Re: [tied] Re: Decircumflexion, N-raising, H-raising: Slavic soundr

On Sat, 24 Apr 2004 13:32:42 +0200, Miguel Carrasquer
<mcv@...> wrote:

>On Sat, 24 Apr 2004 10:39:10 +0000, Sergejus Tarasovas
><S.Tarasovas@...> wrote:
>
>>The main difficulty was the fact that the "LDE shortening" one needs
>>to account for the *o-stems G.pl. conflicts, say, with *y of the *-
>>and *a:-stems Acc.pl. I thought that the shortening was blocked (or
>>the length was restored) before [#h], but the idea seemed too
>>speculative and ad hoc to me. You have overcome the controversy by
>>postulating a lengthening by *-Rs. This raises the question: are
>>there any other examples (except *-o::is, *-a:ms and *-o:ms you
>>listed) to support the rule? And why would one classify *i as a
>>sonorant (or at least put it in the same natural class along with
>>*m)? I think it's the crucial point for the whole idea, and an ad hoc
>>rule would deeply compromise it.
>
>The classical Indo-European and Balto-Slavic sonorants are
>/irlmn/ (or /ywrlmn/, whichever you prefer). Combinations
>of vowel + rlmn still behave like vowel + iu in Lithuanian
>for purposes of accentuation, don't they?
>
>As to the lengthening by final sonorant + -s, it seems
>natural enough (cf. Szemerényi lengthening in PIE, which
>applies to -Cs in general, but where -s is deleted only
>after sonorants). I can't think of any other examples right
>now, but if you look at the i- and u-stems, the fact that
>acc.pl. *-ns (*-ms) lengthens is undeniable.

One example that's not an acc.pl. that I had overlooked:
nom.sg. (masc. & n.) of the ptc.pres.act. *-onts > -y.
Quality is |u| because it's raised twice (by n and s),
length is |:| because of lengthening by -n(t)s (which
happens _after_ LDE).

Kortlandt's FPITS raises an interesting problem that I had
also overlooked: 3pl. aor. Slav. -oN < *-ont. We would
expect -U the way my rules are set up right now (-ont > -on

> -oN > -uN > -U). Kortlandt solves this by putting -om >

-um very early (in Balto-Slavic), but all the examples
adduced are invalid. The acc.sg. is -aN in Lith., the G.pl.
-uN does not come from *-om, and there's no -u(N) in Lith.
às^/ès^ "I" (nor do I think Slav (j)azU comes from *h1egom).

I would much prefer to put the loss of *-t very late (and
independently in Baltic and Slavic). That would
unfortunately divorce it from *-d, which indeed _was_ lost
early on (Kortlandt's example is *tod > to, not *ta, so *-d
was lost before Winter's law). But late loss of -t nicely
explains the Slavic 3rd. person ending -tU (with secondary
-U, as in azU < *e:g, izU < *eg^hz, etc.), besides -0 (both
from preterite/subjunctive *-t), and besides -tI (from
present indicative *-ti). So the ending was *-ont(U) at the
time of the soundlaw *-oN > *-uN, and it remained
unaffected, finally resulting in -oN(tU). In *-onts, the
/t/ was probably also deleted late, but before the *-Nh
lengthening.


=======================
Miguel Carrasquer Vidal
mcv@...