LINGUIST List 5.908
Fri 19 Aug 1994
Editor for this issue: <>
"Reinhard, 5.905 Altaic
Harold Schiffman, Re: 5.905 Altaic
Ecological Linguistics,Anderson,PRT, Altaic
, Altaic and Penutian
Message 1: 5.905 Altaic
Date: Thu, 18 Aug 1994 10:38:41 5.905 Altaic
From: "Reinhard <rhahnu.washington.edu>
Subject: 5.905 Altaic
Thanks to Alexis Manaster Ramer for initiating some discussion about the
Altaic hypothesis (AH) and about the controversy surrounding it, as well
as for the summary of the remaining problems. (Original message above.) I
wholeheartedly agree with the view that the AH offers an excellent
opportunity to discuss universally applicable methodology in genealogical
I would like to add a number of notes.
(1) For the benefit of the uninitiated the following introduction:
(1.1) The AH holds that there is an Altaic language family which
consists minimally of the following branches: Turkic, Mongolic,
Tungusic. These languages are of the SOV type, are agglutinative
and share morphophonological harmony.
(1.1.1) I suggest that we use the generic term "Turkic" in place
of "Turkish" to distinguish it from the specific
language, and in place of "Chuvash-Turkic" or "Bolghar-
Turkic", though there may be merit to the argument that
Bolghar (Old Chuvash) represents a sub-branch.
(1.1.2) I suggest that as a generic label "Mongolic" be used in
place of "Mongolian" found in traditional-style Altaist
and Mongolist literature. This allows us to reserve
"Mongolian" for the specific language.
(1.1.3) I suggest that "Tungusic" be used in place of traditional
"Manchu-Tungus(ic)". There seems to be no justification
for adhering to the composite term.
(1.2) Some have argued that there is a Ural-Altaic super-family. While
the Uralic languages (Finnic, Ugric, Samoyedic) share important
typological (particularly morphophonological) features with
Altaic, the Ural-Altaic hypothesis is currently considered pass'e.
(1.3) Some Altaists have proposed Korean-Altaic affinity. Others,
including solidly mainstream ones like Nicholas (Nikolaj
Nikolaevi^c) Poppe, have supported or at least not
rejected this proposal. Poppe's assumption that such a
Korean-Altaic relationship goes back to a pre-Altaic stage
ought to be noted.
(1.4) Some, notably Roy Andrew Miller, have argued for Japanese-
Altaic affinity. If this affinity exists, it ought to be
sought at an even earlier stage than the possible Korean-Altaic
(1.5) Notable among the less accepted hypotheses is that of Dravidian-
(2) The main arguments used by the critics of the AH (in recent years
particularly Gerhard Doerfer and a number of European Turcologists and
Mongolists) to discredit the hypothesis are mostly based upon criteria
used in Indo-European research, criteria they seem to assume to be
universally applicable. Evidence for this is found in two of the main
weaknesses of the AH as alleged by the anti-Altaists:
(2.1) Absence or scarcity of cognates among numeral.
This poses questions regarding the universal validity of
numerals as mandatory indicators of genealogical affinity. The
fact that numerals seem to supply evidence for the
Indo-European hypothesis ought not lead us to assume that this
applies universally. In this regard it would be interesting to
hear from those who specialize in numeral systems development
as well as from those who deal with genealogical groups in
which numeral systems are relatively less developed (e.g., among
(2.2) Scarcity of cognates among terms for parts of the body.
This appears to be a weak argument. Quite a large number of
cognates have been identified among Altaic terms for parts of
the body. Many of these have undergone considerable semantic
shifts among the main branches, which is why they were not
immediately apparent to earlier investigators.
(2.3) Turkic-Mongolic and Mongolic-Tungusic "pseudo-cognates" but no
direct Turkic-Tungusic ones.
This has been taken as one of the main indicators that we are
dealing with lexical borrowing rather than with genealogical
affinity. This is mostly based on geographical distribution and
thus on the assumption that there used to be no direct
An impressive number of direct Turkic-Tungusic cognates have been
identified especially in recent years. Regrettably, most of these
have only been presented to small audiences or to
institute-internal readers so far. Extensive distribution of
this information is urgently needed. In the light of such
research data, maintaining the lexical borrowing argument (e.g.,
by arguing that in all such cases the corresponding words
were lost in all Mongolic languages) would be going way over the
(3) The Turkic, Mongolic and Tungusic pronominal systems bear astonishing
resemblance, which enables us to reconstruct a proto-Altaic system.
This has been dismissed as irrelevant, since this proto-Altaic system
bears great resemblance not only to the proto-Uralic one but also to
the Indo-European one. (Obviously, the anti-Altaists are among those
least likely to accept the possibility of even earlier genealogical
affinity, "Nostratic" or otherwise.)
(4) Morphological correspondence among the Altaic languages is not only
structural. A number of researchers have been engaged in identifying
cognates among Turkic, Mongolic and Tungusic (as well as Korean and
Japanese) suffixes and enclitics. It would be interesting to hear from
(5) Altaic research is in urgent need of the following:
(5.1) Identification of further lexical cognates, with emphasis upon
direct Turkic-Tungusic cognates.
(5.2) Systematic reconstruction of proto-Turkic, proto-Mongolic and
proto-Tungusic suffixes and enclitics, and ultimately
reconstruction of an Altaic morphology.
(5.3) Creation of opportunities to conduct Inner/Central Asian
research as well as comparative Turkic, Mongolic and Tungusic
research with provisions for formal Altaic studies options.
