Multilateral Comparison mis-evaluated

[ECOLING at aol.com]

Larry Trask writes, in a message concerning retained info
vs. spurious proposed cognates in Multilateral Comparison:

>Anyway, the point is *not* to avoid missing any valid cognates -- a
>pointless and futile enterprise, in my view.  The point is to find
>sufficient positive evidence for relatedness, over and above chance
>resemblances, that the null hypothesis of unrelatedness cannot be
>maintained.

That may be Trask's claim about "the point" to him!

Actually, the legitimate point of Multilateral Comparison is
NEITHER OF THE ABOVE.
It's legitimate use is to be a heuristic for selecting which sets of languages
are MORE LIKELY to be fields for fruitful study of
deep relationships than others.   Or put another way,
starting from any particular language or known family,
what are the MORE LIKELY links of that language or
family to other languages or families.  Not proven, just
more likely as compared to less likely.

Trask agreed with the possible value of Multilateral Comparison as
a heuristic; so it is simply incomprehensible to me that he
then insists on ONLY evaluating it as some kind of proof of
relatedness passing some threshhold of strength of proof.
THAT task is for methods other than Multilateral Comparison,
in Trask's view.  Specifically the Comparative Method,
Internal Reconstruction, and others.

Trask has his own preferences,
but others need not choose the same preferences.
It is not immoral to want to work on pioneering edges,
recognizing that one can be less certain of one's results,
whether overviews or details,
that one will make perhaps more errors,
and surely a different kind of errors
(since those who work within more limited domains may be working
under a framework with unquestioned false assumptions
which make many of their results wrong,
at least in some detail, or worse).

Trask's statement quoted above is thus not what it claims to be
or appears to be.  It is rather a circular expression of the kind
of work Trask prefers to do, but expressed as if other choices
are not legitimate, other choices which are KNOWN to be necessary
for the progress of science.  Heuristics are simply unavoidable.

Trask, like Greenberg, misstates what his logic can legitimately
lead him to.
Given what Trask has said elsewhere, his statement quoted
above would follow IF one took as an assumption that only proof
and not heuristics could be "the point".  Each of those can be
"the point" in some contexts, there is no justification for trying to
throw out either of them.

Therefore, the application to Multilateral Comparison of a "test"
(criterion of sufficiency for "proof") which is not appropriate to
a heuristic, when Trask acknowledges that Multilateral Comparison
might be a useful heuristic, is a form of discourse which seriously
misstates what Multilateral Comparison can do.
Use a test which evaluates it as a heuristic,
and include ALL cases of its use,
including its use for Northern Eurasia under Catherine the Great,
and its use in pioneering stages of many language families,
and the conclusion will be different.

(This comparison with Greenberg is not intended to infuriate Trask,
it is merely to say that we should use the results of the work
of each person and of each tool of analysis
for what they are most useful for, and that will of
course be for different purposes in the cases of different persons.
That is completely obvious to me.
It constantly boggles my mind that eminent academics somehow
so often forget this obligation of respect to the rest of society.)

I don't consider it honorable to apply an inappropriate "test"
to something, when one has oneself said it is not a tool of the kind
for which such a test would be appropriate.
Trask has said MC cannot provide "proof" of the kind CM can.
And he has said it can be a heuristic.
Then he should apply tests to it which are appropriate to heuristics.

Trask at least give the appearance (very strongly so) of wanting
to deny any role for Multilateral Comparison, by focusing attention
only on abuses of it, instead of focusing on where it can be useful.

Trask is of course welcome to choose his own preferred domains
and tools of work.  But he should not try to deny the usefulness
within their own domains of effectiveness (as proven by the cases
mentioned above) of other tools which he himself
simply does not want to use, or to admit that he uses
(since I firmly believe Trask could not have become so competent
and successful without using heuristics himself).

Almost by definition, Multilateral Comparison applied to a larger
number of languages or language families
cannot be as detailed or deep a study
as can the Comparative Method when applied to a smaller
number of languages or language families.
So what?

That does not mean we are best served by using only ONE
of these two approaches.
No more than by never using any other heuristic,
or never using any other method of detailed analysis of evidence.

Sincerely
Lloyd Anderson
Ecological Linguistics