Re: Vasco-Caucasian and the comparatine method [was: Stacking up on

W dniu 2012-04-02 18:01, Tavi pisze:

> As documented by linguists such as Coromines, there's ample toponymy
> evidence of Basque-like dialects being spoken in the Pyrenees at least
> until c. 1000 AD before being replaced by Romance.

So what? Borrowings went both ways.

> > The usual Romance meanings seem to be 'membrane under the shell
> > of an egg; onion skin'. One possible VLat. source is *vinctia-
> > 'wrapping', admittedly a little speculative, but far less speculative
> > than what you propose.
> >
> I disagree. AFAIK this word isn't attested anywhere else in Romance, so
> a "Vulgar Latin" etymology is highly unlikely.

We have post-Class. Lat. vinctio:n-, VLat. *vinctiare (with attested
reflexes), and a regional cluster of Ibero-Romance forms that may
reflect a related noun. It adds up to a much stronger case than either
of your (or rather Starostin's) NC etyma.

> > *Problems on the "North Caucasian" side*
> >
> > The reconstruction is problematic. You use Starostin's reconstruction,
> > but note the author's comment: "The root is not widely attested in EC
> > (only in PTs [Tsetzian]), thus the etymology is somewhat dubious
> > (although phonetically and semantically plausible)." I may add that the
> > affricate of the supposed West Caucasian cognates is not the expected
> > reflex of PNC *3_ according to Starostin's own system. As the affricate
> > is the only segment that WC and Tsezian have in common in this root, the
> > reconstruction is in fact worse than dubious: it should be dismissed.
> >
> An irregular correspondence usually indicates borrowing on either side.

An irregular correspondence *usually* means a false etymology. And if we
are dealing with a wanderwort in the Caucasus area, it can't be used for
the reconstruction of Vasco-Caucasian.

> > *Problems with the comparison*
> >
> > The wide semantic latitude ('membrane' : 'steal, conceal') is the
> nail in the coffin for this etymology.
> >
> Should I remind you of Sanskrit *s´áras* 'cream, film on boiled milk' <
> IE **k´el-* 'to cover, to conceal'?

The fact that we have Old Irish súil 'eye', cognate to the IE 'sun'
word, does not give you a licence for equating 'sun' with 'eye'
everywhere. If you list all the meanings relatable to 'steal' or
'conceal', the list will be very long ('bury, veil, lid, crime, thief,
loot, secret' etc., etc., etc.). If you base your etymologies on
meanings "related" in this way, you are sure to find an accidental match
somewhere. Accepting a loose semantic match *before* you build a
convincing etymology is bad methodology.

> > *Problems on the "North Caucasian" side*
> >
> > To quote the author of the etymology again: "Reconstructed for the PEC
> > level. Not very reliable, because of the strange behaviour of the stem
> > in Lezghian languages; besides, labialised -3w- should not have yielded
> > -t.t.- in a cluster in PN. Contaminations of originally different roots
> > may be the reason". In other words, even admitting all potentially
> > cognate forms (which, however, do not obey Starostin's own rules), the
> > word is not really reconstructable as Proto-North-Caucasian. If one
> > eliminates the aberrant forms, the only thing that remains is Chechen
> > <battam> (not even securely Proto-Nakh), with not quite the right stop
> > in the middle.
> >
> AFAIK, this root corresponds to IE **(s)pondh-* 'wooden vessel', so if
> it exists in IE, then it must exist at all.

Even assuming that something like *spondHo- 'wooden bucket' is
reconstructible, I see no reason to connect it forcefully with Chechen
battam 'brass jug'.

> > *General problems*
> >
> > What are these two pairs of etyma supposed to demonstrate? The
> > correspondence of Basque <tz> : North Caucasian *3_(w)? They don't show
> > any such thing, since most of the NC forms quoted by Starostin have the
> > either the *wrong* consonant or some other irregularity.
> >
> > Even if both etymologies were flawless, two examples would scarcely be
> > enough to define a "regular correspondence".
> >
> But these aren't the only examples available, of course.

Of course? Where are the others? Is their quality better? If so, why did
you pick such hopeless stuff for the purpose of illustration?

> > However, both are seriously
> > flawed even within Starostin's system, and the corresponding PNC
> > reconstructions are unreliable by the author's own admission.
> >
> Nobody has said that Starostin's reconstruction of Proto-NEC (a better
> defined entity than NC) is the last word on the subject. Rather on the
> contrary, it's a pioneer work which allows for further improvements. In
> fact, i've heard about two specialists (Schultze and Nichols) working on
> this field.

The fact remains: the reconstructions on which you base your
Vasco-Caucasian scheme are bad enough to be useless.

> > *Conclusion*
> >
> > No valid evidence of anything here.
> >
> I *strongly* disagree.

No doubt, but who cares?

Piotr