywhere, but there is a small
problem with the o-stems.
On pages 40-46 of Dybo's "Osnovy slavjanskoj akcentologii",
the correspondences are layed out:
Under AP(a), only 2 masculine o-stems are given, 8
adjectives, and 10 o-stem neuters.
Under AP(b), 17 masculine o-stems are given, but they are
all "accent paradigm d", i.e. they follow AP(c) almost
everywhere, exept perhaps a few relict areas. There are 8
adjectives here, and 17 neuter o-stems, which have all been
transferred to masculine gender.
Under AP(c), we have 16 masc. o-stems, 10 adjectives, and
only 2 neuter o-stems, both having a long (circumflex) root
vowel.
I have no explanation for the small amount of o-stem
masculine correspondences in AP(a), but Zaliznjak in his
lists (pp. 131-140) gives a rather healthy amount of o-stem
masculina following AP(a) in Old Russian.
My point is about the distribution of words in AP's b and c.
First, there is the problem of words such as dvòrU and and
sítU. They were originally barytone non-acute o-stem
neuters, so we would expect *dvoró and *sitó. Well,
*lautgesetzlich* we would expect dvòrU and sítU, since PIE
*-om regularly gives Slavic -U, but the question is why this
group of neuters has been singled out to follow the
soundlaws to the letter, while all others have replaced the
NAsg. ending PBS *-am with *-a (presumably from the neuter
pronominal ending -o(d) > *-a).
One might be tempted to think of a conditioned development,
where stressed *-ám becomes -o, while unstressed *-am
becomes -U, but this is completely out of character for a
Slavic soundlaw (no other vocalic developments from PBS to
Common Slavic depend on the position of the ictus), and it's
contradicted by cases where stressed *-ám becomes -U
(acc.sg.masc. *tom > tU) and where unstressed *-am gives -o
anyway (AP(a) síto, AP(c) mêNso).
The 2 neuter o-stem nouns which belong in AP(c) are
accompanied by a footnote where Dybo/Zamjatina/Nikolaev say:
"V ètu gruppu vxodjat lish' dolgotnye <oxytona neutra>,
refleksacija kratkostnyx <oxytona neutra> do konca ne
vyjasnena vvidu nedostatka materiala" (In this group there
are only long oxytona neutra, the reflexation of short
<oxytona neutra> has not been completely explained in view
of lack of material). This is a bit surprising, given that
there must be plenty of material. In no more than 15
minutes, I was able to collect the following sample:
*nisdóm > gne^zdó
*dhubnóm > dUnó
*pteróm > peró
*sidlóm > si(d)ló
*indróm > jeNdró
Apparently, the oxytona neutra with short or acute root
vowel give AP(b) neuters: exactly the category that was
missing because of the masculinity of dvorU and friends.
We seem to have a push or pull chain involving the following
categories:
1) expected AP(b) masculines => AP(c) [AP(d)] masculines
(e.g. zôNbU)
The empty slot is filled by:
2) expected AP(b) neuters => AP(b) masculines;
(e.g. dvorÚ > dvòrU)
The empty slot is filled by:
3) (some) expected AP(c) neuters => AP(b) neuters;
(e.g. peró)
The AP(c) neuters class does not empty, because the
circumflex o-stem neuters remain (e.g. mêNso).
My feeling is that the retention of the PIE ictus on neuter
oxytones (peró) is the most archaic feature.
=======================
Miguel Carrasquer Vidal
mcv@...