Wald on Linguistic classification

[bwald]

----------------------------Original message----------------------------
I read Alexis's respose to me with interest (naturally).  I have to admit
that I found his response to
Tom Cravens at about the same time even more interesting, but I'd like to
clarify the difference between my position and the one which AMR uses my
message to attack.  Quoting from my message, he writes:
 
>Friend Wald writes (i.a.):
 
(uh oh, I thought.  I'm not a Quaker.  How come he's not calling me Benji,
like everybody else does?  As a linguist I am interested in such rhetorical
ploys.  My heart sank when in his message re: Cravens he referred to
"friend Nichols", since her relayed information about the status of Altaic
antagonized him.  So what did *I* do?   AMR quoting me continues...)
 
>Are there any generalizations about where the attempts to unite groups of
>languages tend to become problematic?  What are the arguments we can
>dismiss (typological ones? areal ones?)  Which ones are generally
>bothersome because we don't know quite how to deal with them?
>
>Can we get further than:
>
>NOT ENOUGH DATA  (because of time-depth; that's the biggie, isn't it?)
>CAN'T TELL BORROWINGS FROM INHERITANCES
>BASIC VOCABULARY
>CHANCE RESEMBLANCES
>MULTIPLE SOURCES
>(end of quote)
 
To which Scott DeLancey added: "not enough specialists to deal with all the
relevant data".
So now to get to AMR's point:
 
>I could not disagree more.  The "biggie" involves one of the
>biggest myths in linguistics.  ....this whole sordid mess, for now I will
>ask those of
>you who trust my competence in the relevant areas (i.e.,
>baby mathematics and history of linguistics) that (a) the
>vast majority of publications claiming some such limit on
>time depth contain not even a shred of the required documentation
>that such limit exists or that it is what they claim it is,
>(b) that the works that tried to do the math required to
>demonstrate such a limit (the earliest of which that I know of
>was believe it or not by Swadesh!!!!!!!!!!!!!!!!) are mathematically
>invalid, and (c) that the only competent work on the subject that
>I have been able to find is, I am afraid, the unpublished work
>by my coauthors and me, so that if there is a limit, you will
>have to either find it or ask us (:-).
 
I don't mind Alexis using my
 
>NOT ENOUGH DATA  (because of time-depth; that's the biggie, isn't it?)
 
to criticize speculative theories which try to "prove" that relationships
beyond a certain time-depth will remain undemonstrable.  I don't think such
theories are any more interesting or sound than glottochronological
theories.  So, for one thing, I want to make it clear that I am not
endorsing such theories, and did not even have them in mind  (so, maybe I
can be "Benji" again, instead of "Friend").  Instead I had in mind the use
of "time-depth" and "not-enough data (left)" as excuses for doing
inconclusive work in reconstruction and demonstrating relationship.  It's
not about some abstract theory.  It's about:  how come everybody accepts
Indo-European but not Altaic?  Is it the nature of the data or what?  Is
there enough data to do an "Indo-European" on Altaic?  Does anyone say no,
and try to explain why on the basis of time-depth? Should there be any
problem in doing this, or is it "reasonable" to consider such an argument
for establishing a relationship on evidence which does not attain the
standards set for Indo-European  (and met for some other language
families)?  Now this question is addressed to AMR as well as anybody else.
 
 
My position is  -- and this goes for Indo-European too -- that we want to
*raise* the standards for demonstrating "genetic relationship", NOT *lower*
them, so that we can better understand the real implications of "genetic
relationship" and the principles of linguistic change (of which "genetic
relationship" and its implied "internal linguistic change" is but a part).
 
Alexis continues:
 
As to the other points,
>I will just say briefly that the real issue is not whether we
>can tell borrowings from cognates etc. but whether IN EACH
>INDIVDUAL CASE you accept the proposals of X, Y, or Z linguist.
 
No.  I don't agree.  This does not go far enough, and it dismisses the
legitimate, even inevitable, issue of borrowings vs. cognates.  There is no
disagreement that both such things exist.  On what basis should I accept
the proposals of linguist X, Y or Z.  Is the implication that that's why we
universally accept Indo-European but not Altaic?  Because we're victims of
some authority-complex?  I wouldn't agree that this is why there's a
difference in acceptance of the two theories.  (Though I would agree that
it's why most of us accept that the earth goes around the sun, and not
vice-versa, among other irrelevancies.)
 
