I have found a way to fit Slaaby-Larsen's law into my model
of Slavic accentuation, at least as far as verbs are
concerned (I haven't thought about the nouns).
To summarize my model: after Pedersen's law, Balto-Slavic
verbs fell into three accent paradigms: barytone, oxytone
and mobile (I, II and III). The barytone verbs were
relatively rare (before Hirt's law), as they continued
mainly the small category of PIE athematic barytone verbs.
Mobile thematic verbs were of course mobile (barytone
singular, end-stressed dual/plural), and they were joined by
the barytone thematic verbs (which became mobile by adopting
the stress pattern of the athematic verbs in the
dual/plural). Oxytone (theme-stressed) thematic verbs
constituted the "oxytone" accent paradigm (in Slavic terms:
classes II (ne-verbs), III (je-verbs), and IVa (i/i-verbs)).
[In the nouns, we have quite a different picture, with ap I
consisting of static athematics and barytone thematics, ap
III mobile athematics and oxytone thematics, while ap II was
confined to neuter oxytone thematics (peró, *meNsó, etc.)
plus compound words with dominant (stressed) suffixes like
*-ikós]
There were a number later rearrangements, notably:
- Hirt's law, which retracts the stress if a consonantal
laryngeal follows the vowel/diphthong in the immediately
pretonic syllable (i.e. thematic verbs are not affected, and
only the ablaut shape of the weak (dual/plural) stem is
relevant). Laryngeals were not consonantal in the sequences
*eRH, *arH, *CHC, *ih2/3, *uh2/3, but they _were_ in the
sequences *orH, *VH(C), *ih1, *uh1.
- Winter's law, which caused acute intonation of
vowels/diphthongs immediately preceding a *b, *d, *g, *g^ or
*gW (perhaps only when pretonic in the case of *VD?).
- Meillet's law: barytone forms of mobile paradigms acquire
a special accentuation (or lack of it), by which acutes are
lost (becoming circumflex) and stress is retracted to a
preverb/preposition, if possible.
- The "meNso"-law, by which the presence of a pretonic long
circumflex causes the accent paradigm to become mobile. This
only applies in an open syllable (meN~.só => mê.so, perhaps
sIr.dI.cé => sIr~.dI.ce; je-verbs: da~.jóN => dâ.joN,
ne-verbs: vi~.nó: => vî.noN, etc.).
- The "jablUko"-law, whereby a pretonic acute attracts the
stress (ja_blUkó => ja"blUko; vê_dê'ti => vê"dêti, etc.)
- Stang's law, which eliminates non-acute stress on all
medial (but not final) syllables: xodjóN, xodí~s^I, xodí~tI,
xodí~mU => xodjóN, xódi:s^I, xódi:tI, xódi:mU etc.). An
important side-effect is that mobile thematic verbs adjust
their stress pattern to match: nèsoN, nèses^I, nèsetI,
nesemÚ => nèsoN, neses^Í, nesetÍ, nesemÚ, etc.)
- Dybo's law: non-acute (falling) initial syllables pass the
stress to the next syllable (Stang's law does not feed
Dybo's, because Stang's law had resulted in neo-acute
initial stress).
- Ivs^ic'-s law: weak yers lose the accent to the preceding
syllable, if there is one, else to the next.
Where Slaaby-Larsen's law fits in in this scheme is as a
restriction on Meillet's law: the law fails if the syllable
is closed.
This explains the accentuation of short-vowelled C-verbs:
nés-lU, nes-lá:, nés-lo remain normally stressed (not
enclinomenic), so Dybo's law affects them and gives neslÚ (>
nèslU by Ivs^ic'-s law), neslá, nesló. C-verbs with an
acute remain acute, and are therefore subject to the
ja"blUko-law (kla"d.la, preN"d.la etc., but by.lá, li.lá,
etc.).
There is still one little problem with the athematic verbs.
On the one hand, Slaaby-Larsen's law explains perfectly the
accentuation of the present tense singular of by"ti (which
is anything but mobile):
ésmi Dybo: esmí
ései(?) Dybo: esí:
ésti Dybo: estí
esmú
esté
sontí
On the other hand, da(d)mI, ê(d)mI and vê(d)mI do not behave
as expected. If Meillet's law failed in a closed syllable,
we would expect *da"mI, *ê"mI and *vê"mI etc. Perhaps in
the cases of da"ti and vê"dêti we can count on analogy: the
aorist/l-ptc stem *da: (dâ-) has an open syllable and
therefore circumflex/enclinomenic/mobile accentuation (dâ,
dalá). The first person singular of vêdeti was earlier
vêdê, with an open syllable (so presumably accentuated
vê~dê, with Meillet's law). However, *êmÍ (> ê'm), êsí,
*êstÍ (> ê'stI) is a true counterexample. It must have
adopted to the accent pattern of the other 'mobile'
athematic verbs (by the way, isn't it funny that all
remaining athematics in Slavic --well, except esmI and
jImamI-- come from roots ending in *-d?).
=======================
Miguel Carrasquer Vidal
mcv@...