isolates

[Alexander Vovin]

Reply follows.
 
On Tue, 25 Mar 1997, Larry Trask wrote:
 
> Alexander Vovin writes:
>
> > I would like to offer some corrections for the following passage in
> > Larry Trask's posting, as some of the claims regarding the certain
> > Asian languages as isolates are out of date, I am afraid.
>
> Well, I am happy to be corrected, but I have my doubts about some of
> what follows.
>
> >     First, there is a general consensus nowadays among all linguists
> > working in historical Korean and Japanese that Korean and Japanese
> > are related.
>
> "All"?  I doubt it.  I know that this idea has been gathering support,
> but it's news to me that *all* specialists now accept it as
> established.  Anyway, I can name one specialist who certainly doesn't
> accept this "consensus": Masayoshi Shibatani, who has repreatedly
> characterized the proposed Japanese-Korean link as unsubstantiated.
> And Shibatani also repeatedly points out that views on the possible
> affiliations of Japanese are numerous, varied, and much debated.
 
        Shibatani is a syntactician, not a historical linguist. His most
significant contribution to the historical Japanese linguistics is a
chapter in his "Languages of Japan", which I believe, is the weakest part
of the otherwise excellent book (I often use it as a textbook in my
classes): it discusses indiscriminately various hypotheses of Japanese
origins (without any reference to comparative method), and is seriously
flawed and/or outdated in other respects. There are a number of
"hypotheses" flourishing in Japan about the genetic origins of
Japanese (probably no less than about Basque),
but the general level of comparative linguistics in Japan remains very low
(with several important exceptions, such as Hattori, Murayama, Osada
etc.), as it is not based on comparative method, but on kokugogaku
(national linguistics) methods undistingushable from folk etymology.
Shibatani mentions several of them, but try to find any substantial
discussion of any -- you won't. The center of research on Korean-Japanese
relationship is now in the States, and it is one of the main
cointributions of the scholars who belong to S. Martin's school. Anyway,
let me refer you to the J. Whitman's dissertation: "Phonological Basis for
the Comparison of Japanese and Korean", Harvard 1985, which together with
a couple of articles published recently by S. Martin in Baldi's volume and
in "Sprung from some common source" should persuade you that the two
languages are related (I can provide you with some further reading, if you
wish so). Well, even ground-breaking Martin's 1966 article is still valid
at 75% percent, but you have to disregard the reconstructions proposed
there. Well, anyway, among the people who are actively engaged in research
on Japanese and Korean (and Shibatani is not one of them) it is accepted
that two languages are related.
 
>
> > Their link to the rest of Altaic remains more disputable, but even
> > here majority of the scholars lean toward the acceptance of genetic
> > relationship between Japanese, Korean, and Tungusic.
>
> Well, even if this is true, "leaning toward the acceptance" of a
> hypothesis is hardly the same thing as seeing it established to
> general satisfaction.  As far as I know, Japanese has not been
> demonstrated to be securely related to anything else at all, and it
> remains an isolate, as does Korean.
 
 
Well, please see above. Japanese and Korean have been demonstrated to be
related. That's enough for them not to be isolates.
 
>
> > In addition, none of these proposals are long-range, in
> > the sense of long-range proposals concerning genetic affiliation of
> > Basque or Nihali. Besides, both "Japanese" and "Korean" are to a
> > great extent sociolinguistic terms: under "Japanese" we have a
> > number of mutually unintelligible languages or "dialects", as they
> > are usually called, with the depth comparable to that of
> > Germanic. The same is true of "Korean": there are at least three
> > Korean languages, again normally called "dialects", mutually
> > unintelligible and with the depth comparable to that of
> > Slavic. Thus, even if they were not related (but they are), each of
> > them would represent a mini-family, and not a true isolate.
>
> This is news to me, and I am skeptical.  No reference source available
> to me even so much as mentions the possibility that either Japanese or
> Korean might plausibly be regarded as a family of three or more
> languages.  The most I have seen is a suggestion that the highly
> divergent Ryukyuan varieties of Japanese might be regarded as a
> distinct language, but even this appears to be a minority view today
> (though it was formerly more prominent).
 
I can only say that you rely on sociolinguistically oriented sources. It
is
a matter of national policy in both Japan and Korea that everyone is
"Japanese" and "Korean" and there is a "great national unity". This is not
true as long as language divergence is concerned, and the break-off is not
just bewtween Ryukyuan and Japanese. There are at least 5 branches within
Ryukyuan: Okinawan proper, Northern Okinawan-Amami, Miyako, Hateruma,
Yonaguni. None of them is mutually comprehensible, and they are all very
divergent. The same is true about mainisland Japanese: a person from Tokyo
will not understand a person from Toohoku )north-east), and will barely
understand a person from Kyuushuu. A close situation exists in Korea:
while most dialects spoken on the Korean mainland are mutually
intelligible (Korean proper), a dialect spoken on the island of Ceycwuto
(Chechudo, Quelpart) is too divergent to be understood by a person from
Seoul. THe same is applicable to the Northeastern Hamkyeng dialect, spoken
also in adjacent regions of China and in Russia, on which I myself have
done a fieldwork. The mutual comprehension is almost out of question, and
the languages have diverged to the point where they have almost different
verbal suffixation. Hope this helps to dissolve your doubts.
 
