The Significance of Comecrudan

[bwald]

----------------------------Original message----------------------------
Only because of recent exchanges, I feel the need to disclaim that I have
appointed myself AMR's nemesis in commenting on an argument he made about
language-relatedness.  In fact, my comments repeat things I have already
said on this list in responding to other scholars.

The point that interests me starts where AMR wrote:
"....  If there is to be a discussion of
methodologies, then we should look at language families that have proposed
on the basis of specific methodologies and see what we find, "

That's an excellent idea, but the sentence continues:

"not start out
by a priori accepting some completely arbitrary methodological assertion
made up from whole cloth and then rejecting any language family whose
recognition would force us to abandon that assertion".

 The part about "some completely arbitrary methodological assertion" has to
be defended,
since there is a traditional consensus view that has been accepted over
time, and, because of the tests it has passed in its development and its
successes, does not appear to the field as a whole to be arbitrary.  To be
sure, not all languages are thought to have single ancestors, but the
notion of "language family" does seem to rest on the notion of the related
languages having a single common ancestor, and a demonstration of
reconstruction of that ancestor and diversification of its descendents is
the consensus that banishes skeptics to the sidelines.

The continuation is:
"...I am referring of
course to the two assertions:(I) Language relatedness can only be shown by
reference to morphology  (falsified by the history of how Tai, Comecrudan,
and (an example I forgot to cite earlier) how Uto-Aztecan was discovered),

(II) Language relatedness can only be shown by establishing a system of
sound laws (falsified by the history of how Niger-Kordofanian,
Indo-European, Afro-Asiatic, and indeed probably most of the currently
accepted language families were established)."

The two assertions are opposed to each other, when I think they should be
presented as
cooperative properties of the consensus view.  Of course, morphology is
vulnerable to
phonological erosion and syntactic reorganisation, but demonstration of
relationship
should eventually be able to account for the grammatical structure of the
ancestor -- otherwise, what's the point of assuming a single gramatical
ancestor whose properties are unknown?.  Apart
from that, the "falsification" referred to has to do with the first HUNCHES
that these were
families -- of the same kind as those  families that no one would dispute,
-- not how SUBSEQUENTLY they came to be "established" as families.

In this context, I want to comment on AMR's claim that Niger-Kordofanian
(and probably various other families) falsifies the traditional claim of
how language-relatedness is ESTABLISHED.

These statements remind me of what I wrote in reference to Merritt
Ruhlen's argument that language-relatedness is "established" prior to such
things as the above; he was pushing mass comparison as sufficient to
"establish" relationship.  He argued speciously that if relatedness was not
already being "accepted", what would be the point of further mining the
data for sound correspondences, etc.?  I think we all agree that the way we
understand (genetic) relatedness, if languages are indeed related then we
expect systems of sound laws, and diversification of syntax and morphology
to eventually be demonstrable.  In view of that, I objected to Merritt's
obscuring the "acceptance" (if you will) of a HYPOTHESIS (that mass
comparisons, including some grammatical morphemes, suggest that a set of
languages are related) and a METHOD OF DEMONSTRATION to TEST the hypothesis
(which involves sound correspondences and whatever else).  My scare-quotes
above call attention to tricky words like "establish" and "accept".

Further repeating comments I think I have already made on this list, let's
take the case of Niger-Kordofanian (which generally reverted to the earlier
label Niger-Congo, NC, in a manner similar to the way "Indo-Hittite"
reverted to "Indo-European".)

It is absolutely true that concerned scholars "accept" the notion of an NC
family as the most reasonable HYPOTHESIS (yet offered, and over the dead
bodies of previous hypotheses) for the massive similarities among various
NC languages.  However, in the absence of DEMONSTRATION by classical
methods, there is disagreement about membership in that family -- and the
skepticism is JUSTIFIED until such demonstration is forthcoming.  The
greatest victory in the establishment of the NC family was the "acceptance"
that Bantu is a sub-sub...branch of NC, i.e., that it is related to
numerous West African languages.  Demonstration has proceeded quite easily
with respect to Bantu and its nearest relatives in Cameroon and Nigeria,
and by linkage, between various Nigerian languages and many (but not all)
more eastern and northern NC languages.

