>> -----Original Message-----
>> From: Jens Elmegaard Rasmussen [mailto:jer@...]
>
>>The
>> example is even so good as to be
>> decisive: the Dehnstufe-to-circumflex theory is simply wrong.
>
> What _is_ the source for the Balto-Slavic circumflex then (except
> *-V(H)V-)?
Okay, in Common Balto-Slavic, I take it? That is, discounting Slavic
mobililty and East Baltic accent retractions which produce circumflex. The
only category that remains seems to be that of monosyllables. There is
circumflex on Slav. ta, ty, vy, also Lith. tie~ (as opposed to geríe-ji),
tuo~ (as opposed to gerúo-ju), ju:~s (as opposed to gen. jú:suN, from
where the variant jú:s can have taken its acute, while the reverse in
impossible); there is Fallton on Latv. gùovs 'cow' and sà:ls 'salt' (as
opposed to Lith. sólymas 'Salzlake'); circ. on Sl. aor. by 'was', da
'gave' (Skt. bhu:t, da:t if you drop the augment) opposed to the inf.
by"ti, da"ti (contra aor. bi" 'hit' which complies with the inf. bi"ti and
so may reflect levelling); root nouns like Sl. tva^rI 'shape', va^lU
'roll', re^c^I 'speech' and others ; Lith. dúosiu dúosi dúosiva dúosita
dúosime dúosite vs. 3rd person duo~s is almost too clear.
Long monophthongs with acute in old polysyllables are (further) seen in:
Sl. *se^k-ti 'to cut' (SbCr. sj"ec´i), Latv. smiêtiês 'to laugh', Lith.
nósis (from *na:s-i-, from IE root noun with the ablaut seen in Skt.
na:s-/nas- 'nose, nostril' excluding laryngeals); Lith. z^ve:´riN (without
laryngeals, cf. Lat. ferus with short e); Lith. loky~s, lókiN 'bear'
(derived from làkti 'to lick' despite Kortlandt who finds the connection
semantically strange; the Slavs apparently didn't ask him before they said
medUve^dI). There are others, but these are the ones that come to mind
right now.
The important thing is that this is the only picture in which the forms
that are explained by the rule are the ones that could not well be
analogical. I know there are variants almost everywhere; but if we
discount what is productive and go by froms that are opposed to the
general trend and to allied forms, then these rules present themselves
with great force.
Jens