LINGUIST List 5.911

Sun 21 Aug 1994

Disc: Altaic

Editor for this issue: <>


  1. , Re: 5.910 Altaic

Message 1: Re: 5.910 Altaic

Date: Sat, 20 Aug 94 14:53:19 EDRe: 5.910 Altaic
From: <>
Subject: Re: 5.910 Altaic

Victor Golla's posting about the Penutian panel touched a nerve,
because what prompted my suggestion that we discuss Altaic was
my discovery of the rumors that have been circulating at third
or fourth hand about Altaic as a result of the so-called Altaic
panel at Stanford in 1987. Notably, I fear that much damage
has been done by Johanna Nichols' uncritical reliance on the
report of this panel by Unger in the 1990 book ed. by Philip
Baldi, Linguistic Change and Reconstruction Methodology, and by
perhaps by Unger's report itself. (I refer to Nichols because
her ebook has been so widely acclaimed and so presumably many
linguists and non-linguists--it's even been reviewed in Science--who
know little about Altaic will take what is said there as gospel.)

I say "so-called Altaic" panel, because Unger himself begins his
report by saing that the term was inappropriate given the composition
of the group, and Baldi has an editorial note explaining that some
of the invitees who were supposed to make it more Altaic did not
show up. Now, this panel consisted of three experts on Japanese
and Korean, one expert on Uralic and miscellaneous language groups,
and one person whose expertise I have not been able to trace who
gave a presentation to the panel (unpublished in the proceedings)
about not his own but rather Gerhard Doerfer's and Andras Rona-
Tas's objections to the reconstructions of Proto-Altaic by
Ramstedt and Poppe (which date the sixties and earlier!). No
specific works of Doerfer's or Rona-Tas's are cited and ther
is no discussion of the specific issues, and certainly no
mention of the extensive debates that have been going on ever
since Doerfer and Rona-Tas (following the lead of Clauson) started
attacking Altaic some decades ago.

This panel then, which for example did not include anybody who
is an expert on Turkic or anybody who has tried to work on Altaic,
concluded that they liked Doerfer's and Rona-Tas's objections, and
they also were skeptical of efforts by Roy Andrew Miller and John Street
to relate Japanese to the "so-called" Altaic languages. However,
two of the papers by two of the panelists that do appear in the
proceedings (Whitman and Samuel Martin) do in fact take Korean and
Japanese to be related, and Martin's paper, despite a snide remark
about the "Altaicists" appears to endorse the connection of both
these languages to Turkic, Mongolian, and Tungusic!

Now, I would like at some point in this discussion to address Doerfer's
and Rona-Tas's (and Clauson's) arguments, and Ron Hahn has already
made a beginning in this direction. I would like also to discuss
Nichols proposal that the systematic correspondences between the
pronominal systems of Turkic, Mongolian, and Tungusic are NOT due
to a genealogical connection but to some kind of areal phenomenon.
BUT, as St. Augustine said in another context, NOT YET.

I would first like to establish once and for all that one should
read and cite the relevant literature if one is going to criticize
a theory, rather that citing a panel report written by a non-specialist
about a panel of nonspecialists' views of an unpublished paper reporting
the work not of its author but those of other people entirely about
this theory, when moreover the views of these other people (Doerfer
and Rona-Tas) have been published in many many places over the
decades and have been the subject of an extended debate, in which
a number of distinguished linguists have in fact taken the other
(that is, the pro-Altaic) point of view (but of course we won't
mention these people or their work in our rush to condemn the theory!).

I would also like to establish once and for all that, whether
Altaic is a valid language or not, it is NOT true that as Nichols
claims "the received view now is that Turkic, Mongolian, and Tungusic
are unrelated (UNger 1990)" (Nichols 1992, p. 4). Even Unger does
not say this, and even his little panel did not conclude this, as
noted. And even if they had, they would not in anybody's estimation
represent the received view on a topic on which, among others,
we have the published works of people like Hamp, Menges, Miller,
Starostin, Vovin, Tsintsius, and many many others.

Moreover, the specific claim that Nichols makes about why these
languages are supposedly now perceived to be unrelated,namely
that there are "very few good potential cognates", aside from
the pronouns, is also factually incorrect. Even Doerfer and Rona-
Tas admit that there are many potential cognates, running into
the hundreds, but what they question is whether these are
cognates or borrowings. Moreover, Rona-Tas at least has
often said that he is by no means sure whether the oldest
layer of this "shared vocabulary" is due to borrowing or to

On the other hand, of course, while there have been four or
so major critics of Altaic, who keep propounding the borrowing
thesis, there are many more experts who have argued cogently
(whether we agree with them or not, an issue I want to leave
to another time!) that these must be cognates, in part because
of the formal and semantic properties of the forms of question
and in partbecause, while one can conceive of massive borrowing
from Turkic into Mongolian, it is less easy to see how these
alleged loan words would have ended up in Japanese, and in
part because it is difficult to see how such basic vocabulary
AND morphology would have been borrowed (unless the sense of the
word "borrowed" is watered down so that it subsumes "cognate").

