Comparative Methodology

[Rankin, Robert L]

----------------------------Original message----------------------------
>I'm trying to determine the steps that would be followed in a full
implementation of comparative methodology.  I've searched the web,
including the archives of this e-list, and found nothing helpful.  I
have a half dozen or more historical linguistics textbooks, as well as other
books and articles that contain pertinent information, to include books that
present the results of applications of comparative methodology, but none of
these sources answer all of my questions about what the procedural steps of
such an implementation should be.

*************

I hope I haven't misinterpreted the author's request.  In any event, here
are my views:  In the late '90's I wrote a survey of the comparative method
(CM) for an encyclopedic treatment of a variety of historical linguistic
topics that will appear from Blackwell.  I found it helpful to consult
earlier encyclopedia articles on the subject (W.S. Allen and others) and
also found Terry Kaufman's discussion in the book _Amazonian Linguistics_ a
good source.  I tried to divide the methodology into stages for the
readership, but I didn't try to break it down much more than was already
done in this inquiry.  I guess I have always resisted the temptation to try
to codify the comparative method into a specific finite number of steps
(stages, phases, elements...) for a couple of reasons.  

I think it is a mistake to assume that the CM is a set of airtight
procedures, which, if followed faithfully, will produce the desired answers
-- genetic relationships will automatically emerge (if they are there), and
after that, lexical/phonological and finally  morphosyntactic
reconstructions will be produced.  My classes often seem to want this sort
of methodology.  Frequently it develops that they would like to program a
computer to do the repetitive work and thus to produce quick and accurate
results.  When I tell them it doesn't work like that, they sometimes accuse
me of being a "mere" humanist (and an aging one at that), who likes to sit
behind his desk cherishing the feeling of steeping himself in language
minutiae instead of taking a properly modern scientific approach.  

But the data themselves frequently suggest solutions, or, they suggest which
aspects of the CM may be useful in a given instance.  Bloomfield is said to
have remarked (tho' I'm not sure where) that "if you're going to compare two
languages, it helps to know one of them."  This was his understated way of
insisting on a knowledge of detail and exceptionality in the data being
compared.  In this he followed Meillet, who believed that we reconstruct on
the basis of exceptions, not on the basis of what is the rule.  

I think a lot of problems stem from over-emphasis in introductory
linguistics texts of the oft-repeated claim that what comparativists must
rely on is multiple, recurring sets of phoneme (by whatever definition)
correspondences in basic vocabulary.  Then you get the usual examples of
cognate sets from shallow language families to illustrate the principle.
Introductory textbooks are especially bad about leaving it at that, but some
full-fledged historical linguistics texts aren't much better.  

The reader is left with the assumption that the answers lie in finding
quantitative evidence, whereas in reality the best answers usually lie in
finding qualitative evidence.  Matching idiosyncracies and exceptions are
much more convincing evidence that two morphemes, lexemes or languages are
related than even a pretty fair number of consonant and vowel matches in
basic vocabulary.  There are dozens of sober hypotheses for relationships
circulating that have never risen above the level of speculation (and I'm
not even talking about far-fetched proposals for things like Dené-Caucasian,
Amerind or the like).  Almost none of them has escaped that status unless
significant qualitative evidence has been found to supplement ordinary
correspondences.  

Internal reconstruction is no different.  One of the relatively few places I
disagree with Anthony Fox's excellent book involves his assertion that
internal reconstruction is essentially the same as the establishment of
underlying forms in a synchronic phonology.  The difficulty is that
synchronists value productivity, whereas comparativists value
exceptionality.  The historian realizes that extreme uniformity may well be
the result of analogy or social borrowing; the synchronist doesn't (perhaps
needn't?) care.  So their results should typically be different, even though
both are treating allomorphs as cognates.  I think this needs to be
emphasized.

I'd very much like the answers to be amenable to extraction by an explicit
step-by-step methodology, but I don't think they always are.  Nor is that
really what the comparative method purports to be.  It is, as Johanna
Nichols has put it, a heuristic.  I'm probably just preaching to the choir,
so I'll stop here.

Bob Rankin