isolates

[Larry Trask]

Alexander Vovin writes:
 
> You of course are right that languages are not to be defined solely
> on the mutual intelligibility. Being a native speaker of Russian, I
> can understand Ukranian and Belourussian without any effort at all,
> and with more strain even good chunks of Polish and Bulgarian, which
> are all, of course, different languages. Here I used mutual
> intellgbility solely to demonstrate that there is considerable
> linguistic diversity in both Japan and Korea, which you seemed to
> doubt in your previous posting. I can easily demonstrate that we
> deal with various languages, not dialects by other means: showing
> that they have divergent morphology and lexicon. I can do it if you
> are still in doubt.
 
To be honest, I see little point in pursuing this.  If you prefer to
see Japanese and Korean as small families rather than as single
languages, fine.
 
> L.T.:
 
> > I'm sorry, but I simply cannot understand this.  You are telling me
> > that Ainu is "unlikely" to be an isolate, even though no relationship
> > has been demonstrated between Ainu and anything else at all.  I find
> > this position incomprehensible.
 
> A.V.:
 
> Should I repeat again that there have been established a number of
> Ainu-Austric parallels which do satisfy the principle of phonetic
> regularity? If under these circumstances you still maintain that "no
> relationship has been demonstrated between Ainu and anything else at
> all", I just give up. It has been demonstrated, but there is still
> many things left. Don't you think that this is a different situation
> from your Basque?
 
We are going in circles.  It is already clear from other postings that
the validity of Austric is not generally accepted, in which case no
Ainu-Austric link can possibly be generally accepted.  Moreover, I
think there are grounds for doubting your assertion that Ainu has been
shown, to general satisfaction, to be related to *something*.
 
> L.T.:
 
> > You are joking.  It is not on my shoulders to demonstrate that any
> > languages are not related; this is a logical impossibility.  If you
> > can demonstrate that J and K are related beyond reasonable doubt, I'll
> > be delighted, since I prefer positive results to negative ones.
 
> A.V.:
 
>    If you maintain the point of view that it is impossible to
> demonstrate that two or more languages are not related, then you
> obviously cannot demonstrate that Basque is unrelated to North
> Caucasian or whatever.
 
Yes, of course.
 
> Meanwhile, this is exactly the opposite what you did over the last
> year: showing your audience the impossibility of the connection.
 
Absolutely not: this is a fundamental misunderstanding.  I neither
provided nor even attempted any such demonstration.  What I did was
quite different.  Bengtson and his colleagues had put forth what they
regarded as evidence for a Basque-Caucasian genetic link.  What I did
was merely to demonstrate, to my own satisfaction at least, that the
evidence on offer *did not stand up to scrutiny*, and hence that their
case could not be accepted.  And that is a very different thing from
proving the absence of a relation.
 
So, Bengtson's case fails.  There remain at least three logical
possibilities:
 
(1) Basque and Caucasian really are discoverably related, but no one
has yet uncovered the available evidence.
 
I can't scrutinize a case that has never been made.
 
(2) Basque and Caucasian are indeed very remotely related, but the
evidence for that relationship has long since disappeared and cannot
be recovered.
 
There is no earthly way I could disprove this possibility.
 
(3) Basque and Caucasian are not related at any level.
 
Again, there is no earthly way I could prove this.
 
> What is true for Basque, should be true for other languages.
 
Yes, of course, but I've never maintained otherwise.
 
> Demonstrate that Japanese-Korean etymologies are either faulty,
> because they are build on erroneous reconstructions, or that there
> are no regular correspondences, or even better both -- exactly like
> you do with Basque. This is in my opinion a quite logical way to
> disprove a genetic relationship.
 
It is a very reasonable procedure, and indeed the only possible way of
disposing of a proposed comparison.  But it is most emphatically *not*
a way of disproving a genetic relationship.
 
The evaluation of your case will have to await the scrutiny of
specialists in the relevant languages.  If your case withstands that
scrutiny, you win; if it doesn't, you lose.  That's the way it is in
this business.
 
> A.V.:
 
> Again, neither Yukaghir, nor Ket are in fact a single language:
> there are two living languages in each case and more extinct
> ones. If you are going to call a family consisting of more than one
> language an isolate, then where is going to be the line?  >
 
Again, we are merely quibbling about words.  Being an isolate is not
an intrinsic property of a language; it's merely, as others have
pointed out, a property we project onto a language as a result of our
investigations to date.  It's purely a matter of taste whether we want
to apply the term to a living language with no known living relatives
but with known extinct relatives.
 
Larry Trask
COGS
University of Sussex
Brighton BN1 9QH
UK
 
larryt at cogs.susx.ac.uk