It should also perhaps be pointed out that the arguments for
borrowing and against a genealogical relationship by Clauson,
Doerfer, and Rona-Tas, whether we accept them or not, are
(as has been pointed out many times, e.g., by Hamp, Miller, and
others) based on very unusual methodological premises, which
are NOT (pace Ron Hahn) shared by comparative linguists working
in any other field, esp. not Indo-European (Hamp I think made
this very clear in the 1970's). Thus, we find the assumption
that in order to be related two language (group)s HAVE to
share numerals. Now, this is certainly NOT something assumdd
by Indo-Europeanists, although it happens tobe true of the
Indo-European languages! Also, Clauson, who started the
whole attack on Altaic, took the very unusual tack of trying
to "prove" that these languages are unrelated (which, as again
noted by Hamp and many others) is not something that can be
done by any standard method of comparative linguistixs, and
as part of his "proof" made the further assumption (which
other anti-Altaicists appear to share) that in related languages
cognate forms should have THE SAME meanings, not merely related
ones (so that if a Turkic word's Mongolian counterpart has a
differetn meaning, then Clauson, Doerfer et al., will accept
that there was a semantic shift but it has to be borrowing
rather than a cognate). Clauson et al. make the further
assumption that in related languages certain terms other
than numerals, specifically body part terms, have to be
shared WITH SEMMLA(that is, with complete semantic) identity, so that Turkic
and Mongolian would have to have related forms for 'head'
or 'hand' to be related. Now, as has often been pointed out,
such an assumption is not made in any other field of comprative
linguistics. The fact that French tete and main are unrelated
to English head and hand is nOT taken as an argument against
these being related languages!

So, in conclusion, before we go any further, I would like to
see if we can at least agree that (a) critiques of Altaic
should not be done on the basis of rumor or third- or fourth-
hands reports, (b) we should not accept the findings of
a panel of nonspecialists, esp. when some of them in their
own work do support part or whole of the Altaic theory, whether
they call it that or not, (c) we should once and for all
stop pretending that there is any kind of consensus of
specialists that Altaic is dead, when in fact only a handful
of people have done the critical work and there are more
people than ever actively working ON Altaic, (d) we should
once and for all stop citing old criticisms of Altaic without
mentioning the extensive literature that has been devoted
to refuting these criticisms, (e) we should also not accept
uncritically critiques of Altaic which are based on methodological
principles which no one in any other area of linguistics seems
by Hamp, Ligeti, Krueger, Aalto, Poppe, Miller, and others.

That is, I have nothing against entertaining the possibility that
Clauson, Doerfer, and Rona-Tas (together with the fourth major
critic of Altaic, Shcherbak) might (a) be justified in using
these new methodologies and (b) be right in using them to
criticize Altaic. Or the possibility that Nichols might be
right that languages get to share pronominal systems as an
areal rather than genealogical phenomenon. Indeed, these
are entertaining possibilities. But if I am going to accept
these possibilities, then I want to get my money's worth.
Instead of relying on Nichols' exaggerated report of Unger's
report of a small panel of nonspecialists' reception of
an unpublished paper by one Larry Clark about the work of
Doerfer and Rona-Tas, I would like to see a response by
Unger or Nichols to the decades worth of work by all the
people mentioned above (Hamp, Krueger, Ligeti, Aalto, Menges,
Poppe, Tsintsius), all the many others I have forgotten to
mention, and (now this would really be entertaining) to
the most recent work on Altaic, by, e.g., Starostin and
Anna Dybo in Russia, or Alexander Vovin in this country.

Now just a few responses to specific points:

(a) Ron Hahn is right about the terminology, but 'Turkic'
and 'Tungusic' are already accepted terms. 'Mongolic'
would make a lot of sense, and I will use it in future.

(b) Ron is also right that numerals need not look related
among related languages. Cf. Hebrew and Burji (a Cushitic
language) below:

 H B
1 exad d'ekki, micca
2 shtayim lama
3 shalosh fadiya
4 arba foola
5 xamesh umutta

(c) The latest work on Ainu (Alexander Vovin, "The origins of the
Ainu language", Panasiatic Linguistics, II: 672-685, Bangkok,
and his just published book "A Reconstruction of Proto-Ainu")
conclusively, to my mind, refute the idea that AInu might be
Altaic and suggest a relationship with Austroasiatic that is
at least suggestive.

(d) About Jao(I mean, Japanese) and Altaic, the latest work is
Starostin's book Altajskaja problema i proisxozhdenie
japonskogo jazyka, Moscow 1991, and Vovin's survey
article in a recent issue of Diachronica (I am sorry
I don't have reference handy).
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue