Principled Comparative Method - a new tool

[Robert Whiting]

On Mon, 23 Aug 1999, Sean Crist wrote:

> On Sat, 21 Aug 1999, Robert Whiting wrote:

>> Steve is quite correct here.  A reconstruction is just that.  In
>> the best case it is only the statistically most probable
>> original relationship between the forms found in the daughter
>> languages.

> The Comparative Method is not quantitative or probabilistic.  It's
> incorrect to describe reconstructed forms as "statistically the most
> probable".  That's not how the Comparative Method works; it doesn't
> compute numeric probabilities among competing reconstructions.

And it is incorrect to describe "In the best case it [a reconstruction]
is only the statistically most probable original relationship between
the forms found in the daughter languages" as a statement about the
comparative method.  Please read what is said before going into
knee-jerk reactions.  Nothing was said about the comparative method.
The comparative method doesn't compute anything.  If you want a
paraphrase of the comparative method, it is "similar things that are
in complementary distribution are often aspects of the same thing."

Although the comparative method is one of the major tools of
historical linguistics it is not the only one and it does not
give a "best case" reconstruction.  In general, it doesn't give
you a reconstruction at all.  What it gives you is the idea that
there is something that two different but similar forms that do not
contrast could have both developed from.  It doesn't tell you what
that something is.  If you have /b/ and /f/ in complementary
distribution in two languages (or even in the same language) you
can use the comparative method to show that they might be descended
from the same phoneme.  You can call that phoneme */b/ or */f/ or
*[labial consonanat], but the only thing that you can tell about it
phonologically is that there should be some reasonable path from it
to both /b/ and /f/.

But don't automatically associate "reconstruction" with "comparative
method" in your mind.  And keep in mind that the discussion was about
reconstructions in Middle Chinese which is an attested language so
there may very well be statistical means of arriving at the best
reconstruction.

> I've seen other variations of the phrase "statistically probable" on this
> list in contexts which suggest that what the writer means is something
> like "what we've judged to be most likely."  That's not a correct use of
> the terminology.

This is doubtless true, but then practically everybody plays fast and
loose with terms like "likely" and "probable".  And people often use
the term "statistically probable" without having real statistics
available, in which case they usually mean that they know of a lot of
cases where some particular event has happened and only a few where it
hasn't.

<snip>

>> It is not so much a question of innovation versus preservation.
>> It is a matter of how much innovation there is in each daughter
>> language.  When you have the parent preserved, this can be
>> measured.

> What exactly is being measured?  This whole line of discussion assumes
> that "innovation" is something that can be measured.  I'd like to know
> exactly what this means.  What is being counted?

You seem to be confused about what the topic of discussion is.
Basically it is about measuring pathways of change between two
daughter languages and a parent to see which of the daughters
is closer to the parent *when all three are known*.  The method
was described in a message posted by Jon Patrick on Thu, 12 Aug
1999 (see http://listserv.linguistlist.org/cgi-bin/
wa?A2=ind9908&L=indo-european&O=T&P=11528 for the message; note put this
URL back together before using it).  Essentially the method is like
counting up the little numbers between the stars on your Texaco road
map to figure which route between two places is the shortest.  My
comments were to the effect that the method could only be effective
if all three languages were known and reconstruction of the parent was
minimal (and not limited to the comparative method).

> [...]
>> will also help to reduce the likelihood of the daughters having
>> innovated the same way independently (historical linguists really
>> hate it when this happens because it screws everything up).

> Not as bad as you think, because you can often tell from the relative
> chronology that the change must have happened independently.

Which is why it is easier to do historical linguistics with languages
that have a history.  But if you only have, say, the modern forms, as
in many African or Polynesian languages, it is almost impossible to
tell common retention from independent innovation from influence of
one of the daughters on the other.

Bob Whiting
whiting at cc.helsinki.fi