Language and Anthropology in the Americas

[manaster at umich.edu]

----------------------------Original message----------------------------
 
 
On Sun, 10 May 1998, Miguel Carrasquer Vidal wrote:
 
> The real reason that I reject Amerind is not its lack of sound
> correspondences.  John Bengtson has published sound-correspondences
> between North Caucasian and Basque, and I'm afraid I must reject that
> too.  The real reason is that I have read the evidence as presented,
> and I didn't like it.  I've read the evidence presented by
> Illich-Svitych for Nostratic, and I liked it.  I have not seen the
> evidence for Altaic, so I don't know if I like it.
It is hard to reject Altaic and accept Nostratic, actually
I think it is impossible, except in the technical sense
that you could reject Altaic as a valid node within Nostratic.
>I can explain
> what I like about Nostratic, but this is neither the time nor the
> place.
Actually, I think it would be interesting to have you expand on
this.  I myself have massive doubts about Nostratic but continue
to be sympathetic.
>It is much harder to explain the opposite reaction (apart
> from errors of fact etc.).  It is largely gut, I'm afraid.  That's
> why it's so difficult to have a fruitful discussion about these
> matters.  But we can try...
I dont see why it is hard to say what one does not like about
a proposed theory.  My own reaction to Greenberg is as follows:
 
(a) I like to break a problem up.
(b) Start with Nadene-Haida, which is a manageable chunk.
Greenberg responds in some detail to the critiques of this
connection and he is mostly right but he makes it seem as
though we did not need do anything beyond what Sapir
himself considered merely a "provisional" argument and
indeed has nothing whatever to add.  That is very disturbing
to me (and in my paper on the subject I did try to go beyond
Sapir and I also sketched a research program for doing even more
which I would like one day to undertake).  For one thing, much
of the data Sapir relied on was unreliable for Tlingit and
Haida both, and I at least tried to see what happens if
you look at corrected data I got from the best experts
on these two languages.  But Greenberg did nothing, so that
bothered me.
(c) I then reasoned that as far as Amerind is concerned,
Greenberg's mass comparison can at best mean that most
of these languages are related; it simply cannot mean
that every last one is.  This is actually a simple logical
consequence of using n-ary comparison (comparing n languages
all together), btu I could never get Greenberg to see this,
so THAT bothered me.
(d) Furher, I wanted to break Amerind down into some
manageable pieces too.
(e) One thing I thought was to find test cases, languages
which either have a long-standing history of being difficult
to classify or occur along the Nadene-Amerind frontier or
both. Tonkawa and Zuni qualified immediately. And so I
looked at them and realized that Tonkawa looks more Nadene
than Amerind using Greenberg's methods, and ZUni looks
no more Amerind than it does IE.  Greenberg;s response
to my initial publication on this did not still my
doubts, to say the least, but I am waiting for a response
to my much more detailed argument in IJAL.
(f) I then realized that the other and probably more
significant thing to do is forget about Amerind and
check each of Greenberg's proposed daughters of Amerind,
but I got sick before I could do this. However,
Pam Munro has done one such study and her results
seem to me to very interesting. Indeed, it is the
single most important study of the topic.
(g) I was puzzled by the fact that Greenberg
did not comment on the fact that in earlier
work he hd excluded Oto-Manguean from Amerind,
which would make this a particular interesting
case.
(h) I was like so many others disturbed to some
extent by Greenberg not using teh best or the
most recent data, although I am not convinced that
this is a crucial problem.
(i) I thought Greenberg's teratement of morphology
was suggestive but a bit cavalier.
 
On the other hand, I found almost all critiques
of Greenberg to be either methodological grandstanding
by people who have not done enough work on language
classification or studied the history of the field
enough and who were inventing a mythical "Comparative
Method" which never existed.  It also struck me
that the spate of publications correcting minor
or not so minor factual errors was not very
useful because it did not address the question of
whether the errrors were numerous enough to matter.
I myself did tackle this qustion for Zuni and
Tonkawa and found that the percentahe of errors
was not significant.  I was also struck by how
Greenberg's critics were no better than him at
picking selected test cases and seeing what they
really tell us.  The very fact that no critic
to my knowledge looked at the Tonkawa or Zuni
orOto-manguean question or at such proposed
daughters of Amerind as Central Amerind suggested
(and I said this in IJAL) that they were interested
merely in scoring points and not in classifying the
languages of the Americas.
 
As for Nostratic, it is true that I-S proposed
a mess of sound laws, but there are so many
problems that I do not see it as clearly
more convincing than Amerind.  For ex.,
it is striking that all teh kinship terms
posited for Nostratic are those for in-laws,which
would more natural to find in the case of borrowings
than inherited vocabulary, on the assumption of
exogamy. The sound laws are shaky, and there is
all maner of inconsistency.  For ex., the IE
word for 'milk' has the wrong velar but I-S
did not care.  In reality, as I have argued,
it is much more likely that it was borrowed
from Afro-Asiatic than that it is a cognate of
the AA and Uralic forms.  The sound laws, in
addition to being ignored as in the case just
mentioned, are quite shaky. Serebrennikov
was right to say that I-S took the Altaic
vowel system and the Kartvelian consoannts
and put them together to give us Nostratic
phonology.  And so on.
 
Of course, as I have written in several
places (most recently in JIES), I-S's
critics have been no better than Greenberg's,
except for Brent Vine, but he only deals
with IE and so does not really address
the real issues.
 
So as for me I am stuck in teh uncomfortable
position of not having any way to decide
about either Amerind or Nostratic, simply
because the relevant work has not been
done, except for little bits here and there,
and again Pam Munro's work on a part of
Amerind comes to mind. There is no
comparable work on ANY part of Nostratic,
and to that extent at least one migth say
that Amerinda is BETTER off than Nostratic,
since a significant part of it has been
tested and approved by a major indepenednt
scholar.  I dont think my work on Nostratic
or anybody else's even comes close.
 
The true-believers in each case totally
ignore all the problems and proceed as though
both theories were proven.  The critics are
acting as though they were proven wrong (which
is logically impossible), and as though there
were no point trying to classify the languages
in question AT ALL.  And indeed with people
like Nichols and Ringe getting loud applause
for their claims that no classifications
beyond those which we now know are ever going to
be possible, it is not hard to understand why
that should be.
 
Alexis MR