6.43 Greenberg: techniques and counterevidence

[The Linguist List]

----------------------------------------------------------------------
LINGUIST List:  Vol-6-43. Mon 16 Jan 1995. ISSN: 1068-4875. Lines: 100
 
Subject: 6.43 Greenberg:  techniques and counterevidence
 
Moderators: Anthony Rodrigues Aristar: Texas A&M U. <aristar at tam2000.tamu.edu>
            Helen Dry: Eastern Michigan U. <hdry at emunix.emich.edu>
 
Asst. Editors: Ron Reck <rreck at emunix.emich.edu>
               Ann Dizdar <dizdar at tam2000.tamu.edu>
               Ljuba Veselinova <lveselin at emunix.emich.edu>
               Liz Bodenmiller <eboden at emunix.emich.edu>
 
-------------------------Directory-------------------------------------
 
1)
Date: 11 Jan 95 06:00 GMT
From: ECOLING at AppleLink.Apple.COM (Ecological Linguistics,Anderson,PAS)
Subject: Evaluate techniques not person
 
2)
Date: Wed, 11 Jan 95 15:34:33 EST
From: amr at kali.cs.wayne.edu
Subject: Greenberg and Counterevidence
 
-------------------------Messages--------------------------------------
1)
Date: 11 Jan 95 06:00 GMT
From: ECOLING at AppleLink.Apple.COM (Ecological Linguistics,Anderson,PAS)
Subject: Evaluate techniques not person
 
Mike Maxwell's message on Greenberg's hypotheses and possible parallels with
"Eve" and mitochondrial DNA is very interesting.
 
I would suggest that a PhD dissertation would be much better oriented to
finding out exactly how some of our various techniques of analysis fare at
different time depths and on typologically different sets of languages.  In
other words, trying to prove a particular *person* right or wrong is much less
useful to the field than trying to advance the techniques of the field, and the
more useful work will be more cited and will itself have a longer useful life.
 
On the particulars, I should certainly read "The Search for Eve" since I am so
interested in how people reason and the traps they can fall into.
 
-------------------------------------------------------------------------
2)
Date: Wed, 11 Jan 95 15:34:33 EST
From: amr at kali.cs.wayne.edu
Subject: Greenberg and Counterevidence
 
In response to Pullum and Maxwell's postings on the question
of what would constitute counterevidence to Greenberg's
proposal that the languages of the Americasfall into
three groups (Eskimo-Aleutian, Na-Dene, and Amerind),I
wanted to suggest a few things:
 
First, I think it may well be the case (although I know
of no unambiguous evidence for this) that Greenberg would
basically ask for a superior classification (not a merely
negative result).  If I am right in this, then Greenberg
would be taking the same position here as some theoretical
linguists do who laugh at counterexamples and insist on
a better analysis instead (a position I should say I do not
accept).
 
Second, a concrete way to do something like this would be to
show that, by Greenberg's own criteria, some language(s) he
classified as Amerind really fits in better with some family
outside of Amerind (Na-Dene or Eskimo-Aleutian or something
else).  As far as I know, the only attempt to do this occurs
in my own paper which will come out some time in IJAL.
 
Three, I think that the suggestion that has been floating around
that Greenberg does not care if his data are incorrect, however,
is not fair.  I believe that if you could show that there are
NO data for some classification, he would accept that (as I hope
any theoretical linguist would in the case where ALL the data
supporting an analysis turn out to be mistakes).  However, what
I think he maintains is that the amount of error in the data
he has used is not sufficient to overthrow the proposed
classification (or the subclassification of Amerind into
subfamilies).  Note that percentages are not relevant here:
it could well be that 90% of the data on some classification
were wrong, but the remaining 10% could still be enough.  The
point bein, of course, that only the correct data count.
The fact that somebody's examples are, hypothetically, 90%
wrong does not have any bearing on the validity of the theory,
only on such things as the person's credibility or whatever.
In the cases I have looked at, Tonkawa and Zuni, the percentage
of errors in the data seems to be around 30% at most, by the way.
 
So, it seems to me that Greenberg's classification is capable
of refutation even by his own criteria (which seem to be those
of many theoretical linguists, even though they must seem shocking
to most historical linguists, for example), not to mention by
other, more stringent, criteria that could be proposed.
 
Alexis Manaster Ramer
 
--------------------------------------------------------------------------
LINGUIST List: Vol-6-43.