On Thu, 06 Jan 2005 13:19:14 +0000, Sergejus Tarasovas
<
s.tarasovas@...> wrote:
>--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:
>
>> As to your question how a hypothetical PIE/PBS *gWr.H-jó:
>> (syllabified thusly) would have come out in Modern
>> Lithuanian, I cannot tell. It depends on the exact nature
>> of the acute diphtongs: did *r.H give /i:r/ with acute
>> _length_ on the vowel (I don't think so, because that would
>> have made the _syllable_ circumflex, as e.g. acc.sg. fem.
>> *-ah2m with acute vowel, but circumflex syllable)
>
>I must admit that the notion of a circumflex syllable with an acute
>nucleus doesn't make much sense to me.
It's the further tonal elaboration of the scheme I presented
here last year, I think, to explain the (Balto-)Slavic case
endings. As you'll recall, I posited five principles, only
one of which had something to do with tone:
1) circumflex raising (a~ > ó:, e~ > í:, o~ > ú:, u~ > ú:,
i~ > í:)
2) N-raising (-aN > -uN, -a:N > -o:N, -o:N > -u:N)
3) h-raising (-as > -uh, -a:s > -o:h, -o:s > -u:h)
4) Rh-lengthening (-VRh > -V:Rh)
5) long diphthong shortening (-V:R > -VR),
(4+5 simultaneous, of course)
followed by:
6) j-umlaut
I didn't say anything then about the tones or the accent.
What I would like to propose is the following procedure for
establishing the tone contour, at least for final syllables:
(short) vowels have falling tone (\), syllable-final
resonants (-i, -u, -m, -n, -r, -l) too (\). The combination
of short vowel + resonant (\\) gives /\ = circumflex.
Laryngeals (consonantal or vocalized) give rising tone (so
they must have passed through a stage where they had merged
into glottal stops). Lengthened vowels (\\) give
falling-rising (\/ = acute) intonation, but contracted
(superlong) vowels are (falling)-rising-falling (i.e.
circumflex = (\)/\).
If we apply this to acute diphthongs, e.g. {e/r\} vs.
{e/r\H/} (or {i/r\} vs. {i/r\H/}), the tone contours are: /\
(circumflex) vs. [/]\/ (acute). After the loss of the
laryngeal, the difference is maintained as {e/r\} vs. {e\r/}
(reflected as e.g. Russ. ére vs. eré). I don't think
there's any reason to assume length on the vowel. It may
exist in mod. Lith. dialects, which I'm assuming participate
in the general Lith. accent reversal (acute is falling,
circumflex is rising).
This simple procedure clears up a few difficulties in the
way the Slavic case endings developed vis à vis the PIE
prototypes. If we take the ah2-stems, the nom. sg. is PIE
*-a\h2/ (acute), giving Slav. *-a\:/ => -á. The acc.sg. is
PIE *-a\h2/m\ (circumflex), but the vowel is itself acute
(cf. Greek -á:n, -é:n). If the _vowel_ had been circumflex,
*-a:: would have been raised to *-o:, which would have been
further raised before -N to *-u:N, and the acc. sg. would
have been *-U instead of what it is: -oN (the development
was: -a:/m\ [2]=> -o:/m\ [5]=> -a/m\ => -oN(~)). The
genitive should have been *-a\h2/o\s => -a\:/:\s (and
further [1]-o\:/h, [3]-u\:/h => -ý), except that the ending
was mixed up with the acc.pl. ending *-a\h2/ms [2]=>
*-o\:/Nh [3]=> *-u\:/Nh => -ý / *-ja\h2/ms [2]=> *-jo\:/Nh
[3]=> *-ju\:/Nh [6]=> *-ji\:/Nh => *-je\:/(N) (i.e. -jeN ~
-jê).
The n- and r-stem m./f. nom.sg. endings can also be easily
analyzed using this procedure. PIE *-o\:/n\ (e.g.
*h2ak^mo:n) is circumflex, but the vowel is lengthened grade
(therefore acute, cf. Grk. akmó:n). PBS continues -o~,
which has the circumflex contour (\)/\ of *-o:n. *-o\:/:\
[1]=> *-u\:/ => -ý, as in <kamy>. The same goes for
*mah2té:r, *dhug(h)&té:r, where *-e\:/r\ => *-e\:/:\ [1]=>
*-i\:/ (mati, dUkti).
