I was trying various queries against Derksen's Slavic
Inherited Lexicon, and I noticed some trisyllabic a.p. b
words which may be relevant to the question of which came
first: Dybo's law or the retraction of the ictus from weak
yers (Ivs^ic''s law). The words in question are bIc^elá
"bee" and vIdová "widow". Now, I believe that before Dybo's
law, Proto-Slavic had no non-acute interior stressed
syllables. Words had either fixed stress (a.p. I) on the
first syllable (or acute stress on an interior syllable:
jeNzy"kU, kopy"to, etc.), or were mobile (a.p.'s II and
III), with stress on the first or the last syllable (II and
III were in complementary distribution: where III had
initial stress (meN~so, ne``soN), II had final (peró, idóN),
and vice-versa (meNsá, neses^í vs. péra, ídes^I)).
Therefore, the proto-forms cannot have been *bIc^éla and
*vIdóva. The solution is to start from a mobile paradigm
*vIdová, acc. vÍdovoN. Then, weak yers lose their
stressability: *vIdová, *vIdóvoN. Now the newly stressed
vowel in *vIdóvoN acquires an accentuation which makes it
susceptible to Dybo's law, and we get a.p. b *vIdová,
*vIdovóN. Therefore, Ivs^ic' comes before Dybo. I had
earlier thought this was the right solution on the basis of
the accentuation of Russian nájden- "found": na-jÍden-
becomes ná-jIden (Ivs^ic'), and then Dybo cannot shift the
stress forward to a weak yer. But I cannot disprove the
alternate hypothesis that the stress was originally
*ná-jIden- > Dybo *na-jÍden > Ivs^ic' *nájIden-.
Another theme: final lengths. Dybo in various places
suggests that endings which were stressed in mobile (a.p. c)
paradigms were also long. In "Osnovy", he goes on
convincingly for several pages on the effects of this length
on Slovenian and other Slavic languages. It is also
confirmed by Kortlandt's analysis of the Slovene paradigm
for "horse" <kònj>, where, from memory, we have short <ò>
before a short yer in the ending (NAsg.), long <ó.> before a
long yer [in the Gpl.], <ó> before a formerly stressed short
vowel in the ending, and (neo-) circumflex <ô.>/<ô,> before
a long vowel in the ending [Lsg. -u [from the u-stems], Lpl.
-jih and Ipl. -ji]. I thought I was smart, and had come up
with the idea that the peculiar shortening in Czech kráva,
Gpl. krav may also be due to neo-circumflex metatony, but
Dybo already says that in "Osnovy".
The long endings should in theory be:
o/jo a:/ja: i u C
N -a: -y:,-i:
G -y: (-i:) (-u:) (-eC-e:)
L -i: -u:
I -ojoN:
n -a: (-eC-a:)
G -U: -U: -IjU: -ovU: -eC-U:
D -omU: -a:mU -ImU: -UmU: -eC-ImU:
L -ê:xU: -a:xU -IxU: -UxU: -eC-IxU:
I -y: -a:mi -Imi: -Umi: -eC-Imi:
NA
GL -u: -u: -Iju: -ovu: -eC-Iju:
DI -oma: -a:ma -Ima: -Uma: -eC-Ima:
This rule should also apply to the verb, but I'm not sure if
Dybo says anything about that. In the present thematic I
would expect:
1 -oN, always short
2 -es^I:
3 -etI:, -e:
1 -emU:
2 -ete:
3 -oNtI:, -oN:
1 -evê:
2 -eta:
3 -ete:
That the 1sg. was always shortened is an obvious fact (e.g.
Pol. 1sg. -eN vs. 3pl. -oN). I have the vague feling that
I've seen something somewhere about a 2pl. -ete:, but
perhaps I'm wrong. The 3pl. -oN(tI) and 1/2du. forms -evê:,
-eta: are long by nature, so their remaining long doesn't
surprise (although it should: 1sg. -oN was also long by
nature, but it was shortened). The often vocalized reflex of
1pl. -mU: (-me, -mo, -my) may, at least in part, be better
explained if the yer was lengthened. But the most important
thing is of course 3sg. -e:, which must be relevant to the
existance of lengthened thematic vowels in West- and
South-Slavic. If we had 3sg. nesé: (as well as boN"de: and
jìde:), the transfer of the length to 2sg. neses^Í > nesé:s^
follows immediately, without the need to invoke neo-acute
lengthening of a short vowel after the retraction of the
stress from the final yer (neo-acute doesn't hurt, but the
essential point is the length on the 3sg.: with that we can
just as easily explain a.p. a boN"de:s^ and a.p. b
jìde:s^). Transfer to the plural (and dual) forms with
thematic vowel -e- is a small step, and a paradigm like
Slovak nesiem, nesies^, nesie, nesieme, nesiete, nesú
follows quite naturally.
=======================
Miguel Carrasquer Vidal
miguelc@...