As I have mentioned here before, I think that Balto-Slavic
had three basic accent paradigms instead of two, as is
usually assumed. Besides the accent paradigm with fixed
stress on the root (a.p. I, Lithuanian a.p. 1/2), and the
accent paradigm with mobile stress alternating between
initial and final position (a.p. III, Lithuanian a.p. 3/4),
I believe there was also an accent paradigm (I call it a.p.
II) with fixed stress on the thematic vowel. In the verbal
system, this accentual type is abundant, comprising the
large majority of verbs in -je-, -ne-, -de- and -i-, and a
sizeable minority of the e-verbs.
In the nominal system, however, only two categories belonged
to type II: oxytone neuters in *-ó, and compound nouns with
a stressed suffix like *-ikó- or *-otá:.
Somewhere in the Proto-Slavic period, a sound law operated
which removed the stress from a medial non-acute syllable
(medial acute stress was allowed to remain: jeNzy"kU,
kolê"no, kopy"to, roNka"mi, etc.). This primarily affected
accent paradigm II (a.p.'s I and III were stressed on the
initial or the final syllable). In the verb, the
rearrangemnent is clear:
1 steljóN = steljóN
2 steljés^I > stéljes^I
3 steljétI > stéljetI
1 steljémU > stéljemU
2 steljéte > stéljete
3 stelóNtI > stéljoNtI
(The mobile type was rearranged to match:
1 bèroN = bèroN
2 bères^I > beres^Í
3 bèretI > beretÍ
1 beremÚ = beremÚ
2 bereté = bereté
3 beroNtÍ = beroNtÍ)
In the noun, we would expect:
NA peró = peró
G perá = perá
D perú = perú
L perê' = perê'
I perómI > *péromI
NA perá = *perá
G perÚ = *perÚ > pérU
D perómU > péromU
L perê'xU > pérêxU
I perý = *perý
The ins. sg. is actually perómI, which is not surprising
(regularization of the singular stress), but in the plural,
all forms are barytone, including NApl péra and Ipl péry.
This is seen in Slovene ókno (*oknó) pl. ó.kna, ó.k&n,
ó.knom, ó.knih, ó.kni (where /ó./ points to a vowel that was
"always" stressed), but in Slovene, retraction from a final
long vowel (NApl -a:, Ipl -y:) is regular, so doesn't prove
much. More significant are Serbo-Croatian, where we have
sèlo (< seló), pl. se``la, sêla:, se``lima (the falling
intonation (``, ^) indicates old initial stress), and
Russian seló, pl. sëla, sël, sëlam, sëlax, sëlami.
Illich-Svitych, in "Imennaja akcentuacja v baltijskom i
slavjanskom", tries to derive the initial stress in NA pl.
pèra from a Proto-Indo-European stress alternation in the
neuter plural (collective). He adduces only two examples
(Grk. neu~ron, neurá [with oxytone NApl.] and Skt. cakrám
vs. Grk. kúkla [with barytone NApl.]), which is not
surprising, as mobility of the stress in o-stems is
otherwise utterly unknown in PIE outside of Balto-Slavic.
In principle, it's possible that the o-stem NApl. ending was
athematic *-h2, which would explain the short -a in Latin
and Greek, and the stress of putative *pteróm, pl. *ptérh2.
But in that case, where does the Slavic ending -a (< *-ah2)
come from, which is the only NApl. ending in the language
and has spread even to C-stems, where the PIE form was
doubtlessly *-h2? In short, the possibility that the initial
stress of NApl. pèra is inherited from PIE, although I have
endorsed it in the past for want of a better explanation,
has never quite convinced me.
I am now convinced that it wasn't so. Because we also have
the other group of Proto-Slavic a.p. II nouns, the ones with
the stressed suffixes, which belong to the masculine o-stems
and the feminine a-stems. No stress alternation can be
inherited from PIE in those words, and still we have it.
My attention was drawn to the phenomenon by the Slovene Gpl.
form ó.v(&)c, with unexpected root-stress (the word is
otherwise regular a.p. c in Slovene). It struck me that the
stress and quality of the vowel were identical to those in
ó.k&n. In Serbo-Croatian, the "sheep" word is a.p. b in the
singular, a.p. c in the plural (òvca, òvcu, pl. ôvce,
ováca:, òvcama). In Russian, the final stress in the
singular shifts to the initial in the plural (ovcá, ovcú,
pl. óvcy, ovéc, óvcam, óvcax, óvcami). This pattern is
followed by a number of common a.p. b feminines (which
acquired b-stress by Dybo's law, so _not_ original members
of a.p. II: zená, z^enú, pl. z^ëny, z^ën, z^ënam etc. [no
doubt a key role was played here by the retracted stress in
the Gpl.]), but also as a rule by feminines carrying the
stressed suffixes -ota and -ina (temnotá, pl. temnóty,
temnót, temnótam, temnótax, temnótami; like the neuters of
the type res^etó, pl. res^ëta, res^ët, res^ëtam, res^ëtax,
res^ëtami).
In the masculines, we have Serbo-Croatian a.p. b kònac (<
konÍcI), with barytone plural kônci, ko``na:ca:, kôncima
(likewise, from Stankiewicz's list: lonac, novac, stolac,
la:nac, s^te:nac, te:lac and vije:nac). In Russian, a small
number of a.p. b masculines have a barytone plural. Most of
them end in -ok (*-ukós): glazók, glazká pl. glázki, glázok,
glázkam, glázkax, glázkami; further: zubók, roz^ók, kazák,
sapoz^ók, s^c^enók.
It looks as if the retraction in the NA and I pl. was
analogical after all, in order to create a new opposition
between singular (stress on thematic vowel) and plural
(retracted stress).
R = regular, A = analogical
(neuters)
NA peró R
G perá R
D perú R
L perê' R
I perómI A
NA pèra A
G pèrU R/A
D pèromU R
L pèrêxU R
I pèry A
(masculines)
NA konIcÍ R
G konIcá R
D konIcú R
L konIcê' R
I konIcémI A
N kònIci A
A kònIcy A
G kònIcI R/A
D kònIcemU R
L kònIcêxU R
I kònIcy A
(feminines)
N ovIcá R
A ovIcóN R
G ovIcý R
D ovIcê' R
L ovIcê' R
I ovIcéjoN A
NA òvIcy A
G òvIcI R/A
D ovIca"mU ~ òvIcamU R ~ A
L ovIca"xU ~ òvIcaxU R ~ A
I ovIca"mi ~ òvIcami R ~ A
In the feminine oblique, the acute stress on -a"- should not
have been retracted, so Russian forms like óvcam, -ax, -ami
are stressed completely on the analogy of the types of sëlam
and glázkam.
The creation of new mobility (oxytone sg. vs. barytone pl.)
in the a.p. II nouns must be placed before the working of
Dybo's law, because the accent was advanced again in forms
such as temnóty, temnót, temnótam, -ax, -ami and res^ëta,
res^ët, res^ëtam, -ax, -ami (but not where applying Dybo's
law would destroy the new mobilty, as in disyllabic pèra
etc.).
=======================
Miguel Carrasquer Vidal
miguelc@...