The o-stem abl.sg. was PIE *-o\o\d, PBS *-a\a\. The
uncontracted form was maintained in the Slavic pronominal
forms (*-a\a\ => *-a\ha\ => -o\go\ [N.Russ. -o\vo\]), but in
the nominal forms it was contracted to *-a\:/:\ (=> Lith.
*-a~ > -o~). Slavic has [1]=> *-o\:/ => -á.
The dative has circumflex syllable _and_ circumflex vowel.
PIE *-o\e\i\ contracted to *-o\:/:\i\. This gives Slavic
[1]=> *-u\:/i\ [5]=> *-u/i\ => *-ô (= *-[u/o\]) => -u(~).
The instr.pl. has a similar ending (Skt. -a:is, Grk. -oi~s
(with Osthoff-shortening)), but it's not possible to tell
from Slavic if the vowel itself was circumflex or acute: <a>
PIE *-oy-is > *-o\:/is [3]=> *-u\:/ih [4]=> -ý or <b> PIE
*-oy-is > *-o\:/:\is [1]=> *-u\:/ih [4]=> -ý. Given Lith.
-ai~s (not *-ui~s), I'd opt for acute *-o\:/is (cf. i- and
u-stem loc.sg. below, with a similar PIE contraction).
The instrumental sg. is a bit awkward. For systemic
reasons, I would like to reconstruct PIE *-o\e\h1/,
contracted to *-o\:/:\h1/. The result is still acute,
because of the final laryngeal, so we may write simplified
*-o\h1/ => -o\:/ => Lith. -ù, Slav. -á (in e.g. Russ.
vc^erá).
Then there are the problematical -oi endings: loc.sg. *-o\i\
= PBS *-a/i\, nom.pl. *-o\y = PBS *-a\j(??), NA.du.n.
*-o\i\h1/ => PBS *-a/i\(??). Cf. also the verbal optative
ending *-o-yh1(-s/t) => PBS *-a\j(??). The nom.pl. counts
as short in Greek prosody [consistent with *-o\j], and
doesn't trigger Saussure's law in Lithuanian, so it wasn't
acute *-a\i/, at least not initially. The Slavic
development *-ai => *-ee => -ê vs. *-aj => *-ej => -ii => -i
is also consistent with vocalic /i/ (mutual assimilation
with /a/) vs. consonantal /j/ (no change). But it's hard to
reconcile the Slavic and Lith. developments here.
The acc.pl. maybe was once *-oy-ms, contracted to *-o\:/ns
(the -n- is not syllable final, so prosodically neutral).
This gives acute *-u\:/Nh > -ý (and Saussure's law in
Lith.). The gen.pl. was *-oy-m > *-o\:/:\m\ (Grk. -o~n).
In the acc.pl., the result of the contraction *-oj- > -o::-
had probably been shortened before -CC#, which did not
happen in the gen.pl., hence the difference in vowel length.
Slavic has in the G.pl. *-o\:/:\m\ [1]=> *-u\:/N\ [5]=> *-uN
=> -U(~).
Another PIE contraction can be seen in the loc.sg. of the
i-and u-stems: *-ei-i > *-e:i, *-eu-i > *-eu-u > *-e:u (or
*-ou-i > *-ou-u > *-o:u). Slavic -i(~) can come from either
circumflex-vowel *-e\:/:\i\ (> i:i > i: > i) or acute-vowel
*-e\:/i\ (> ei > i: > i), but circumflex -o\:/:\u\ would
have given [1]=> *-u\:/u\ => *-y~ rather than -u~. I opt
then for acute length (within circumflex syllable).
The length in the acc.pl. forms *-ins, *-n.s > -i, *-uns >
-y is caused by the -Rh-lengthening law [4]. Likewise in
the ptc. endings *-o\nts > *-u\Nh > *-u\:/Nh > -ý and
*-e\nts > *-e\:/Nh > -eN/-ê.
The ins.pl. ending -mi is from acute *-mi\:/s [< *-bhi-is /
*-mi-is, vs. *-bh(j)-is / *-m(y)-is elsewhere], not
circumflex *-mi\:/:\s, because of Lith. -mìs (both would
have given the same result in Slavic).
=======================
Miguel Carrasquer Vidal
mcv@...