Re: [tied] Some lengthened vowel Slavic verbs

On Sub, lipanj 24, 2006 2:06 am, Miguel Carrasquer reče:

>>> The lengthening of the root vowel in these verbs must
>>> therefore be relatively ancient, as it follows the PIE
>>> distribution where the only long /i:/ and /u:/ were acute
>>> (from /iH/ and /uH/), while a:, e: and o: could be either
>>> acute or circumflex.
>>
>>I don't understand the point. Do you wish to propose that these new long
>>*i: and *u:'s got the acute because *i: and *u:'s had the acute always
>>originally?
>
> They got the acute beacuse at the time there was no such
> thing as circumflex long /i~/ or /u~/. Only /a~/, /e~/ and
> /o~/ had been inherited from PIE. I agree with your earlier
> remarks that the story of long (lengthened) vowels in Slavic
> is complex and contains several chronological layers, but
> realizing that lengthened /i/ and /u/ could only become
> acute /i:/ and /u:/ at a time when circumflex /i~/ and /u~/
> simply didn't exist in the phonological inventory allows us
> to economize on the number of layers. Sy"pati, my"kati,
> smy"kati, ty"kati, sy"sati, pry"skati, bry"zgati, sti"gati;
> and skaka"ti, xapa"ti, xrama"ti, maka"ti, kaza"ti, dre^ma"ti
> can be put in the same chronological layer without
> contradictions.

Yes, that seems like a possible solution. However, I think that the
situation in Slavic iteratives is a little bit more complicated and that
the distribution of the acute and circumflex is not so clear as it seems.
For instance, in Croatian there is o`tvarati besides otvárati "open" and
hra``mati besides hrámati. And usually, the long rising accent is
secondary in cases like these.
Could you provide me to an exact reference where Dybo talks about these
iteratives?

> The whole point of the exercise into lengthened-grade
> i-verbs was precisely to determine whether PIE lengthened
> grade yields acute in Balto-Slavic or not, as the evidence
> usually adduced in discussions on the subject is largely
> irrelevant or inconclusive.

I agree completely.

> My preliminary conclusion is
> that it does: the verbal category itself can be traced back
> to PIE times rather solidly (and Jens has provided a
> rationale for the presence of the long vowel). If the long
> vowel is inherited from PIE, then the fact that ga"ziti,
> pa"riti and va"diti are a.p. a can only be explained if this
> lengthened vowel indeed yields a Balto-Slavic acute.

Where is the proof for the PIE lengthened grade for these Slavic verbs?

> The intonation of Lithuanian akmuo~, dukte:~, Slavic
> z^e``ra:vU (< *gero:us) is therefore as expected: the long
> diphthongs *-o:n, *-e:r and *-o:u yield an acute vowel
> wrapped in a circumflex diphthong.

Slavic z^erav is not a proof for anything. Croatian z^e``ra:v is mobile
and thus irrelevant (the possibility of Meillet), Czech z^eráv does not
agree with Croatian as it points to *z^era~vU (or to *z^e``ra:vU but with
a secondary lengthening in front of the final -v), and Slovene z^era`v
does not agree with any of the mentioned forms. Thus, it does not seem to
be the best example to take as evidence for the reflection of the old PIE
lengthened grade...

> Original root nouns
> ending in a resonant must also yield Balto-Slavic circumflex
> as is the case with Lith. ge:lŕ / Slav. z^âlI (*gWe:l-s),
> Lith. z^ole:~,

Lithuanian vocalism here points to a secondary Baltic ablaut. Thus, it is
not relevant. Slavic z^alI is mobile and also irrelevant.

> Lith. me:sŕ, Latv. měesa, Slavic męNso
> (*me:m-sóm),

Lithuanian me.sa` is possibly a Slavic loanword or a word influenced by
Slavic (Fraenkel), thus not the best evidence. Slavic męso is mobile and
thus irrelevant.

> Slav. jâje (*o:u-yóm),

Mobile > irrelevant/inconclusive.

>Latv. sŕ:ls (*sa:ls),

I cannot accept long PIE *a: of non-laryngeal origin, but we've discussed
this before.

> guňvs (*gWo:us). But when a (non-resonant) consonant
> follows, we have Latv. năss, năsi (*na:ss, *nasm.),

The same as with the 'salt' word.

> with
> what looks like an acute. The Slavic examples with a
> consonant after the long vowel given by Kortlandt are:
> (vodo-)tęc^I, ręc^I, (noc^-)lęgU, sapU, slępU, krâsU,
> (u-)z^âsU. If I'm not mistaken, they are all mobile (as
> they should be, if from original root nouns), so Slavic is
> inconclusive: Meillet's law should have eliminated the
> acute. Kortlandt does not give Baltic counterparts for any
> of them.

Many of those examples are indeed mobile and irrelevant, that is true. But
not all. Cf. Czech c^ára, Ukr. c^ará with the exact parallel in Avestan
c^a:ra:. There are more examples like this, but Kortlandt as usual gives
only hints and one has to look for the examples by himself.

> The only solid piece of evidence I can think of of a word
> that indeed suggests a development long vowel > circumflex
> is Lithuanian me:nuo~ (< PIE *méh1no:ts, PBSl *meh1nó:ts),
> where we would have expected *me:nů. However, the ending
> -uo~ here could easily be analogical here after s^uo~,
> píemuo, akmuo~ etc.

I think you are right. It could be analogical, so it's not really conclusive.

> The main part of Kortlandt's article is devoted to the
> lengthened grade s-aorist. I don't have the time to go into
> that right now, and I am still investigating the matter. My
> first impression is that the evidence from the aorist is
> again largely inconclusive, because of the complete mixture
> of forms from s-aorist, sé-aorist, root-aorist, é-aorist and
> thematic imperfect. A form like 1sg. ręxU (with neo-acute
> *ręxÚ > *rę'xU), while it has the vowel of the s-aorist,
> continues the _accentuation_ of the sé-aorist (*-sóm), which
> never had a long vowel; and 1pl. ręxňmU, while it has the
> desinential structure of the sé-aorist, must continue the
> accentuation of the root-aorist or imperfect (*-(x)omÚ), as
> original *ręxómU (*-só-mos) would have had to retract the
> stress to *rę'xomU by Stang's law.

Kortlandt's aorist examples are irrelevant since the long vowel is always
pretonic, like in his Paradebeispiels do`nijeh and zakle~ (also ri``jeh
etc.). The acute would be eliminated in the pretonic syllable so the
examples are not relevant.

Mate