AMR goes on:
 
>And since the same person may use different methods at different
>times or in different cases, you cannot even generalize that you
>will or will not accept anything proposed by, say, Sapir or Greenberg
>or Kroeber or Swadesh or Hamp.
 
Right.  So what I want to know is what are these METHODS that one *will*
accept, and what or where do they have limitations?  Enough with the
innuenedos about personalities and politics.  We are fortunate to have a
pretty democratic list here, where AMR's views are certainly no less
privileged in getting through to the readership than anybody else's.  Let's
stick with that and make some progress.
 
AMR goes on at length to give numerous examples where the same linguist has
made some claim that is generally accepted and some other claim that is
generally rejected, relevant to classification. But there was nothing in
that list of examples about METHODS, or what the problems are.  AMR's
message only touches on what the (methodological) problems are NOT.  I
asked"  "what ARE the problems?"  I still want to know.  I repeat MY
question, quoted by AMR above:
 
>Are there any generalizations about where the attempts to unite groups of
>languages tend to become problematic?
 
(By the way, What's the issue AMR alluded to with Armenian?  Or was that a
joke I didn't get?)
 
As long as I'm at it, I'd like to ask some questions about AMR's message in
response to Tom Cravens.  AMR writes:
 
>So, I still say that when a relationship is proposed,
the burden of proof is on the proponents, the opponents
need do no more than show that teh burden has not been
>met (most obviously by showing that the work is incompetent).
 
I'm not sure where the line is between "obviously" incompetent work and
unsound (or "speculative") scholarship.  For example, since I know Bantu I
can recognize when somebody incompetently make a wrong morpheme cut to
offer a cognate with some other language group, but I don't even have to
know anything about Basque and Caucasian to stop reading a book that starts
off trying to relate the two by arguing that since "man" has free-will, any
consonant in Basque can unconditionally correspond to any consonant in
"Caucasian"  (I'm thinking of an actual case, though I'm not sure the
languages were Basque and "Caucasian" -- nor does it matter.)  Obviously,
what AMR says here becomes more interesting where things are NOT "most
obvious(ly)".
 
AMR continues:
 
On[c]e however a case is presented which stands up well to
whatever criticisms have been offered, then it obviously
becomes much more difficult, though not impossible, to
reject the putative relationship
 
What does "stands up well" mean?  Like Indo-European?  Not THAT well?  HOW
well?  What are the standards for "stands up well".
 
--and the more strongly
that it is argued and the more widely that is accepted
by competent scholars, the more work will be required to
refute it.
 
I guess "strongly argued" and "accepted by *competent* scholars" inevitably
coincide.  Nevetheless, I remain more interested in the crtieria for
"strongly argued".
 
Finally, AMR gets to his main point, which is to shift the burden of proof
from one side to the other:
 
Hence, in the case of Basque-anything I would
say the burden is entirely on the proponents of such
relationships, in the case of Uto-Aztecan or Afroasiatic,
 
[Benji, HERE IT COMES]
 
just as obviously it is on the opponents (there are none
in the case of UA that I know of but believe it or not
there seem to be some in teh case of AA).
 
Why not use Indo-European as the example of a non-controversial family?  Is
any reader unfamiliar with the basis for it?  The same is not true of any
other language family, regardless of the fact that historical accident of
the development of the discipline is involved in this state of affairs.
 
In the case of
Altaic (like Sino-Tibetan or perhaps Penutian), it seems
that the situation is a little less clear, but the burden
is mostly on the opponents.
 
I don't understand the implications of "mostly".  The argument is
unsatisfactory with respect to clarity.  What about "leastly"?  What
responsibility remains to the proponents of Altaic (cf. below)?
 
AND it is crucial to note that
since 1956 or so teh opponents of ALtaic have accepted this
disadvantageous position and done so very loudly and emphatically.
 
No.  I don't agree that that's crucial.  That does not concern me AT ALL in
trying to understanding the issues involved in what the problems ARE with
Altaic.  NB, even Indo-European has problems, how many "laryngeals" -- in
which words -- etc etc.