>
> >     Second, over last seven years there has been presented quite a
> > chunk of evidence allowing us tentatively to link Ainu with
> > Austronesian and Austroasiatic.
>
> > There is a consensus between anyone who ever tried to venture in
> > this field (late Murayama S., L. Reid, J. Bengtson, P. Sidwell,
> > I.Pejros, Yanagizaki Y. , and myself) that Austric connection is the
> > likeliest for Ainu. Although, none has yet come up with a proof
> > beyond the reasonable doubt, this is the only connection among
> > proposed for Ainu that cannot be easily disproved (contrary to the
> > case, e.g., with Ainu and Altaic). Thus, this is situation
> > remarkably different from Basque situation, where majority of
> > specialists reject any remote proposals.
>
> But no relationship has been established between Ainu and anything
> else at all, and it remains an isolate.  Having a favorite conjecture
> is a far cry from having a persuasive case.  Anyway, even Austric
> itself is not generally accepted as a valid construct.
 
It has been established that Ainu is UNLIKELY to be an isolate: for me it
is enough to take off the list of definite isolates, if we want to be
completely honest with ourselves. Let's place it into intermediate group.
Austric (as consisting of Austronesian and Austroasiatic, but not
including Kadai) is accepted
nowadays by all known to me leading scholars in both Austroneasian and
Austroasiatic. It is even accepted by R. Blust, who is one of the most
carteful historical linguists known to me.
 
>
> >      Third, since both Ket and Yukaghir did have relatives in the
> > recent past, this very fact seems to work against Larry Trask point
> > that there are more isolates among us than we suppose. As a matter
> > of fact, it shows us how isolates come into being: by extinction of
> > the other family members. Thus, there are probably less true
> > isolates than we suppose.
>
> I don't follow this at all.  If we can watch isolates come into
> existence before our eyes in historical times, then why is that an
> argument against the existence of numerous isolates which lost any
> relatives they may have had ages ago?
>
 
Nope, it is an argument against your point that there are more isolates
around us than we tend to think. About modern status of Ket and Yukaghir,
please see my reply to Thomason.
 
> > In particular, as I intended to demonstrate above, Larry Trask's
> > Asia list can be safely reduced to Gilyak, Nihali, and Burushaski.
>
> "Safely"?  I don't think so.
 
   As I always say, let us discuss the evidence.  I remove Japanese
and Korean from your list, and since it is accepted by virtually everyone
who works in the historical and comparative Japanese field (using of
course, comparative method, and not folk etymology), I believe that the
burden of proof that Korean and Japanese are not related rests on your
shoulders. Please present us with the evidence that the two languages in
question are not related, using exactly the same technique as you apply
for Basque: that is, showing that we have faulty etymologies etc. Then I
will be happy to present counterevidence, showing, e.g. why such highly
divergent words as Tokyo Japanese isi and Seoul Korean tol "stone" are in
fact cognates.
 
>
> > Among those remaining three, Gilyak looks like a Nostratic language
> > (although it has yet to be proven),
>
> Er, um...but *Nostratic* is still very far from being established.
 
I don't know anything about the Southern branches of Nostratic:
Kartvelian, Afrasian, and Dravidian (and I have grave doubts about
Nostratic validity of the latter), but there are very few doubts in my
mind that IE, Uralic and Altaic are related -- I have an article in a
forthcoming volume on Nostratic from John Benjamins. Anyway, I believe
Nostratic is not yet finally established, but I wouldn't call it "very
far".
 
>
> > and I even will not be very surprised if ultimately it turns out to
> > be a very abberant member of Altaic family: after all, scholars so
> > far applied only internal reconstruction to Gilyak, but we should
> > not overlook the fact that what we call Gilyak actually consists of
> > three or four mutually untelligible and pretty much divergent
> > languages, the fact that allows comparative reconstruction that is
> > yet to be done.
>
> Interesting, certainly, but Gilyak is still an isolate.
 
Rather, we should call it a small family with no apparent relatives. But
it is not a single isolate -- the same as about Ket and Yukaghir.
 
Alexander Vovin
 
>
> Larry Trask
> COGS
> University of Sussex
> Brighton BN1 9QH
> UK
>
> larryt at cogs.susx.ac.uk
>
>