Renewed access to Inner Asia, dealings with newly founded
Central Asian countries and the availability of relevant
linguistic data (including access to large numbers of native
speakers) are generating increased interest in Turkic, Mongolic
and Tungusic studies, but and governmental and institutional
response to the growing demand is very slow in coming.
Reinhard F. Hahn
Message 2: Re: 5.905 Altaic
Date: Thu, 18 Aug 1994 10:39:59 Re: 5.905 Altaic
From: Harold Schiffman <haroldfsu.washington.edu>
Subject: Re: 5.905 Altaic
Although I am not an Altaicist, I am in a department where some of the
primary proponents of the Altaic hypothesis, including the
Altaic-includes-Korean-and-Japanese version, were once members. I have
been interested also because some mega-versions of Altaic would include
Dravidian, which is typologically quite similar, though few phonological
correspondences between Proto-Dr. and Altaic are evident. This has not
stopped some people in Japan, esp. Susumo Ohno, emphatically NOT an
Altaicist, from claiming that Japanese and Dravidian, esp. Tamil, are
related. Ohno simply bypasses Altaic and compares Dravidian with
My own reading of the controversy impels me to the conclusion that it is
primarily political, i.e. people accept the notion the Altaic includes
Japanese NOT on linguistic grounds (phonolog. corresp. etc.) but on the
grounds of what this would entail socio-culturally: that the Japanese
are/would be related to Koreans and other despised primitive peoples
(hunters and gatherers, mares'-milk drinkers) of northeast Asia. This
idea seems anathema to many Japanese, so they search elsewhere for
relationships of a more "noble" sort, i.e. with "noble" civilizations such
as the Dravidians, or Polynesians, or whatever. I have corresponded with
Murayama, a war-time student of Poppe, and the only linguist in Japan who
seemed to be strongly in the Altaic camp; his arguments made sense to me,
at least. And Poppe himself told me once that he thought of the Dravidian
relationship to Altaic as an "Urverwandtschaft" that might be established
after proto-Altaic was more firmly accepted.
I also believe that Poppe was not firmly convinced that Japanese belonged
within Altaic, despite the fact that he wrote a foreword to Roy A.
Miller's book "Japanese and the other Altaic languages".
I'd be interested in hearing more technical arguments about real data
such as phonological correspondences etc.
Message 3: Altaic
Date: 18 Aug 94 18:56 GMT
From: Ecological Linguistics,Anderson,PRT <ECOLINGapplelink.apple.com>
We should be grateful to Alexis for giving us his analysis of where studies of
the Altaic hypothesis now are. I got frustrated years ago by the
non-communication and the many discussions in the absence of sufficient
evidence or neglecting what had been presented (both ways).
Just on one margin, readers might like to know that at least a few years ago,
the closest relation of Korean to Tungus had been specifically claimed with the
Manchu branch of Tungus, and presumably not from borrowings.
I reworked the material in a paper by the dean of Korean linguists on this
connection, and found the sound correspondences even closer than he claimed.
Where there were dual correspondences for vowels, they sorted themselves out
fairly neatly depending on the grammatical part of speech of the cognate set,
in ways which recognizably can reflect typologically plausible segmental
reflexes of recessive vs. non-initial stress positions, in more nominal vs.
more verbal forms respectively.
Since more detailed analysis made the set of correspondences look better rather
than less good, I assumed the hypothesis was on the right track. I look
forward to getting back into Altaic in the next year or two, catching up on the
more recent literature, publishing some unpublished papers, etc.
Message 4: Altaic and Penutian
Date: Thu, 18 Aug 1994 19:52:00 Altaic and Penutian
Subject: Altaic and Penutian
I was very interested in Alexis Manaster Ramer's summary of the state
of opinion regarding Altaic, since much the same summary could be made,
_mutatis mutandis_, of the state of Sapir's Penutian hypothesis.
Fortunately in Penutian we have -- as of this summer -- gotten past
the sterile yea/nay debate. 20 linguists who work on Penutian languages
from Tsimshian to Yokuts met for a 2-week workshop on Comparative Penutian
Linguistics at the University of Oregon (a meeting organized by Scott
DeLancey and myself). We came to the meeting with a variety of attitudes
towards the validity of the Penutian hypothesis, but by the time the
workshop was over we were in unanimous agreement on several important
First and foremost we agreed that the Penutian relationship is real one
historically, not just the mechanical epiphenomenon of someone's
classificatory "method". There was a strong feeling, however, that we
need to fashion new tools to deal with connections this old and involving
such a diversity of structures. There clearly ARE historical inter-
connections among these languages--it's not just a matter of similar
typologies, or chance--but the Stammbaum model is inadequate for the
task. However, we also agreed that lexical comparisons, to be probative
and historically interesting, needed to be made in the context of a
hypothesis of a proto Penutian morphosyntax, and it became clear that
future research is going to have to be strongly reconstructive in nature,
however those reconstructions are arrived at or supported. We spent a
lot of time, for instance, talking about how we might reconstruct
the proto Penutian case system, or how the templatic morphology of Yokuts,
Miwok-Costanoan, and (perhaps) Takelma might be projected back to a
The "vitriol" of recent discussions of deep relationship has largely
focused on the acceptance or rejection of shortcuts to real historical
understanding. Whether or not Greenberg's "multilateral comparison" or
any other specific method usefully functions to generate testable
historical hypotheses, the fact is that such hypotheses are already thick
on the ground, while the real work of deep historical linguistics largely
remains to be done. Altaic, like Penutian, is clearly a useful arena
for historical investigation. Let's get on with it!
Humboldt State University
Arcata, CA 95521