Re: [tied] Balto-Slavic accentology

From: Miguel Carrasquer
Message: 35502
Date: 2004-12-19

More a.p. c verbs:

Inf. in -ati (e.g. bIrati). The present tense is mobile (<
PIE barytonic thematic *bhéro:, *bhéreti). The infinitive
*bhr.-ah2-téi gives bIrátei with Hirt's = bIra"ti. The
l-ptc. is mobile, after the present and/or after the 2/3sg.
aorist (bhr.-ah2-ét > bIrã[tU]), without Hirt.

We would expect verbs with a long circumflex root vowel
(e.g. daja"ti < *doh3-eie-téi ?) or with an acute root
syllable (no such cases for AP(c) -ati, buf cf. verbs in
-êti such as bêz^a"ti or sêdê"ti, instead of *bê"z^ati,
*sê"dêti) to have further retracted the accent in the
infinitive to *da"jati (by -Dybo or "circumflex metatony"),
but this does not happen, AFAIK (as it does in AP(a) verbs
such as vi"dêti, sly"s^ati, pry"gati, bê"gati). The
retraction was blocked here by analogy with the vast
majority of AP(c) verbs with infinitive in -a"ti, -ê"ti.

Verbs in -noNti. These have no business being mobile, and
indeed only a handful (those that have a root ending in a
vowel such as vi-noN-ti, -ma-noN-ti, mi-noN-ti, -meN-no-Nti)
are in AP(c), presumably because they can be interpreted as
plain e-stems in the present (vin-e-, man-e-, etc. instead
of vi-né, ma-né, etc.).

Verbs in -iti and -êti. These share a present tense made
with the theme -i-. Historically, we can distinguish
between causatives/iteratives (PIE R(o), rarely R(z) +
*-éie-), denominatives (PIE *-i-yé-) and essives/fientives
(aorist and inf. *-éh1-, present *-h1-yé-, according to
LIV). I just saw in Dybo et al. the suggestion that the
causatives simply have the a.p. of the non-causative verb
they are derived from. Of course. The same goes for the
denominatives. The iteratives, on the other hand, must have
already been a closed (non-productive) category in
proto-Slavic, and kept their original mesotonic accent
(-éie- > -í~). The AP(a) and AP(c) causatives had
unstressed -i:- as a marker, and this was transferred to the
AP(b) causatives, which therefore acquired stressed (acute)
-í:-, like the denominatives from -iyé- > -í:-, instead of
circumflex (superlong) -í~- (as preserved in the
iteratives). This explains the split between a.p. b1
(subject to Stang's law: nosi~ti > nòsitI) and a.p. b2 (not
subject to Stang's law, at least not in all dialects:
pojí:ti > poji'tI). The PIE origin of the ê/i-verbs is
obvious for the aorist/infinitive stem (*-eh1-, with Hirt's
law Slav. -ê'-ti), but the present tense formant -i- (Lith.
-i-, with _short_ i!) is problematic. It cannot derive from
*-h1-yé-, which would surely have merged with plain -yé- and
have given Slavic a.p. b jé-verbs and Lith. class I
ia-verbs. It also cannot derive from *-eh1-ye- (as the
Italo-Celtic e:-verbs), whch would have given, and
occasionally gives, Slav. -êj- and Lith. -e:j-verbs. Dybo
et al. suggest plain /i/, secondarily lengthened in Slavic
(like the thematic vowel /e/ > /e:/) and retained as such in
Lith., but there is no PIE verbal category with plain /i/ as
the "thematic vowel". Furthermore, there simply *must* be a
connection with the aorist/infinitive formant *-eh1-. The
only thing that occurs to me is a metathesized variant
*-ih1é- (instead of *-h1-yé-), which merged with
denominative *-iyé- in Slavic, but not in Lithuanian (where
-ie- > -i-).

The accent paradigm of -ê/i- verbs in Slavic is a.p. c (with
some a.p. a's due to Hirt's). The a.p. b's are missing,
which suggests another merger. If the accentuation was
*-ih1é-, the bulk of the ê/i-verbs are properly a.p. b, and
became mobile only secondarily (cf. Dybo, Zamjatina,
Nikolaev OSA - Slovar' 1, 1993, pp.31-43).

In the a.p. c iti-verbs, the fixed stress infinitive and
s-aorist forms show stress retraction by -Dybo (*lavi:téi >
lovi"ti, lavi:xám > lovi"xU). It is patently absurd to claim
(Lehfeldt 2001, p. 66), that "[d]er Akzent solcher Formen
wie avi"ti, lovi"ti; avi"xU, lovi"xU [] auf das uns bereits
bekannte HIRTsche Gesetz zurück[geht]".

Miguel Carrasquer Vidal