The situation is not so clear for some hypothesised branches of NC.
Mukarovsky in Vienna has doubted the Northern Atlantic (?sub-)branch and
the Mande branch.  While I, for one, find it quite likely that both these
groups will turn out to be bona fide NC ( most NC-ists seem to agree), I
look on with great interest to the demonstration that will relieve the
doubts of critics like Mukarovsky.  I am particularly interested because
NC-ists tend to consider Mande to be among one of the earliest separations
from NC as a whole, and  if demonstration of relatedness is successful, I
want to see what it indicates about the nature of NC at that stage in its
history.  Indeed I am more interested in that than the "fact" of
relatedness itself.  (Actually, I suspect that the commonalities of the
less problematic NC branches might already tell us more about early
versions of NC, and that Mande has diverged from those commonalities more
than most languages -- but that's just a suspicion, not even a hypothesis,
and still of great interest to me: how and why has Mande diverged so much?)

I think I have already mentioned in earlier discussions that Kordofanian
also has problematic aspects, and that is involved in general reversion to
the label "Niger-Congo".  It was originally assumed that the Kordofanian
languages were all related.  On the basis of the evidence in SOME of these
languages, it was hypothesised that it was also related to NC, as a sister.
Some of the Kordofanian languages, however, showed more striking apparent
resemblance to Nilo-Saharan than to Niger-Congo.  On the basis of mass
comparisons it was hypothesised that Kordofanian was a link between
Niger-Congo and Nilo-Saharan.  Gregersen, in particular, published a paper
with this proposal (he was a student and admirer of Greenberg, a teacher of
mine at Columbiaand highly knwoledgeable, as he would have to be, to argue
for such a proposal).  On closer inspection, however, esp. by Schadeberg,
it turned out that this proposal was not sustainable, and that the
similarities among the Kordofanian languages were the result of
convergence.  Some of them (the ones mainly used as the NC-NS link) turned
out to be more likely to be Nilo-Saharan languages, while others still
seemed  most likely to be Niger-Congo.  Hence the backoff on
"Niger-Kordofanian", and reversion to "Niger-Congo".  "Mainstream" opinion
is still hesitant about whether Mande or "Kordofanian" split off first.
All of these things are HYPOTHESES guiding and propelling further research.
Terms like "accepted" (NC is an "accepted" family) must be understood in
this context.  Terms like "established" (NC is an "established" family) is
open to misinterpretation, because of the blurring of "hypothesis" and
"demonstration".

AMR immediately continues:
"But even more important is the fact that, as I maintain, it is NOT
reasonable to take a quick look at some proposed linguistic family and
immediately take either a negative or a positive stand and then hold on to
it for dear life.  Rather, a new theory, if correct, will over time
undergo considerable refinement and, crucially, find more and more
evidence to support it--and will explain more and more data."

There is nothing I disagree with there.  I do not object to AMR arguing
that there may be sufficient reason to further consider the possibility
that certain languages are related, for which the standard statement is
"further research is needed".   I only object to blurring the lines between
a hypothesis (which may be "accepted" due to "opinion-makers" -- and turn
out to be wrong) and a methodically acceptable demonstration (which
"establishes" the relatedness beyond reasonable skepticism/criticism).  On
the basis of earlier friendly conversations with AMR I do not think he will
disagree.  I think I am doing more than mincing words to bring to attention
the need for an argument in defense of more speculative relationships to be
more carefully and precisely laid out, esp what makes a speculative
relationship more or less promising (for eventual demonstration).   If that
is all that is being proposed -- great!  Leave out the part about the
consensus view being "arbitrary" or "made up from whole cloth", or defend
it.                       --  Benji