Re: [tied] Albanian intervocalic s > h

From: alexandru_mg3
Message: 38808
Date: 2005-06-20

--- In cybalist@yahoogroups.com, Piotr Gasiorowski <gpiotr@...>
wrote:
> alexandru_mg3 wrote:
>
> > There is no complex PIE cluster CCC in <djathtë>, because is a
recent
> > formation, Piotr. djathë is attested at BUZUKU (so with no -të). \
> > See Demiraj that consider djathë too. So we have an intervocalic
k^s
> > in djathë: PIE *-k's- > PAlb c > Alb th
>
> It can't be that simple. The <ja> diphthong (sorry for misspelling
the
> word in my previous posting) indicates a closed syllable even
before the
> addition of the secondary <-të>. Either *-k^s- was still a cluster
at
> the time when -je- and -ja- were differentiated in Proto-Albanian,
or
> there was another consonant that followed it. The prototype must
have
> been PIE *dek^si- with one of the few suffixes that normally go
with it
> (*-wo-, *-no-, *-tero-), with some word-final simplification later
on,
> most likely something like this:
>
> *dek^siwo- > *dec'(s)iw- > *diäcw- > *djacw- > *djac- > djath-
>
> The cluster *cw would have arisen too late to fall together with
> inherited *k^w > *c'W > *c^(W) > s, so the *w was simply lost, as
it was
> in most other *Cw clusters in Albanian.
>
> I accept the possibility that *-k^s- and *-k^- may have fallen
together
> (eventually yielding <th>) in Albanian; that would be similar to
what
> happened in Slavic. But even so, the alternative metathesis
*swek^uro- >
> *wesk^uro- remains an option, and *-sk^- certainly gives Albanian <-
h->.
>
> Piotr







Hello Piotr,

I. I agree with your derivation ja supposed still a PAlb. dets-sV-
so the stages were: k^s > cs > c > th

When I presented the rule: k^s > cs > c > th, based on djathë
(Buzuku), I didn't make reference to a wrong derivation for *swek^uro-

On *swek^uro- my reserve was related to your supposed metathesis:

*swe > *wes on which I have the following doubt:
How many examples showing *swe > *wes you know? If this is the
single one it will be difficult for your derivation to be accepted.

The 'alternance c<->s' is well attested (even its cause is not clear)
so a derivation like:
*swek^uro- > *swesura- > *wesura- > *wehura- is more probable.

(So a k^ <-> s like in sumbull <-> thumbull will be more easy to be
accepted )


II. To come back, I presented the rule: k^s > cs > c > th, based on
djathë (attested at Buzuku) to show that your derivation below:
*keh1k^-sah2 > kohë 'time' (cf. Slavic *c^asU 'id.')

(posted awhile ago) is not correct due to impossible k^s > h :

Now you have changed your mind (that is ok) and you said that Slavic
*c^asU is not linked in any way to gjuhë (even Demiraj, Derksen, Orel
still keep them as cognates (but of course doesn't matter how many
scholars support a different idea if there are arguments against it).

But for instance kohë > PIE *ke:s-a: is more more probable, showing
intervocalic s > h in Albanian.


III. Also the derivation gjuhë < PIE *gl.sa: seems ok too.

In conclusion : is difficult to reject intervocalic s>h in PAlbanian.

If we accept this conclusion : a second output for an intervocalic
PIE *s in Proto-Albanian become obvious.


IV. "Regarding *-sk^- certainly gives Albanian <-h->"

It could be very probable that sk^ merged to sk in PIE times (see
Lubotsky's article), so if true, this fact will put in cause your
metathesis's timeframe (*swek' > *wesk') that should be moved in
PIE's time, that for sure it wasn't the case.
Also there are not other examples showing *swe > *wes.

But if sk' merged with sk already in PIE is another subject. Now
regarding Proto-Albanian: we cannot detect any opposition sk / sk'.
There are no examples showing a different treatement so for sure we
can talk in Proto-Albanian only about sk > ks > h doesn't matter the
original source: PIE sk or sk'.

Best Regards,
Marius