Hello Piotr,
The existence of an intervocalic s that passed to h in Proto-
Albanian is strongly related to the correct derivations for Alb. kohë
and Alb. gjuhë
(also vjehërr is a strong candidate too)
I. Alb. hohë - Sl. c^as& - OPruss. kisman < PIE *ke:s-a:
=========================================================
See Derksen Derivation: PIE *ke:s-
====================================
"Proto-Slavic form: c^as&
Old Prussian: ki:sman `time' [Accs]
Other cognates: Alb. kohë `time'
Proto-Balto-Slavic reconstruction: ke:s-"
See Demiraj derivation: PIE *keh1s-(u-)
========================================
"
Albanian form: kohë [f] (tg)
Meaning: time; span; weather; epoch
Proto-Albanian: ke:ua: {1}
Page in Demiraj AE: 221
IE reconstruction: keh1s-(u-)
Meaning of the IE root: time?
Page in Pokorny: 636
Other IE cognates: OCS c^as& `time'; OPr. ki:sman `time'
"
Demiraj preferred s > zero and u > h but the original root is
considered ke:s- too.
Your proposed derivation was:
==============================
*keh1k^-sah2 > kohë 'time' (cf. Slavic *c^asU 'id.')
Now to analyse your derivation:
=======================
1. For kohë you have used an additional k^-sa: in place of -sa: again
only not to explain the h from a simple s
But there is no need to add an 'additional k^' (see also Demiraj
and Derksen derivations above using ONLY an *V-s-V).
So when I previously have talked about 'additional k^ used as
an 'workaround' in your derivation' I made reference at the k^ in your
derivation above.
2. In addition : an original k^s will give in Albanian th and NOT h:
k^s > tss > ts > th see:
Alb. djathë and djath-të 'right' but also ndjath 'to the right' < PAlb
*deca- /detsa/ < *decsa /detssa/ < PIE dek^s-
See also Albanian ith < PIE *eg^hs (for a similar output: g's > dzs
> ts > th )
3. So you need to use an original *sk^ or a *sk but not a *k^s in
order to obtain the h.
Only sk^ or sk gave h via a LATER Metathesis to k(^)s.
But this later k^s (I don't think that existed, it was only ks, but
let's accept k^s too for the argumentation) it was at least another
k^s with no link with the original one that merged already k's >
c /ts/ > th
Note-1: In a very good article (2001) Lubotsky came back to Meillet
position showing that there was no oposition between sk' and sk in
Satem Languages.
So sk' > sk already in IE. But anyway an original PIE *k^s will not
generate h in Proto-Albanian, you need an original sk^/sk for this.
II. swek^uro > swesuro
========================
For swek^uro you have used a supposed metathesis wesk^ura to obtain a
sk^ for the Albanian h in place of a simple s
But you don't have any other example of such Metathesis: swe > wes
Also a derivation like the one below don't raise any issue:
*swek^ura < [alternance k^<->s as in thumbull<-> sumbull] < *swesura >
[sw>w] > *wesura < [V-s-V > V-h-V] > *wehura
III. gjuhë
===========
I don't know your derivation for Alb. gjuhë 'tongue' - but I'm sure
that a k^, k^-sa:, s-k^a: will appear as an workaround to obtain the h.
Alb. gjuhë is very probable cognate with Sl. go:ls& -> Once again we
have s in both forms with no additional k^ etc...
See Derksen:
=============
"Proto-Slavic form: go:ls&
Old Church Slavic: glas& `voice' [m o]
Indo-European reconstruction: golH-so-
Page in Pokorny: 350
Other cognates: W galw `call' [verb]
"
For the derivation of Albanian gjuhë:
PIE *gl.sa: (zero-grade) > *gulsa: [l. > ul > lu] > *glusa: [gl > gj;
V-s-V > V-h-V] > gjuhë
For both forms, Slavic and Albanian, we have only an original s (no
trace of k^), that became intervocalic s in Albanian after l. > ul > lu
(ul>lu because otherwise an *ls will never gave h in Albanian (also gl
> gj is the regular output))
Based on all these example the existence of an intervocalic s that
became h in Early Proto-Albanian is evident.
From here the next step is : we need to manage two different ouputs in
Alb for an intervocalic PIE *s: h and sh.
For this we need to find another solution than to simple add a k^ each
time when we have an h related to an intervocalic s.
Best Regards,
Marius Alexandru