[tied] Re: Convergence in the formation of IE subgroups

> "On Fri, 9 Apr 1999, Adam Hyllested wrote:
>
> [This is a reply to a posting on the IE list; the moderator has

asked

> for the thread to be moved to Nostratic.]
>
> [on Bengtson and Ruhlen's proposed Proto-World etymon for `dog']
>
> > Proto-World allright, but don't worry. The science of cross-
> linguistic
> > comparison is still at a cradle stage, and nobody can be *sure*
> about
> > anything. However, the examples are striking, at least some of

them

> (I
> > already mentioned the Nostratic ones):
>
> > Archaic Chinese: *khiw@... 'dog'
> > Tibetan: khyi 'dog'
> > Ket (Yenisey-Ostyak): ku:n~e 'wolverine'
> > Basque: haz-koin 'badger' (lit. 'bear-dog')
> > Proto-(North)Caucasian: xHweje 'dog'
> > Achomawi (a Hokan language): kua:n 'silver fox'
> > North Yana (Hokan): kuwan-na 'lynx'
> > Esmeralda (Equatorial): kine 'dog'
> > Pila (Papuan): kawun 'dog'
> > /'Auni (Khoisan): /ka~i~n 'hyena'
> > /Xam (Khoisan): !gwa~i~ 'hyena'
> > etc. etc.
> >
> > Coincidence? Maybe. Certainly not borrowing everywhere.
>
> Coincidence, almost certainly, apart from the odd borrowing,
>perhaps.

Because?

> But there's something even more fundamental to consider: do the

cited

> forms exist at all? B and R adopt a very cavalier approach to the
> treatment of data. Let me illustrate this with the Basque word

cited

> above.
>
> B and R cite the Basque word for `badger' as "<hazkoin>", insert a
> morpheme boundary not justified in Basque, and unhesitatingly gloss
> the
> supposed formation as `bear-dog', thus yielding them their required
> "<koin>" for `dog'. But most of this is fantasy.
>
> First, the Basque word for `badger' appears in at least seventeen
> regional variants. B and R have no business selecting just one of

the

> numerous variants merely because it suits their purposes: that is
> unprofessional. With this shabby methodology, they could equally

have

> selected some other single form if that had happened to suit their
> purposes better, such as <asku~> or <azkenarro> or even the very
> dubious
> hapax <akomarra>.

If we accept Trask's argument here, nor do B and R, or any other
linguist, have any business selecting English 'hound' over 'dog' for
purposes of linguistic reconstruction.

> Second, their cited variant, *<hazkoin>, does not exist. No such

form

> is recorded anywhere in any of my scholarly sources, and B and R,

who

> provide no source for the Basque data they cite, appear to have
> invented
> it. Apparently not one of the seventeen or more genuinely attested
> variants suits their purposes so well as this imaginary form.
>
> Third, it is *far* from clear that the Basque word is a compound
> meaning
> `bear-dog', or even that it is a compound at all. It is even far

from

> clear that the word is native.
>
> The Basque word for `bear' is <(h)artz>, itself often suspected of
> being
> a loan from an early IE language. The two Basque words for `dog'

are

> <(h)or(a)>,

which, if I'm not mistaken, could be from Proto-Basque *kora,
suspiciously similar to English cur, Finnish koira, Estonian koer

>today confined to the east but sparsely recorded centuries
> ago in the west, and <zakur>, found today throughout the country,
> either
> as such or in its diminutive form <txakur>.

for which another etymology has been proposed, but it escapes me

>Both words are recorded
> early. But there exists nothing resembling B and R's required
> *<koin>,
> which in any case would have a bizarre form for a native Basque
> lexical
> item. So, the cited Basque *<koin> `dog', presented as fact by B

and

> R,
> is a fantasy without justification. The cited comparandum is not
> Basque, but a kind of fantasy Basque apparently constructed merely

to

> allow B and R's comparison to go through.
>

> Anyway, the Basque word for `badger' is widely suspected of being a
> borrowing from late Latin <taxone>, or perhaps even from Celtic;

this

> was discussed a few days ago on the IE list.
>
> If you are willing to massage the data to this extent, you can
> probably
> prove any damn thing you like.
>
> Larry Trask
> COGS
> University of Sussex
> Brighton BN1 9QH
> UK"
>

Torsten